Topic: NxtMint Java minter (Read 51651 times)

ScripterRon · « **on:** January 16, 2015, 10:53:52 pm »

NxtMint 1.6.0

Download the latest version from NxtMint Releases

Unpack the files in a directory of your choice
Copy sample.NxtMint.conf to the application data directory and rename it to NxtMint.conf

Linux: <user-home>/.NxtMint/NxtMint.conf
Mac: <user-home>/Library/Application Support/NxtMint/NxtMint.conf
Windows: <user-home>\AppData\Roaming\NxtMint\NxtMint.conf

Copy sample.logging.properties to the application data directory and rename it to logging.properties
Edit NxtMint.conf to specify your desired NRS server (you can use scripterron.dyndns.biz if you don't run a local server) and your secret passphrase (the passphrase will not be sent to the server). Update the other parameters as needed.
Install OpenCL if you want to use the GPU for minting
Rename and edit sample.mint.bat (Windows) or sample.mint.sh (Linux/Mac) as desired

NxtMint requires the Java 8 runtime since it uses language features that are not available in earlier versions of Java.
Nxt NRS works with Java 8, so you can upgrade without worrying that you will break the server.

NxtMint mints currencies defined by the Nxt Monetary System. A single currency can be minted as specified by the NxtMint configuration file. The minting algorithm is executed using one or more CPU threads or GPU devices. Newly-minted coins will be added to the account specified in the configuration file.

The NRS node used to create the mint transactions must accept API connections. This is done by specifying nxt.apiServerPort, nxt.apiServerHost and nxt.allowedBotHosts in nxt.properties. The account secret phrase is not sent to the Nxt server since the mint transactions are created and signed locally.

OpenCL is used to mint using the GPU and is not needed if you are using just the CPU. You will need to obtain OpenCL from your graphics card vendor (OpenCL may be automatically installed as part of the graphics card driver installation). The directory containing the OpenCL runtime library needs to be in either PATH (Windows) or LD_LIBRARY_PATH (Linux). On Windows, OpenCL.dll is often placed in \Windows\system32 by the device driver installer, so it is always available and you won't need to update PATH.

Runtime Options
The following command-line options can be specified using -Dname=value

nxt.datadir=directory-path
Specifies the application data directory. Application data will be stored in a system-specific directory if this option is omitted:

Linux: user-home/.NxtMint
Mac: user-home/Library/Application Support/NxtMint
Windows: user-home\AppData\Roaming\NxtMint

java.util.logging.config.file=file-path
Specifies the logger configuration file. The logger properties will be read from 'logging.properties' in the application data directory. If this file is not found, the 'java.util.logging.config.file' system property will be used to locate the logger configuration file. If this property is not defined, the logger properties will be obtained from jre/lib/logging.properties.

JDK FINE corresponds to the SLF4J DEBUG level
JDK INFO corresponds to the SLF4J INFO level
JDK WARNING corresponds to the SLF4J WARN level
JDK SEVERE corresponds to the SLF4J ERROR level

Configuration options
The following configuration options can be specified in NxtMint.conf. This file is required and must be in the application data directory.

connect=host
Specifies the NRS host name and defaults to 'localhost'

apiPort=port
Specifies the NRS API port and defaults to 7876.

secretPhrase=phrase
Specifies the account secret phrase and must be specified. The secret phrase will not be sent to the NRS server. Note that the configuration file is processed using the current user locale which can cause problems if the characters in the secret phrase aren't defined in the locale.

currency=code
Specifies the code for the currency to be minted.

units=count
Specifies the number of units to generate for each hash round and defaults to 1. The hash difficulty increases as the number of units increases but the transaction fee is 1 Nxt no matter how many units are generated. Thus you want to increase units as much as possible to reduce the cost of minting the currency but don't set it so high that you don't mint anything during a session. The count can be specified as an integer value or as a decimal value with a maximum number of digits following the decimal point as defined for the currency.

cpuThreads=count
Specifies the number of CPU threads to be used and defaults to 1. Specifying a thread count greater than the number of CPU processors will not improve minting since the mint algorithms are CPU-intensive and will drive each process to 100% utilization. Decrease the thread count if your computer becomes too hot or system response degrades significantly.

gpuIntensity=count
Specifies the GPU computation intensity and defaults to 0. Your graphics card must support OpenCL in order to use the GPU. You will need to try different values to determine an acceptable hash rate. Specifying too large a value will result in performance degradation and GPU memory errors. Start with an initial value of 10 and raise or lower needed. A GPU will not be used if gpuIntensity is 0.

gpuDevice=index,wsize,gcount
Specifies the GPU device number (0, 1, 2, ...), the work group size and the work group count. The first GPU device will be used if this parameter is omitted. This parameter can be repeated to use multiple GPU devices. The GPU devices that are available are listed when NxtMint is started with a non-zero value for gpuIntensity.

The work group size specifies the number of work items per work group and defaults to 256. Performance can sometimes be improved by setting the work group size to the number of cores in a compute unit. You can determine this value by dividing the number of cores on the card by the number of compute units. For Scrypt, local memory contention can occur if the work group size is too large, so experiment with smaller values.

The work group count specifies the number of work groups that will be executed by OpenCL before returning control to the application. Specifying too large a value can result in adapter global storage being exceeded and can cause sluggish display performance as display work groups get delayed by mining work groups.

enableGUI=true|false
Specifies whether or not to enable the GUI and defaults to true. Disabling the GUI allows NxtMint to run in headless environments such as a disconnected service.

coretechs · « **Reply #1 on:** January 16, 2015, 11:22:09 pm »

Really nice work.

After getting it to build I ran into problems with a class dependency issue with LoggingFactory. I was able to resolve it by adding an entry for slf4j-log4j12 to the pom.xml file for NxtMint, but I've never used Maven before so I'm not sure if I missed some other step.

colin012 · « **Reply #2 on:** January 16, 2015, 11:24:04 pm »

Cool!

KrusherPT · « **Reply #3 on:** January 16, 2015, 11:27:58 pm »

This is very good

I'm trying to use it on windows but it says secret phrase not specified and i have all set, so i dont know my mistake

EDIT:
Nevermind i get it i need to get my conf file on appdata

colin012 · « **Reply #4 on:** January 16, 2015, 11:43:28 pm »

Hmmm, needs Https support!

faddat · « **Reply #5 on:** January 17, 2015, 12:06:54 am »

I think I've things set, but I do not know any of the possible currency codes. Thousand thanks to ScripterRon for taking the time to build this.

coretechs · « **Reply #6 on:** January 17, 2015, 12:24:47 am »

Quote from: faddat on January 17, 2015, 12:06:54 am

I think I've things set, but I do not know any of the possible currency codes. Thousand thanks to ScripterRon for taking the time to build this.

The most popular currency on the NXT MS right now is EGOLD which has the currency code EGOLD and is mintable using this GPU minter.

http://nxtportal.org/search/5775213290661997199

bidji29 · « **Reply #7 on:** January 17, 2015, 12:35:29 am »

WHat are the result in term of speed?

Modern cpu can do 1-2Mh/s on keccak with the default minter.
How many this one do with cpu and gpu?

ScripterRon · « **Reply #8 on:** January 17, 2015, 01:05:36 am »

Quote from: coretechs on January 16, 2015, 11:22:09 pm

Really nice work.

After getting it to build I ran into problems with a class dependency issue with LoggingFactory. I was able to resolve it by adding an entry for slf4j-log4j12 to the pom.xml file for NxtMint, but I've never used Maven before so I'm not sure if I missed some other step.

I use the JDK version of SLF4J (slf4j-api and slf4j-jdk14). In order to use a different logger implementation, you should replace slf4j-jdk14 in the pom for both NxtCore and NxtMint and then package both of them. Or you could just replace slf4j-jdk14.jar with slf4j-log4j12.jar in the lib subdirectory without recompiling anything (slf4j-api works with any slf4j logger implementation). You will need to start NxtMint using -cp instead of -jar in this case since the NxtMint.jar file contains the classpath as part of the manifest.

For example,
java -Xmx256m -cp "NxtMint-1.0.0.jar;lib" org.ScripterRon.NxtMint.Main

ScripterRon · « **Reply #9 on:** January 17, 2015, 01:10:45 am »

Quote from: KrusherPT on January 16, 2015, 11:27:58 pm

This is very good
I'm trying to use it on windows but it says secret phrase not specified and i have all set, so i dont know my mistake

EDIT:
Nevermind i get it i need to get my conf file on appdata

Yep - the default data directory is %APPDATA%\NxtMint

If you want to keep everything in one place, you can specify the application data directory on the command line.

For example, if you unpacked the zip into C:\NxtMint, then start it this way:
java -Xmx256m -Dnxt.datadir="C:\NxtMint" -jar "C:\NxtMint\NxtMint-1.0.0.jar"

You should also set up logging.properties to control the output. I forgot to include a sample in the first release, so I'll it on the forum. If you set NxtMint logging to FINE, you will see the hashing rates every minute.

ScripterRon · « **Reply #10 on:** January 17, 2015, 01:13:36 am »

I forgot to include a sample logging properties file for use with Java logging, so here it is. Put it into the application data directory as logging.properties.

Code: [Select]

######################################
#  	NxtMint Logging Configuration File                               #
#                                                                                      #
# Rename this file to logging.properties and place it in          #
# the NxtMint application directory.  Change the log             #
# file name as desired (java.util.logging.FileHandler.pattern) #
######################################

# Log to console and file
handlers = java.util.logging.FileHandler, java.util.logging.ConsoleHandler

# Default logging level (FINE, INFO, WARNING, SEVERE)
.level = SEVERE

# NxtMint logging level (FINE, INFO, WARNING, SEVERE)
org.ScripterRon.NxtMint.level = FINE

# NxtCore logging level (FINE, INFO, WARNING, SEVERE)
org.ScripterRon.NxtCore.level = INFO

# Create log file in temporary directory, accept FINE or higher messages,
# maximum file size is 500KB with a single file
java.util.logging.FileHandler.pattern = %t/NxtMint.log
java.util.logging.FileHandler.limit = 500000
java.util.logging.FileHandler.count = 1
java.util.logging.FileHandler.level = FINE

# Limit the message that are printed on the console to FINE and above
java.util.logging.ConsoleHandler.level = FINE

ScripterRon · « **Reply #11 on:** January 17, 2015, 01:15:48 am »

Quote from: colin012 on January 16, 2015, 11:43:28 pm

Hmmm, needs Https support!

Why? It doesn't send or receive confidential data. The minting transaction is created and signed locally without any interaction with the server. The signed transaction bytes are then sent to the server for broadcast. Since these bytes will appear in the blockchain for all to see, it doesn't really make any sense to add the overhead of https.

ScripterRon · « **Reply #12 on:** January 17, 2015, 01:19:24 am »

Quote from: faddat on January 17, 2015, 12:06:54 am

I think I've things set, but I do not know any of the possible currency codes. Thousand thanks to ScripterRon for taking the time to build this.

The currency codes are listed on the Monetary System tab in the NRS UI. For example, go to http://scripterron.dyndns.biz:7876 and login to the UI. Then select the Monetary System tab on the left. I've been mining EGOLD while testing out NxtMint. Note that each minting transaction will cost you 1 NXT, so you need some NXT in your account before you can start minting.

ScripterRon · « **Reply #13 on:** January 17, 2015, 01:25:55 am »

Quote from: bidji29 on January 17, 2015, 12:35:29 am

WHat are the result in term of speed?

Modern cpu can do 1-2Mh/s on keccak with the default minter.
How many this one do with cpu and gpu?

Here is an excerpt from the NxtMint log. Workers 0-3 are using the CPU and worker 4 is using the GPU (NVIDIA GT 720). If you have an AMD graphics card, you should get even better hash rates. The GPU intensity determines how heavily the card is used (I have it set to 15 at the moment). I have gotten rates as high as 16 MHash/s on my NVIDIA card but the video display becomes too sluggish for me to use it for anything else. 15 seems to be a good compromise but even then I get resource failures if I open a graphics-intensive application such as a browser.

Code: [Select]

08:03:06 INFO Main.main: Java Nxt Mint Version 1.0.0
08:03:06 INFO Main.main: Application data path: C:\Users\Ronald\Appdata\Roaming\NxtMint
08:03:06 INFO Main.main: Using Nxt node at scripterron.dyndns.biz:7876
08:03:06 INFO Main.main: Minting 25.000000 units of EGOLD for account NXT-XM86-4ZNA-65L5-CDWUE: 4 CPU threads, 15 GPU intensity
08:03:06 INFO Nxt.init: API node=scripterron.dyndns.biz, API port=7876
08:03:07 FINE MintWorker.run: Worker 4 starting on counter 166
08:03:07 FINE MintWorker.run: Worker 1 starting on counter 166
08:03:07 FINE MintWorker.run: Worker 0 starting on counter 166
08:03:07 FINE MintWorker.run: Worker 3 starting on counter 166
08:03:07 FINE MintWorker.run: Worker 2 starting on counter 166
08:04:07 FINE MintWorker.run: Worker 1 hash rate 1100 KH/s
08:04:07 FINE MintWorker.run: Worker 4 hash rate 12551 KH/s
08:04:07 FINE MintWorker.run: Worker 3 hash rate 1140 KH/s
08:04:07 FINE MintWorker.run: Worker 0 hash rate 1125 KH/s
08:04:07 FINE MintWorker.run: Worker 2 hash rate 1120 KH/s
08:05:07 FINE MintWorker.run: Worker 3 hash rate 1144 KH/s
08:05:07 FINE MintWorker.run: Worker 1 hash rate 1108 KH/s
08:05:07 FINE MintWorker.run: Worker 0 hash rate 1131 KH/s
08:05:07 FINE MintWorker.run: Worker 4 hash rate 12614 KH/s
08:05:07 FINE MintWorker.run: Worker 2 hash rate 1130 KH/s
08:06:07 FINE MintWorker.run: Worker 0 hash rate 1142 KH/s
08:06:07 FINE MintWorker.run: Worker 4 hash rate 12734 KH/s
08:06:07 FINE MintWorker.run: Worker 1 hash rate 1118 KH/s
08:06:07 FINE MintWorker.run: Worker 3 hash rate 1151 KH/s
08:06:07 FINE MintWorker.run: Worker 2 hash rate 1141 KH/s
08:07:07 FINE MintWorker.run: Worker 1 hash rate 1116 KH/s
08:07:07 FINE MintWorker.run: Worker 4 hash rate 12720 KH/s

achim · « **Reply #14 on:** January 17, 2015, 02:13:29 am »

why not wait with this until tomorrow when we are all awake in europe?

Jimmy2011 · « **Reply #15 on:** January 17, 2015, 02:16:13 am »

I got an error as following. Seems that I need to compile it myself

Code: [Select]

Exception in thread "main" java.lang.UnsupportedClassVersionError: org/ScripterR
on/NxtMint/Main : Unsupported major.minor version 52.0
        at java.lang.ClassLoader.defineClass1(Native Method)
        at java.lang.ClassLoader.defineClass(Unknown Source)
        at java.security.SecureClassLoader.defineClass(Unknown Source)
        at java.net.URLClassLoader.defineClass(Unknown Source)
        at java.net.URLClassLoader.access$100(Unknown Source)
        at java.net.URLClassLoader$1.run(Unknown Source)
        at java.net.URLClassLoader$1.run(Unknown Source)
        at java.security.AccessController.doPrivileged(Native Method)
        at java.net.URLClassLoader.findClass(Unknown Source)
        at java.lang.ClassLoader.loadClass(Unknown Source)
        at sun.misc.Launcher$AppClassLoader.loadClass(Unknown Source)
        at java.lang.ClassLoader.loadClass(Unknown Source)
        at sun.launcher.LauncherHelper.checkAndLoadMain(Unknown Source)

ScripterRon · « **Reply #16 on:** January 17, 2015, 02:39:43 am »

Quote from: Jimmy2011 on January 17, 2015, 02:16:13 am

I got an error as following. Seems that I need to compile it myself

What Java runtime is installed? You must have Java 8 since I use language features that are not available in earlier versions. You won't be able to recompile it on an earlier JDK since that compiler won't recognize the new language constructs.

I'll update the README to make the requirement more prominent.

qqNxt · « **Reply #17 on:** January 17, 2015, 02:41:20 am »

Thank YOU!! been waiting for this!

Jimmy2011 · « **Reply #18 on:** January 17, 2015, 02:48:36 am »

Quote from: ScripterRon on January 17, 2015, 02:39:43 am

Quote from: Jimmy2011 on January 17, 2015, 02:16:13 am
I got an error as following. Seems that I need to compile it myself
What Java runtime is installed? You must have Java 8 since I use language features that are not available in earlier versions. You won't be able to recompile it on an earlier JDK since that compiler won't recognize the new language constructs.

I'll update the README to make the requirement more prominent.

Thanks for the quick response. I installed Java 8. Could you please write a detailed guide for us to compile or install the program?

qqNxt · « **Reply #19 on:** January 17, 2015, 02:58:31 am »

I think we need a easy mint.bat to help us computer dumbs.
I have no clue how to run .jar extension

I better go grab those dusty miners out of my closet and connect them up!

ScripterRon · « **Reply #20 on:** January 17, 2015, 03:00:04 am »

Quote from: Jimmy2011 on January 17, 2015, 02:48:36 am

Thanks for the quick response. I installed Java 8. Could you please write a detailed guide for us to compile or install the program?

I updated the header post with the steps needed to install and run the program. Let me know if there are any steps missing or unclear.

The README has the build steps. I use Maven which automates everything, including downloading any required dependencies. The only manual step is obtaining Aparapi since it is not in the Maven central repository. If you want to build NxtMint, you will need to install the aparapi.jar file in your local Maven repository. This is done using the mvn install:install-file target. I'll add the complete command syntax to the README.

colin012 · « **Reply #21 on:** January 17, 2015, 03:01:43 am »

Quote from: qqNxt on January 17, 2015, 02:58:31 am

I think we need a easy mint.bat to help us computer dumbs.
I have no clue how to run .jar extension

We can package a run.bat and run.sh file like NXT does so you just have to double click something and it will run for you. (NXT used to run out of a jar too if I am remembering right).

ScripterRon · « **Reply #22 on:** January 17, 2015, 03:03:24 am »

Quote from: qqNxt on January 17, 2015, 02:58:31 am

I think we need a easy mint.bat to help us computer dumbs.
I have no clue how to run .jar extension

You should be able to just double-click the jar file. If that doesn't work, use a text editor to create mint.bat and enter the following command:
java -Xmx256m -jar NxtMint-1.0.0.jar

I'll add mint.bat and mint.sh files for the next release.

Jimmy2011 · « **Reply #23 on:** January 17, 2015, 03:18:00 am »

I just installed JDK 1.8, and got a new error.

Code: [Select]

Error: Registry key 'Software\JavaSoft\Java Runtime Environment'\CurrentVersion'

has value '1.8', but '1.7' is required.
Error: could not find java.dll
Error: Could not find Java SE Runtime Environment.

ScripterRon · « **Reply #24 on:** January 17, 2015, 03:19:23 am »

Got a real-time test of the minting queuing (NRS took 25 minutes to find a block from 9:06 to 9:31). Had me worried because I thought NxtMint was dropping transactions, but the backlog cleared once NRS started confirming transactions again.

Code: [Select]

9:06:52 INFO MintWorker.run: Worker 0 found solution for counter 169
09:06:52 FINE Mint.mint: Solution for counter 169 added to pending queue
09:06:53 INFO Mint.mint: Solution for counter 169 submitted
09:17:09 INFO MintWorker.run: Worker 4 found solution for counter 170
09:17:09 FINE Mint.mint: Solution for counter 170 added to pending queue
09:23:06 INFO MintWorker.run: Worker 4 found solution for counter 171
09:23:11 FINE Mint.mint: Solution for counter 171 added to pending queue
09:31:13 INFO Mint.mint: Solution for counter 170 submitted
09:45:47 INFO Mint.mint: Solution for counter 171 submitted
09:57:05 FINE Mint.mint: Solution for counter 172 added to pending queue
09:57:05 INFO Mint.mint: Solution for counter 172 submitted

Djinou94 · « **Reply #25 on:** January 17, 2015, 03:20:13 am »

Minting with this minter is better? faster?

ScripterRon · « **Reply #26 on:** January 17, 2015, 03:25:27 am »

Quote from: Jimmy2011 on January 17, 2015, 03:18:00 am

I just installed JDK 1.8, and got a new error.

Code: [Select]
Error: Registry key 'Software\JavaSoft\Java Runtime Environment'\CurrentVersion' has value '1.8', but '1.7' is required. Error: could not find java.dll Error: Could not find Java SE Runtime Environment.

What were you doing when you got that error? NxtMint requires 1.8, so that shouldn't be the source of the error. What is displayed for the command: java -version

qqNxt · « **Reply #27 on:** January 17, 2015, 03:30:18 am »

java -Xmx256m -jar Nxt-1.0.0.jar
Error: Unable to access jarfile Nxt-1.0.0.jar

Jimmy2011 · « **Reply #28 on:** January 17, 2015, 03:37:50 am »

Quote from: ScripterRon on January 17, 2015, 03:25:27 am

Quote from: Jimmy2011 on January 17, 2015, 03:18:00 am

I just installed JDK 1.8, and got a new error.

Code: [Select]
Error: Registry key 'Software\JavaSoft\Java Runtime Environment'\CurrentVersion' has value '1.8', but '1.7' is required. Error: could not find java.dll Error: Could not find Java SE Runtime Environment.
What were you doing when you got that error? NxtMint requires 1.8, so that shouldn't be the source of the error. What is displayed for the command: java -version

Done.

'java -version' reported the same error.

I installed JDK 1.8 when Nxt is running. So I think I need to reboot computer and reinstall JDK 1.8. After that and reinstall, everything goes well.

Thanks.

colin012 · « **Reply #29 on:** January 17, 2015, 03:45:53 am »

Quote from: qqNxt on January 17, 2015, 03:30:18 am

java -Xmx256m -jar Nxt-1.0.0.jar
Error: Unable to access jarfile Nxt-1.0.0.jar

I think he meant NxtMint-1.0.0.jar rather than Nxt-1.0.0.jar. Try that instead.

qqNxt · « **Reply #30 on:** January 17, 2015, 04:16:45 am »

edit: I forgot to copy to application data.

SEVERE Main.main: Exception during program initialization
java.lang.IllegalArgumentException:Secret phrase not specified
at org.ScripterRon.NxtMint.Main.main<Main.java:186>

I did provide the secret phrase. and I did copy and rename to NxtMint.conf
currency=EGOLD
units=100
cpuThreads=1
gpuIntensity=0
enableGUI=true

ScripterRon · « **Reply #31 on:** January 17, 2015, 04:26:10 am »

Quote from: Jimmy2011 on January 17, 2015, 03:37:50 am

Done.

'java -version' reported the same error.

I installed JDK 1.8 when Nxt is running. So I think I need to reboot computer and reinstall JDK 1.8. After that and reinstall, everything goes well.

Thanks.

That makes sense. The java.dll would have been in use and the installer couldn't replace it with the new version. Thus the mismatch error.

ScripterRon · « **Reply #32 on:** January 17, 2015, 04:28:38 am »

Quote from: colin012 on January 17, 2015, 03:45:53 am

I think he meant NxtMint-1.0.0.jar rather than Nxt-1.0.0.jar. Try that instead.

Whoops - NxtMint-1.0.0.jar is right. I'll modify the post.

qqNxt · « **Reply #33 on:** January 17, 2015, 04:31:30 am »

one last problem and i'll graduate from running the java minter!

the program is running now but I can't tell if its doing any calculations/mhs
the log box is just empty.

is this normal? im using gpu and I have aparapi in the nxtmint directory.

jasemoney · « **Reply #34 on:** January 17, 2015, 04:36:46 am »

Quote from: qqNxt on January 17, 2015, 04:16:45 am

edit: I forgot to copy to application data.

SEVERE Main.main: Exception during program initialization
java.lang.IllegalArgumentException:Secret phrase not specified
at org.ScripterRon.NxtMint.Main.main<Main.java:186>

I did provide the secret phrase. and I did copy and rename to NxtMint.conf
currency=EGOLD
units=100
cpuThreads=1
gpuIntensity=0
enableGUI=true

ahh which data directory?? same trouble. all my nxt is in a folder on my desktop. copied everything aparapi and nxtminet there. changed logger and conf, and copied them into conf.. still actin weird

*edit ya didnt realize it made a folder nxtmint in %appdata%
not on to next bit, figuring out the PATH

qqNxt · « **Reply #35 on:** January 17, 2015, 04:43:28 am »

Quote from: qqNxt on January 17, 2015, 04:31:30 am

one last problem and i'll graduate from running the java minter!

the program is running now but I can't tell if its doing any calculations/mhs
the log box is just empty.

is this normal? im using gpu and I have aparapi in the nxtmint directory.

my computer is heating up and fan running louder, it should be running but all I see is blank box.

Quote from: jasemoney on January 17, 2015, 04:36:46 am

ahh which data directory?? same trouble. all my nxt is in a folder on my desktop. copied everything aparapi and nxtminet there. changed logger and conf, and copied them into conf.. still actin weird

found this in readme.md
- Linux: user-home/.NxtMint
      - Mac: user-home/Library/Application Support/NxtMint

- Windows: user-home\AppData\Roaming\NxtMint

Admiral Ackbar · « **Reply #36 on:** January 17, 2015, 04:43:38 am »

I am on Linux 64bit. CPU hashing works. Now I am trying to get the AMD started. Got the AMD SDK and aparapi. But I get this:

11:36:01 SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load aparapi native library aparapi_x86_64 or possibly failed to locate opencl native library (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD_LIBRARY_PATH (linux).

Seems something is not in the right folder. Can you give a hint which file needs to be where?

Jimmy2011 · « **Reply #37 on:** January 17, 2015, 04:50:02 am »

Quote from: qqNxt on January 17, 2015, 04:31:30 am

one last problem and i'll graduate from running the java minter!

the program is running now but I can't tell if its doing any calculations/mhs
the log box is just empty.

is this normal? im using gpu and I have aparapi in the nxtmint directory.

I think you may need to download https://github.com/ScripterRon/NxtMint/blob/master/sample.logging.properties

and

Code: [Select]

Copy sample.logging.properties to the application data directory and rename it to logging.properties.

BTW, could the log show how many hashes have been calculated as that in the original CPU mint.exe ?

ScripterRon · « **Reply #38 on:** January 17, 2015, 04:51:10 am »

Quote from: jasemoney on January 17, 2015, 04:36:46 am

Quote from: qqNxt on January 17, 2015, 04:16:45 am
edit: I forgot to copy to application data.

SEVERE Main.main: Exception during program initialization
java.lang.IllegalArgumentException:Secret phrase not specified
at org.ScripterRon.NxtMint.Main.main<Main.java:186>

I did provide the secret phrase. and I did copy and rename to NxtMint.conf
currency=EGOLD
units=100
cpuThreads=1
gpuIntensity=0
enableGUI=true

ahh which data directory?? same trouble. all my nxt is in a folder on my desktop. copied everything aparapi and nxtminet there. changed logger and conf, and copied them into conf.. still actin weird

The default application data directory on Windows is C:\Users\user-name\AppData\Roaming\NxtMint (this is a Microsoft convention since they want to separate shared program data and user-specific data). However, you can put everything into the same directory by specifying -Dnxt.datadir=directory-path on the Java command line where directory-path is the directory where you installed NxtMint.

I generally prefer the separation of programs and data since that allows me to install new versions of a program in a separate directory and still have access to the program data without copying files back and forth between versions.

ScripterRon · « **Reply #39 on:** January 17, 2015, 04:55:46 am »

Quote from: qqNxt on January 17, 2015, 04:31:30 am

one last problem and i'll graduate from running the java minter!

the program is running now but I can't tell if its doing any calculations/mhs
the log box is just empty.

is this normal? im using gpu and I have aparapi in the nxtmint directory.

Did you create logging.properties in the application data directory? The default logging.properties does not log debug (FINE) messages.

I forgot to include a sample in the release archive. I posted the sample in an earlier post or you can get it from GitHub (http://github.com/ScripterRon/NxtMint) and look for sample.logging.properties. Paste it into a text editor and save it as logging.properties.

ScripterRon · « **Reply #40 on:** January 17, 2015, 05:04:17 am »

Quote from: Admiral Ackbar on January 17, 2015, 04:43:38 am

I am on Linux 64bit. CPU hashing works. Now I am trying to get the AMD started. Got the AMD SDK and aparapi. But I get this:

11:36:01 SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load aparapi native library aparapi_x86_64 or possibly failed to locate opencl native library (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD_LIBRARY_PATH (linux).

Seems something is not in the right folder. Can you give a hint which file needs to be where?

Are the files in one of the directories specified by the LD_LIBRARY_PATH environment variable? I don't have a Linux desktop system to try this out on. You could also try setting the Java library path to point to the directory containing the aparapi and opencl library files. This is done by specifying -Djava.library.path=<library-directory> on the Java command line when you start NxtMint.

ScripterRon · « **Reply #41 on:** January 17, 2015, 05:08:54 am »

Quote from: Jimmy2011 on January 17, 2015, 04:50:02 am

BTW, could the log show how many hashes have been calculated as that in the original CPU mint.exe ?

Sure - the worker hash count since the last solution is what is used to compute the hash rate, so I can easily add that to the log line. In addition, the GUI display contains the total number of hashes performed by the worker that found the solution.

jasemoney · « **Reply #42 on:** January 17, 2015, 05:10:59 am »

path stuff isnt going so well for me, i put aparapi files, opencl.dll, and opencl64.dll like in a bunch of spots. and changed my PATH user variable to the nextmint folder where everything is.
I even changed the run.bat to this:

Quote

java -Xmx256m -jar NxtMint-1.0.0.jar
-Djava.library.path=C:\nxtminter
pause

i suck at this!

qqNxt · « **Reply #43 on:** January 17, 2015, 05:18:12 am »

all working now! thanks!

@jasemoney I put aparapi files all in the nxtmint directory. like NxtMint-1.0.0\aparapi_x86.dll
and this for the mint.bat
java -Xmx256m -jar NxtMint-1.0.0.jar

these here
\AppData\Roaming\NxtMint\NxtMint.conf
\AppData\Roaming\NxtMint\logging.properties

hope this helps.

Admiral Ackbar · « **Reply #44 on:** January 17, 2015, 05:22:32 am »

Quote from: ScripterRon on January 17, 2015, 05:04:17 am

Quote from: Admiral Ackbar on January 17, 2015, 04:43:38 am
I am on Linux 64bit. CPU hashing works. Now I am trying to get the AMD started. Got the AMD SDK and aparapi. But I get this:

11:36:01 SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load aparapi native library aparapi_x86_64 or possibly failed to locate opencl native library (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD_LIBRARY_PATH (linux).

Seems something is not in the right folder. Can you give a hint which file needs to be where?
Are the files in one of the directories specified by the LD_LIBRARY_PATH environment variable? I don't have a Linux desktop system to try this out on. You could also try setting the Java library path to point to the directory containing the aparapi and opencl library files. This is done by specifying -Djava.library.path=<library-directory> on the Java command line when you start NxtMint.

Yes, I tried them all in one folder and also copied them around like the other cell mates here. No luck.
Can you name the specific files it is looking for? For example I do not have a opencl.so. They are named differently

/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libamdhsasc64.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libamdocl64.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libGLEW.a
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libGLEW.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libglut.a
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libglut.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libOpenCL.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libOpenCL.so.1

jasemoney · « **Reply #45 on:** January 17, 2015, 05:55:43 am »

ok so I updated to JDK 8 latest version from Java, workes great now and now asking for silly .dlls anymore

manrus · « **Reply #46 on:** January 17, 2015, 10:21:50 am »

Works fine. Thanks

achim · « **Reply #47 on:** January 17, 2015, 10:42:32 am »

if my gpu usage is 99%, my GPU is succesfully calculating hashes, right?

Just wondering because i set units=1 for EGOLD for testing, and I only minted 2 EGOLD in like 10 mintes...

edit: Only 217MH in 8 mintes.

manrus · « **Reply #48 on:** January 17, 2015, 10:46:23 am »

Quote from: achim on January 17, 2015, 10:42:32 am

if my gpu usage is 99%, my GPU is succesfully calculating hashes, right?

Just wondering because i set units=1 for EGOLD for testing, and I only minted 2 EGOLD in like 10 mintes...

My gpu 70-80% load. In last 45 min found 20 units egold. [GPU - Intel Iris Pro 5200]

first 10 units found at 4763 MH
second 10 units at 12110 MH

achim · « **Reply #49 on:** January 17, 2015, 11:03:12 am »

Quote from: manrus on January 17, 2015, 10:46:23 am

Quote from: achim on January 17, 2015, 10:42:32 am
if my gpu usage is 99%, my GPU is succesfully calculating hashes, right?

Just wondering because i set units=1 for EGOLD for testing, and I only minted 2 EGOLD in like 10 mintes...

My gpu 70-80% load. In last 45 min found 20 units egold. [GPU - Intel Iris Pro 5200]

first 10 units found at 4763 MH
second 10 units at 12110 MH

Hmm something looks not right with my minter. You have ca. 6 MH/s, while I'm at 0,45 MH/s
With an AMD 8870M

Although I get no error, GUI comes up and I occasionally mint something

manrus · « **Reply #50 on:** January 17, 2015, 11:13:45 am »

Quote from: achim on January 17, 2015, 11:03:12 am

Hmm something looks not right with my minter. You have ca. 6 MH/s, while I'm at 0,45 MH/s
With an AMD 8870M

Although I get no error, GUI comes up and I occasionally mint something

Set intensity lower. Big value make more errors.

crimi · « **Reply #51 on:** January 17, 2015, 01:06:51 pm »

Nice work. Much fun.

With r290x it is cutting trough EGOLD like butter. 35-40 MH/s with a little tweaking and cpu mining aswell around 60 MH/s

Intensity 14 is limit for me... further bluescreen ^^

Sat Jan 17 14:32:23 CET 2015 Units: 20.0 Hashes: 1110 MH
Sat Jan 17 14:32:29 CET 2015 Units: 20.0 Hashes: 227 MH
Sat Jan 17 14:39:05 CET 2015 Units: 20.0 Hashes: 15952 MH
Sat Jan 17 14:41:13 CET 2015 Units: 20.0 Hashes: 5160 MH
Sat Jan 17 14:45:08 CET 2015 Units: 20.0 Hashes: 9473 MH
Sat Jan 17 14:45:33 CET 2015 Units: 20.0 Hashes: 1029 MH

ScripterRon · « **Reply #52 on:** January 17, 2015, 02:48:03 pm »

Quote from: jasemoney on January 17, 2015, 05:10:59 am

path stuff isnt going so well for me, i put aparapi files, opencl.dll, and opencl64.dll like in a bunch of spots. and changed my PATH user variable to the nextmint folder where everything is.
I even changed the run.bat to this:
Quote
java -Xmx256m -jar NxtMint-1.0.0.jar
-Djava.library.path=C:\nxtminter
pause
i suck at this!

All options must be specified before -jar since anything after that is treated as program arguments. Try:
java -Xmx256m -Djava.library.path=C:\nxtminter -jar NxtMint-1.0.0.jar

However, since you are on Windows, it should work if you place runtime libraries in C:\nxtminter and make sure that is the current directory when you start NxtMint (Windows looks in the current directory before searching the PATH).

ScripterRon · « **Reply #53 on:** January 17, 2015, 02:55:02 pm »

Quote from: Admiral Ackbar on January 17, 2015, 05:22:32 am

Quote from: ScripterRon on January 17, 2015, 05:04:17 am
Quote from: Admiral Ackbar on January 17, 2015, 04:43:38 am
I am on Linux 64bit. CPU hashing works. Now I am trying to get the AMD started. Got the AMD SDK and aparapi. But I get this:

11:36:01 SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load aparapi native library aparapi_x86_64 or possibly failed to locate opencl native library (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD_LIBRARY_PATH (linux).

Seems something is not in the right folder. Can you give a hint which file needs to be where?
Are the files in one of the directories specified by the LD_LIBRARY_PATH environment variable? I don't have a Linux desktop system to try this out on. You could also try setting the Java library path to point to the directory containing the aparapi and opencl library files. This is done by specifying -Djava.library.path=<library-directory> on the Java command line when you start NxtMint.

Yes, I tried them all in one folder and also copied them around like the other cell mates here. No luck.
Can you name the specific files it is looking for? For example I do not have a opencl.so. They are named differently

/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libamdhsasc64.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libamdocl64.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libGLEW.a
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libGLEW.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libglut.a
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libglut.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libOpenCL.so
/opt/AMDAPPSDK-3.0-0-Beta/lib/x86_64/libOpenCL.so.1

I'm guessing it wants libOpenCL.so, but you should copy all the files in lib/x86_64. Did you try setting the Java library path?
java -Xmx256m -Djava.library.path="/opt/AMDAPPSDK-3.0-0-Beta/lib/X86_64:<directory containing libaparapi_x86_64.so> -jar NxtMint-1.0.0.jar

I don't have a Linux system to test this on.

Admiral Ackbar · « **Reply #54 on:** January 17, 2015, 02:55:58 pm »

Quote from: ScripterRon on January 17, 2015, 05:04:17 am

Quote from: Admiral Ackbar on January 17, 2015, 04:43:38 am
I am on Linux 64bit. CPU hashing works. Now I am trying to get the AMD started. Got the AMD SDK and aparapi. But I get this:

11:36:01 SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load aparapi native library aparapi_x86_64 or possibly failed to locate opencl native library (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD_LIBRARY_PATH (linux).

Seems something is not in the right folder. Can you give a hint which file needs to be where?
Are the files in one of the directories specified by the LD_LIBRARY_PATH environment variable? I don't have a Linux desktop system to try this out on. You could also try setting the Java library path to point to the directory containing the aparapi and opencl library files. This is done by specifying -Djava.library.path=<library-directory> on the Java command line when you start NxtMint.

I got it to work somehow by pushing every possible file into every possible folder.
It starts now without error message for the GPU but I don't think it works correctly. It appears to still run on CPU (which is set to 0 threads). The CPU usage is high. No speed improvement visible.
In a mixed mode (CPU and GPU) the worker with the highest number (presumably the GPU) gets less and less hashes the more CPU cores I activate.
Any idea how to check the GPU is crunching?

edit: oops, our posts just criss crossed

2nd edit: I threw everything in the OpenCl installation and started it successfully with java -Xmx256m -Djava.library.path=/blablabla.../AMDAPPSDK-3.0-0-Beta/lib -jar NxtMint-1.0.0.jar

ScripterRon · « **Reply #55 on:** January 17, 2015, 02:59:02 pm »

Quote from: achim on January 17, 2015, 10:42:32 am

if my gpu usage is 99%, my GPU is succesfully calculating hashes, right?

Just wondering because i set units=1 for EGOLD for testing, and I only minted 2 EGOLD in like 10 mintes...

edit: Only 217MH in 8 mintes.

The hash count for a solution is just the number of hashes done by the worker that found the solution. The log file displays the hash rate for each worker every minute (make sure you created logging.properties in the application data directory so the debug messages are logged).

ScripterRon · « **Reply #56 on:** January 17, 2015, 03:02:48 pm »

Quote from: crimi on January 17, 2015, 01:06:51 pm

Nice work. Much fun.

With r290x it is cutting trough EGOLD like butter. 35-40 MH/s with a little tweaking and cpu mining aswell around 60 MH/s Intensity 14 is limit for me... further bluescreen ^^

Sat Jan 17 14:32:23 CET 2015 Units: 20.0 Hashes: 1110 MH
Sat Jan 17 14:32:29 CET 2015 Units: 20.0 Hashes: 227 MH
Sat Jan 17 14:39:05 CET 2015 Units: 20.0 Hashes: 15952 MH
Sat Jan 17 14:41:13 CET 2015 Units: 20.0 Hashes: 5160 MH
Sat Jan 17 14:45:08 CET 2015 Units: 20.0 Hashes: 9473 MH
Sat Jan 17 14:45:33 CET 2015 Units: 20.0 Hashes: 1029 MH

Great! Guess I need to work on a GPU version for Scrypt coins since you are going to corner the EGOLD market

achim · « **Reply #57 on:** January 17, 2015, 03:04:01 pm »

Quote from: ScripterRon on January 17, 2015, 02:59:02 pm

Quote from: achim on January 17, 2015, 10:42:32 am
if my gpu usage is 99%, my GPU is succesfully calculating hashes, right?

Just wondering because i set units=1 for EGOLD for testing, and I only minted 2 EGOLD in like 10 mintes...

edit: Only 217MH in 8 mintes.
The rate is MH/sec and is the average since the last successful minting transaction. However, 217 MH/s is pretty low. What graphics card do you have? NVIDIA is usually slower than AMD for integer operations and the older NVIDIA cards are even slower (4-digit series are older than 3-digit series, which is kind of confusing).

217 MH/s for egold would be pretty nice

it's all good now, I'm crunching along with 18 MH/s, probably set the intensity too high before.

ScripterRon · « **Reply #58 on:** January 17, 2015, 03:12:54 pm »

Quote from: Admiral Ackbar on January 17, 2015, 02:55:58 pm

I got it to work somehow by pushing every possible file into every possible folder.
It starts now without error message for the GPU but I don't think it works correctly. It appears to still run on CPU (which is set to 0 threads). The CPU usage is high. No speed improvement visible.
In a mixed mode (CPU and GPU) the worker with the highest number (presumably the GPU) gets less and less hashes the more CPU cores I activate.
Any idea how to check the GPU is crunching?

edit: oops, our posts just criss crossed

2nd edit: I threw everything in the OpenCl installation and started it successfully with java -Xmx256m -Djava.library.path=/blablabla.../AMDAPPSDK-3.0-0-Beta/lib -jar NxtMint-1.0.0.jar

Aparapi and OpenCL need CPU resources to move data to/from GPU memory and to initiate each kernel instance. So pushing the CPU load up will actually reduce the GPU throughput since OpenCL needs to wait for time on the CPU. The best method is to start with cpuThreads=0 and then adjust gpuIntensity to get the best GPU rate without errors (GPU throughput will actually decline at some point as the cores become saturated, eventually failling due to lack of resources). Once you have done this, you can then start to add CPU threads until you notice a degradation in GPU throughput.

As far as the GPU not being used at all, look at the log file. The GPU worker will fallback to CPU hashing if the GPU fails. This happens if the OpenCL environment cannot be initialized, GPU resources are not available or become unavailable during hashing (for example, a graphics-intensive program is started while NxtMint is running). The CPU workers are started first, then the GPU worker is started. So, if cpuThreads=4, then workers 0-3 are CPU and worker 4 is GPU.

ScripterRon · « **Reply #59 on:** January 17, 2015, 03:22:35 pm »

Quote from: achim on January 17, 2015, 03:04:01 pm

217 MH/s for egold would be pretty nice
it's all good now, I'm crunching along with 18 MH/s, probably set the intensity too high before.

Yep - I misread your append. Out of curiousity, is that an NVIDIA or an AMD card? I'd like to get an idea of the differences between the various cards. Since the GPU kernel program is built just once and then reused for each kernel instance, Java vs native shouldn't be an issue for GPU hashing. CPU hashing benefits from JIT although a native implementation would certainly be faster (but I don't know what the performance difference actually is).

achim · « **Reply #60 on:** January 17, 2015, 03:27:28 pm »

Quote from: ScripterRon on January 17, 2015, 03:22:35 pm

Quote from: achim on January 17, 2015, 03:04:01 pm
217 MH/s for egold would be pretty nice
it's all good now, I'm crunching along with 18 MH/s, probably set the intensity too high before.
Yep - I misread your append. Out of curiousity, is that an NVIDIA or an AMD card? I'd like to get an idea of the differences between the various cards. Since the GPU kernel program is built just once and then reused for each kernel instance, Java vs native shouldn't be an issue for GPU hashing. CPU hashing benefits from JIT although a native implementation would certainly be faster (but I don't know what the performance difference actually is).

GPU is 8870m AMD, on a samsung chronos 7 notebook.
GPU does 17 MH/s on a slight overclock 900/1300

lopalcar · « **Reply #61 on:** January 17, 2015, 04:13:43 pm »

Any ways to enable more than one GPU in the same machine? I have one double core GPU from the scrypt mining times and need to heat a bit the room

And why the GPU core in use isn't at 100% even using 30 intensity?, I checked with 25 and 30 but this isn't increassing the usage :S

anyways, post your nxt address and will send you some EGOLD for your work! Thanks for the miner

Pfool · « **Reply #62 on:** January 17, 2015, 04:20:10 pm »

Thank you ScripterRon!

I was trying to make a GPU miner with aparapi like you but only achieve to get a 2.2 MH/s speed. I have integrated your kernel in my miner and I am now at 145 MH/s with my HD 7950 @ 950MHz. I am running your kernel with an equivalent intensity of 256.

Good job!

Jimmy2011 · « **Reply #63 on:** January 17, 2015, 04:21:14 pm »

I remote login to my computer with GPU, and then shutdown the NxtMint to change some parameters. I got the following error when I restart it again. Can I remotely start NxtMint with GPU?

Code: [Select]

# A fatal error has been detected by the Java Runtime Environment:
#
#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x000000005ee0e685, pid=2220, tid=6868
#
# JRE version: Java(TM) SE Runtime Environment (8.0_25-b18) (build 1.8.0_25-b18)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (25.25-b02 mixed mode windows-amd64 compressed oops)
# Problematic frame:
# V  [jvm.dll+0x12e685]

lopalcar · « **Reply #64 on:** January 17, 2015, 04:23:35 pm »

Quote from: lopalcar on January 17, 2015, 04:13:43 pm

Any ways to enable more than one GPU in the same machine? I have one double core GPU from the scrypt mining times and need to heat a bit the room
And why the GPU core in use isn't at 100% even using 30 intensity?, I checked with 25 and 30 but this isn't increassing the usage :S

anyways, post your nxt address and will send you some EGOLD for your work! Thanks for the miner

Just realized that maybe the cause to not use more than 80% of GPU could be a bottleneck with the CPU cause only use one thread, and since I have 6 cores-> 12 threads one of the threads get full of work quite easy...
Do you have planed making the process which feeds the GPU with the CPU multicore? Or should I disable hyperthreathing

crimi · « **Reply #65 on:** January 17, 2015, 04:40:44 pm »

Quote from: Pfool on January 17, 2015, 04:20:10 pm

Thank you ScripterRon!

I was trying to make a GPU miner with aparapi like you but only achieve to get a 2.2 MH/s speed. I have integrated your kernel in my miner and I am now at 145 MH/s with my HD 7950 @ 950MHz. I am running your kernel with an equivalent intensity of 256.

Good job!

Thanks for the tip with intensity of 256, long ago i did mine any coin. I was using 10-14 and got 55 MH/s now i have 245 MH/s @r290x tri oc water cooled. 512 even better 265 MH/s.

ScripterRon · « **Reply #66 on:** January 17, 2015, 04:41:36 pm »

Quote from: lopalcar on January 17, 2015, 04:13:43 pm

Any ways to enable more than one GPU in the same machine? I have one double core GPU from the scrypt mining times and need to heat a bit the room
And why the GPU core in use isn't at 100% even using 30 intensity?, I checked with 25 and 30 but this isn't increassing the usage :S

anyways, post your nxt address and will send you some EGOLD for your work! Thanks for the miner

I'll look into that. I don't have the hardware, so I'll need somebody to test it for me once I get it coded. Aparapi just added support for enumerating devices, so it should be interesting times

Nxt account: NXT-XM86-4ZNA-65L5-CDWUE

ScripterRon · « **Reply #67 on:** January 17, 2015, 04:55:32 pm »

Quote from: Jimmy2011 on January 17, 2015, 04:21:14 pm

I remote login to my computer with GPU, and then shutdown the NxtMint to change some parameters. I got the following error when I restart it again. Can I remotely start NxtMint with GPU?

Code: [Select]
# A fatal error has been detected by the Java Runtime Environment: # # EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x000000005ee0e685, pid=2220, tid=6868 # # JRE version: Java(TM) SE Runtime Environment (8.0_25-b18) (build 1.8.0_25-b18) # Java VM: Java HotSpot(TM) 64-Bit Server VM (25.25-b02 mixed mode windows-amd64 compressed oops) # Problematic frame: # V [jvm.dll+0x12e685]

That is an error in the JVM itself. The type of login really shouldn't make a difference unless it has something to do with login access permissions or stdin/stdout/stderr redirection. I don't know if an error in the Aparapi/OpenCL JNI code would show up as a JVM error. Since it points to jvm.dll, that doesn't seem like the problem. If the problem is repeatable, you could try disabling GPU hashing to see if the problem goes away.

Jimmy2011 · « **Reply #68 on:** January 17, 2015, 05:00:03 pm »

Quote from: ScripterRon on January 17, 2015, 04:55:32 pm

Quote from: Jimmy2011 on January 17, 2015, 04:21:14 pm
I remote login to my computer with GPU, and then shutdown the NxtMint to change some parameters. I got the following error when I restart it again. Can I remotely start NxtMint with GPU?

Code: [Select]
# A fatal error has been detected by the Java Runtime Environment: # # EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x000000005ee0e685, pid=2220, tid=6868 # # JRE version: Java(TM) SE Runtime Environment (8.0_25-b18) (build 1.8.0_25-b18) # Java VM: Java HotSpot(TM) 64-Bit Server VM (25.25-b02 mixed mode windows-amd64 compressed oops) # Problematic frame: # V [jvm.dll+0x12e685]
That is an error in the JVM itself. The type of login really shouldn't make a difference unless it has something to do with login access permissions or stdin/stdout/stderr redirection. I don't know if an error in the Aparapi/OpenCL JNI code would show up as a JVM error. Since it points to jvm.dll, that doesn't seem like the problem. If the problem is repeatable, you could try disabling GPU hashing to see if the problem goes away.

It goes well if I disable GPU. It may be system restricted remotely access GPU, because I can't open NVIDIA control panel, which shows me it can't be shown from remote client.

ScripterRon · « **Reply #69 on:** January 17, 2015, 05:06:37 pm »

Quote from: lopalcar on January 17, 2015, 04:23:35 pm

Just realized that maybe the cause to not use more than 80% of GPU could be a bottleneck with the CPU cause only use one thread, and since I have 6 cores-> 12 threads one of the threads get full of work quite easy...
Do you have planed making the process which feeds the GPU with the CPU multicore? Or should I disable hyperthreathing

It could just be that Aparapi/OpenCL isn't fast enough to saturate your GPU. The GPU workload is divided into groups based on the number of GPU processing units on the card. As each group completes, Aparapi starts the next group. Once all groups have finished, Aparapi retrieves the data from GPU memory and restarts the Java thread. There isn't really anything that multi-threading could help with since the data is fairly small (40 bytes of input and 32 bytes of output plus some control fields). The biggest chunk is the kernel code which needs to be loaded into GPU memory. I don't have any control over the internal operation of Aparapi, so I can't change anything there. Increasing gpuIntensity will increase the number of kernel executions before control returns to Java, so that is the best way to increase GPU utilization.

achim · « **Reply #70 on:** January 17, 2015, 05:39:49 pm »

Quote from: achim on January 17, 2015, 03:27:28 pm

Quote from: ScripterRon on January 17, 2015, 03:22:35 pm
Quote from: achim on January 17, 2015, 03:04:01 pm
217 MH/s for egold would be pretty nice
it's all good now, I'm crunching along with 18 MH/s, probably set the intensity too high before.
Yep - I misread your append. Out of curiousity, is that an NVIDIA or an AMD card? I'd like to get an idea of the differences between the various cards. Since the GPU kernel program is built just once and then reused for each kernel instance, Java vs native shouldn't be an issue for GPU hashing. CPU hashing benefits from JIT although a native implementation would certainly be faster (but I don't know what the performance difference actually is).

GPU is 8870m AMD, on a samsung chronos 7 notebook.
GPU does 17 MH/s on a slight overclock 900/1300

cool, with intensity=512 I now get 29 MH/s

jasemoney · « **Reply #71 on:** January 17, 2015, 06:57:14 pm »

I just upped the intensity as it wasnt fully saturated was using like 75%, for 10 I was getting low 40Mhash/s now at 350 im getting 180Mhash/s now its over 90% utilized 73*C at 2.8k fan rpm. FYI the kernel uses very little gpu memory, about 20mb dynamic and 20 mb dedicated

good luck all!

*edit, noted issue.. if solving faster than blocks are being found for nxt if a que forms it doesnt seem to recover, it will just stack addition aolutions into a que and not get around to sending them once a block is found.

ScripterRon · « **Reply #72 on:** January 17, 2015, 10:18:25 pm »

Quote from: jasemoney on January 17, 2015, 06:57:14 pm

I just upped the intensity as it wasnt fully saturated was using like 75%, for 10 I was getting low 40Mhash/s now at 350 im getting 180Mhash/s now its over 90% utilized 73*C at 2.8k fan rpm. FYI the kernel uses very little gpu memory, about 20mb dynamic and 20 mb dedicated
good luck all!

*edit, noted issue.. if solving faster than blocks are being found for nxt if a que forms it doesnt seem to recover, it will just stack addition aolutions into a que and not get around to sending them once a block is found.

I had a queue build up last night and it cleared as blocks were found. Only 1 transaction can be submitted per block. Do you still have the log so you can match solution timestamps to Nxt block timestamps?

ScripterRon · « **Reply #73 on:** January 17, 2015, 10:36:02 pm »

Queuing appears to be working. I mined an easy coin and built up a queue. It cleared as blocks were found. Note that I wait 30 seconds between polls when waiting for the next block to be generated (don't want to overload the server with nuisance API requests)

Code: [Select]

----> Block 335535 generated at 5:16 <----
05:23:15 INFO MintWorker.run: Worker 2 found solution for counter 1
05:23:15 FINE Mint.mint: Solution for counter 1 added to pending queue
05:23:15 INFO Mint.mint: Solution for counter 1 submitted
05:23:17 FINE Mint.mint: Solution for counter 2 added to pending queue
05:23:47 FINE Mint.mint: Solution for counter 3 added to pending queue
05:25:48 FINE Mint.mint: Solution for counter 4 added to pending queue
----> Block 335536 generated at 5:25 <----
05:25:48 INFO Mint.mint: Solution for counter 2 submitted
----> Block 335537 generated at 5:25 <----
05:26:18 INFO Mint.mint: Solution for counter 3 submitted
05:26:48 FINE Mint.mint: Solution for counter 5 added to pending queue
----> Block 335538 generated at 5:27 <----
05:26:49 INFO Mint.mint: Solution for counter 4 submitted
05:28:19 FINE Mint.mint: Solution for counter 6 added to pending queue
----> Block 335539 generated at 5:27 <----
05:28:19 INFO Mint.mint: Solution for counter 5 submitted

jasemoney · « **Reply #74 on:** January 18, 2015, 01:12:12 am »

Thanks for reply, sorry i didnt save it. I could make a new one by submitting for SILVR hehe. If it happens again ill post the log but I'm on EGOLD for the moment.

Side thoughts: what happens when the POW is over for these coins? minting up to 1/1000 at a time, I dont see EGOLD or SILVR lasting more than 10-20 days to run through the POW phase.

ScripterRon · « **Reply #75 on:** January 18, 2015, 02:19:50 am »

Quote from: jasemoney on January 18, 2015, 01:12:12 am

Thanks for reply, sorry i didnt save it. I could make a new one by submitting for SILVR hehe. If it happens again ill post the log but I'm on EGOLD for the moment.

Side thoughts: what happens when the POW is over for these coins? minting up to 1/1000 at a time, I dont see EGOLD or SILVR lasting more than 10-20 days to run through the POW phase.

SILVR will go even faster shortly - I'm testing the GPU version of SHA-256 at the moment

I suppose coins will be created with greater difficulties. I think we are in sandbox mode at the moment, just building sand castles to see what happens.

MAMLCOM · « **Reply #76 on:** January 18, 2015, 02:34:53 am »

Why it's say.. currency code is not valid

ScripterRon · « **Reply #77 on:** January 18, 2015, 02:51:14 am »

Quote from: MAMLCOM on January 18, 2015, 02:34:53 am

Why it's say.. currency code is not valid

What currency code is not valid? Did you update NxtMint.conf with a valid currency? Is NxtMint.conf in the application data directory?

MAMLCOM · « **Reply #78 on:** January 18, 2015, 02:59:06 am »

Quote from: ScripterRon on January 18, 2015, 02:51:14 am

Quote from: MAMLCOM on January 18, 2015, 02:34:53 am
Why it's say.. currency code is not valid
What currency code is not valid? Did you update NxtMint.conf with a valid currency? Is NxtMint.conf in the application data directory?

SEVERE Main.main: Exception during program initialization
java.lang.illegalArgumentException:Currency code is not valid

That miner said... I did update NxtMint.conf with currency EGOLD.. and path i think it's right..

egold · « **Reply #79 on:** January 18, 2015, 03:42:31 am »

Quote from: jasemoney on January 18, 2015, 01:12:12 am

Thanks for reply, sorry i didnt save it. I could make a new one by submitting for SILVR hehe. If it happens again ill post the log but I'm on EGOLD for the moment.

Side thoughts: what happens when the POW is over for these coins? minting up to 1/1000 at a time, I dont see EGOLD or SILVR lasting more than 10-20 days to run through the POW phase.

EGOLD POW phase will last much longer, you should check EGOLD properties

ScripterRon · « **Reply #80 on:** January 18, 2015, 04:07:29 am »

Quote from: MAMLCOM on January 18, 2015, 02:59:06 am

SEVERE Main.main: Exception during program initialization
java.lang.illegalArgumentException:Currency code is not valid

That miner said... I did update NxtMint.conf with currency EGOLD.. and path i think it's right..

It sounds like the currency code is not set in NxtMint.conf.

What operating system are you on? What directory contains NxtMint.conf?

MAMLCOM · « **Reply #81 on:** January 18, 2015, 04:16:57 am »

I dont know.. every time i change NxtMint.conf.. It's always false, like that error..

ScripterRon · « **Reply #82 on:** January 18, 2015, 04:13:18 pm »

Quote from: MAMLCOM on January 18, 2015, 04:16:57 am

I dont know.. every time i change NxtMint.conf.. It's always false, like that error..

I need to know the operating system (Windows, Mac, Linux) and the name of the directory containing NxtMint.conf. Also, post NxtMint.conf so I can look at it (blank out your secret phrase)

Hachoir · « **Reply #83 on:** January 18, 2015, 05:02:34 pm »

Hi I got the minter to work for CPU minting but not for GPU minting, the aparapi installation doesn't work when I type the command "java -jar arapapi.jar", I get the message "no main manifest attribute". So I found an .exe installer and proceeded to execute it. Still doesn't work when I run the minter. I obtained the Opencl.dll designed for my graphics card and copied it in the NxtMint directory. What's left ?

colin012 · « **Reply #84 on:** January 18, 2015, 05:14:22 pm »

I have talked with Ron about setting up encrypted password storage and it looks like I will be working on that feature while he focuses his efforts on the core of the minter.

Here are some things I will include:

Password based encryption using PBKDF2
Key file based encryption with users choice of algorithms. The default Algo will be latice based for security against quantum computing (whenever that is ready). The key file may additionally be encrypted using PBKDF2 or another key file
After those, I will work on built in CD burning so that the key file can be stored on a CD without having to leave the program. If two key files are used, this can be set up so that two CDs or a CD and a key file are needed to unlock the password.
Next, I will work on setting things up so that minting can be run automatically from a CD containing a key file. The CD will hold a link to the directory used in addition to the key file and will run automatically when popped in. If the encrypted password file if present in the directory and can be unlocked by the key file.
Next, I will work on setting up optional password protection for the whole CD.
Finally, for the most advanced users, I will make it so that 2-step authentication is needed to unlock the password. This can be achieved by integrating a special messaging service into the encryption software. I plan to make support for multiple messengers, but the first one I plan to support is i2p Bote. This would require the machine to be running i2p and have the Bote add on installed for the PC and a special app I will designe for the mobile device. It will use a special kind of encryption so that not even the computer sending the message can read it after it is encrypted. Another one I want to support is TextSecure which will send a special encrypted message to the phone that will be used for the 2FA. I am open to other suggestions.

Please note that the 2FA is the lowest priority item on the list as it is pretty much only for the hyper paranoid.

This code will be public domain and may he used for other NXT security implementations. In fact, I will even reward STNLC to anyone who can implant it for general use with NXT to pretty much eliminate use of passwords to log in all together (except for in encrypted form) and add 2FA to general NXT login. (Nxt KeyStash is a likely candidate).

ScripterRon · « **Reply #85 on:** January 18, 2015, 05:43:53 pm »

Quote from: Hachoir on January 18, 2015, 05:02:34 pm

Hi I got the minter to work for CPU minting but not for GPU minting, the aparapi installation doesn't work when I type the command "java -jar arapapi.jar", I get the message "no main manifest attribute". So I found an .exe installer and proceeded to execute it. Still doesn't work when I run the minter. I obtained the Opencl.dll designed for my graphics card and copied it in the NxtMint directory. What's left ?

java -Xmx256m -jar NxtMint-1.0.0.jar

This assumes the Aparapi and OpenCL runtime libraries are in the system execution path (PATH environment variable) or in the current directory (Windows adds the current directory to the execution path

Hachoir · « **Reply #86 on:** January 18, 2015, 06:07:19 pm »

So I don't need to execute aparapi.jar at all, just put aparapi.dll and OpenCl.dll in the same directory in which the minter is executed ?

I get this error.

Quote

SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load ap
arapi native library aparapi_x86 or possibly failed to locate opencl native libr
ary (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD
_LIBRARY_PATH (linux).

PATH environment variable is set with this command ?

set PATH="c:\NxtMint-1.0.0";%PATH%

ltcrstrbrt · « **Reply #87 on:** January 18, 2015, 06:31:44 pm »

I installed SDK for Amd GPU and then copied openCL.dll file( from C:\Program Files (x86)\AMD APP SDK\3.0-0-Beta\bin\x86_64) in nxtminter folder (main folder not nxtminter\lib). I used " java -Xmx256m -jar NxtMint-1.0.0.jar " but doesn't work, also Cpu works fine..

WARNING MintWorker.gpuHash: GPUexecution did not complete probably due to GPU resource shortage
INFO MintWorker.gpuHash: Disabling GPU hashing and reverting to CPU hashing

ScripterRon · « **Reply #88 on:** January 18, 2015, 07:10:50 pm »

Quote from: Hachoir on January 18, 2015, 06:07:19 pm

So I don't need to execute aparapi.jar at all, just put aparapi.dll and OpenCl.dll in the same directory in which the minter is executed ?

I get this error.

Quote
SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load ap
arapi native library aparapi_x86 or possibly failed to locate opencl native libr
ary (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD
_LIBRARY_PATH (linux).

PATH environment variable is set with this command ?

set PATH="c:\NxtMint-1.0.0";%PATH%

That put double quotes in the path. Try this instead:
PATH C:\NxtMint-1.0.0;%PATH%

To check that it worked, enter
PATH

However, on Windows, it isn't necessary to update the PATH since Windows automatically looks in the current directory first. So just copy the dll files to C:\NxtMint-1.0.0 and make it the current directory when you issue the Java command:
cd C:\NxtMint-1.0.0
java -Xmx256m -jar NxtMint-1.0.0.jar

What card are you using? NVIDIA ships OpenCL.dll and OpenCL64.dll. If you are using the 64-bit JVM on Windows, you need to copy OpenCL64.dll and rename it to OpenCL.dll.

ScripterRon · « **Reply #89 on:** January 18, 2015, 07:14:47 pm »

Quote from: ltcrstrbrt on January 18, 2015, 06:31:44 pm

I installed SDK for Amd GPU and then copied openCL.dll file( from C:\Program Files (x86)\AMD APP SDK\3.0-0-Beta\bin\x86_64) in nxtminter folder (main folder not nxtminter\lib). I used " java -Xmx256m -jar NxtMint-1.0.0.jar " but doesn't work, also Cpu works fine..

WARNING MintWorker.gpuHash: GPUexecution did not complete probably due to GPU resource shortage
INFO MintWorker.gpuHash: Disabling GPU hashing and reverting to CPU hashing

There should have been an error message from Aparapi before the NxtMint warning message. This message explains why it didn't execute on the GPU (the only information returned to NxtMint is whether the request was executed on the GPU or in a Java thread pool).

Hachoir · « **Reply #90 on:** January 18, 2015, 07:48:40 pm »

Quote from: ScripterRon on January 18, 2015, 07:10:50 pm

Quote from: Hachoir on January 18, 2015, 06:07:19 pm
So I don't need to execute aparapi.jar at all, just put aparapi.dll and OpenCl.dll in the same directory in which the minter is executed ?

I get this error.

Quote
SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load ap
arapi native library aparapi_x86 or possibly failed to locate opencl native libr
ary (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD
_LIBRARY_PATH (linux).

PATH environment variable is set with this command ?

set PATH="c:\NxtMint-1.0.0";%PATH%
That put double quotes in the path. Try this instead:
PATH C:\NxtMint-1.0.0;%PATH%

To check that it worked, enter
PATH

However, on Windows, it isn't necessary to update the PATH since Windows automatically looks in the current directory first. So just copy the dll files to C:\NxtMint-1.0.0 and make it the current directory when you issue the Java command:
cd C:\NxtMint-1.0.0
j
What card are you using? NVIDIA ships OpenCL.dll and OpenCL64.dll. If you are using the 64-bit JVM on Windows, you need to copy OpenCL64.dll and rename it to OpenCL.dll.

I use the command java -Xmx256m -Dnxt.datadir="C:\NxtMint-1.0.0" -jar "C:\NxtMint-1.0.0\NxtMint-1.0.0.jar" because when I run ava -Xmx256m -jar NxtMint-1.0.0.jar from the current directory I get an error

Quote

08:43:09 SEVERE Main.main: Exception during program initialization
java.lang.IllegalArgumentException: Secret phrase not specified
at org.ScripterRon.NxtMint.Main.main(Main.java:186)

Even though the secret phrase is specified in NxtMint.conf which I copied in C:\Users\AppData\NxtMint-1.0.0 (that NxtMint-1.0.0 was not created automatically) alongside logging.properties.

My graphics card is nvidia geforce 6150se. The OS isn't 64 bit so the file already is named OpenCl.dll. I did copy it in the current directory with aparapi_x86.dll.

So I still set the PATH variable and I get the same message.

Quote

SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load ap
arapi native library aparapi_x86 or possibly failed to locate opencl native libr
ary (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD
_LIBRARY_PATH (linux).

ScripterRon · « **Reply #91 on:** January 18, 2015, 09:01:05 pm »

Quote from: Hachoir on January 18, 2015, 07:48:40 pm

I use the command java -Xmx256m -Dnxt.datadir="C:\NxtMint-1.0.0" -jar "C:\NxtMint-1.0.0\NxtMint-1.0.0.jar" because when I run ava -Xmx256m -jar NxtMint-1.0.0.jar from the current directory I get an error
Quote
08:43:09 SEVERE Main.main: Exception during program initialization
java.lang.IllegalArgumentException: Secret phrase not specified
at org.ScripterRon.NxtMint.Main.main(Main.java:186)
Even though the secret phrase is specified in NxtMint.conf which I copied in C:\Users\AppData\NxtMint-1.0.0 (that NxtMint-1.0.0 was not created automatically) alongside logging.properties.

My graphics card is nvidia geforce 6150se. The OS isn't 64 bit so the file already is named OpenCl.dll. I did copy it in the current directory with aparapi_x86.dll.

So I still set the PATH variable and I get the same message.

Quote
SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load ap
arapi native library aparapi_x86 or possibly failed to locate opencl native libr
ary (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD
_LIBRARY_PATH (linux).

The default data directory on Windows is C:\Users\your-user-name\AppData\Roaming\NxtMint

I don't know why it is failing. What does "java -version" display?

ScripterRon · « **Reply #92 on:** January 18, 2015, 09:11:36 pm »

NxtMint Version 1.1.0 is available.

Change log
- Add sample.logging.properties, sample.mint.bat, sample.mint.sh
- Add minting counter to transaction table in GUI
- Rebuild GPU kernel following execution failure
- Add total hash count to worker status message
- Check for unconfirmed transactions
- GPU support for SHA-256

I'm looking into GPU support for Scrypt but this one is a bit more difficult due to memory requirements. Graphics cards with a decent amount of memory will probably work (Scrypt needs 128KB of memory for each execution instance to avoid recomputing previous hash digests for each hash round). So 256 instances would require 32 MB of graphics card memory just for the data array. Since this is temporary storage, it doesn't need to be preserved between kernel executions.

The actual algorithm is more complex as well since it uses multiple HmacSha256 hashes for each round.

crimi · « **Reply #93 on:** January 18, 2015, 09:39:46 pm »

Quote from: ScripterRon on January 18, 2015, 09:11:36 pm

NxtMint Version 1.1.0 is available.

Change log
- Add sample.logging.properties, sample.mint.bat, sample.mint.sh
- Add minting counter to transaction table in GUI
- Rebuild GPU kernel following execution failure
- Add total hash count to worker status message
- Check for unconfirmed transactions
- GPU support for SHA-256

I'm looking into GPU support for Scrypt but this one is a bit more difficult due to memory requirements. Graphics cards with a decent amount of memory will probably work (Scrypt needs 128KB of memory for each execution instance to avoid recomputing previous hash digests for each hash round). So 256 instances would require 32 MB of graphics card memory just for the data array. Since this is temporary storage, it doesn't need to be preserved between kernel executions.

The actual algorithm is more complex as well since it uses multiple HmacSha256 hashes for each round.

Ty works like a charm(EGOLD), if you need someone to test multi gpu send it over to me.

Hachoir · « **Reply #94 on:** January 18, 2015, 09:52:41 pm »

Quote from: ScripterRon on January 18, 2015, 09:01:05 pm

Quote from: Hachoir on January 18, 2015, 07:48:40 pm
I use the command java -Xmx256m -Dnxt.datadir="C:\NxtMint-1.0.0" -jar "C:\NxtMint-1.0.0\NxtMint-1.0.0.jar" because when I run ava -Xmx256m -jar NxtMint-1.0.0.jar from the current directory I get an error
Quote
08:43:09 SEVERE Main.main: Exception during program initialization
java.lang.IllegalArgumentException: Secret phrase not specified
at org.ScripterRon.NxtMint.Main.main(Main.java:186)
Even though the secret phrase is specified in NxtMint.conf which I copied in C:\Users\AppData\NxtMint-1.0.0 (that NxtMint-1.0.0 was not created automatically) alongside logging.properties.

My graphics card is nvidia geforce 6150se. The OS isn't 64 bit so the file already is named OpenCl.dll. I did copy it in the current directory with aparapi_x86.dll.

So I still set the PATH variable and I get the same message.

Quote
SEVERE OpenCLLoader.<clinit>: Check your environment. Failed to load ap
arapi native library aparapi_x86 or possibly failed to locate opencl native libr
ary (opencl.dll/opencl.so). Ensure that both are in your PATH (windows) or in LD
_LIBRARY_PATH (linux).
The default data directory on Windows is C:\Users\your-user-name\AppData\Roaming\NxtMint

I don't know why it is failing. What does "java -version" display?

Well it shouldn't matter since I manually changed the data directory anyway with all the needed files into it (for GPU minting I just need OpenCL.dll and aparapi_x86.dll right ? Do I need libaparapi_x86.so too ?).
java -version says :

Quote

java version "1.8.0_25"
Java(TM) SE Runtime Environment (build 1.8.0_25-b18)
Java HotSpot(TM) Client VM (build 25.25-b02, mixed mode, sharing)

I'll try with your latest release too.

Hachoir · « **Reply #95 on:** January 18, 2015, 10:28:03 pm »

Still the same with version 1.1.0. It can't find OpenCL.dll and aparapi_x86.dll while it's in the same directory ?

ScripterRon · « **Reply #96 on:** January 18, 2015, 11:16:40 pm »

Quote from: Hachoir on January 18, 2015, 10:28:03 pm

Still the same with version 1.1.0. It can't find OpenCL.dll and aparapi_x86.dll while it's in the same directory ?

You don't need libaparapi.so since that is for Linux.

It sounds like Aparapi/OpenCL doesn't work on your system. I have 64-bit Windows 8, so I can't try the 32-bit version. It could also be that your graphics card doesn't support Aparapi. You could try adding some Aparapi debug options to the command line, although I'm not sure it will make any difference if it is failing to load the dll.

-Dcom.amd.aparapi.logLevel=FINE
-Dcom.amd.aparapi.enableVerboseJNI=true

Daedelus · « **Reply #97 on:** January 19, 2015, 01:01:26 pm »

I have started a thread on MSCoin mining: https://bitcointalk.org/index.php?topic=929548

I mentioned NxtMint for GPU miners. Can you keep an eye on it to help me out with any questioning?

Thanks in advance

ScripterRon · « **Reply #98 on:** January 19, 2015, 03:07:27 pm »

Quote from: Daedelus on January 19, 2015, 01:01:26 pm

I have started a thread on MSCoin mining: https://bitcointalk.org/index.php?topic=929548

I mentioned NxtMint for GPU miners. Can you keep an eye on it to help me out with any questioning?

Thanks in advance

OK - I added it to my watch list on BitcoinTalk

Daedelus · « **Reply #99 on:** January 19, 2015, 03:14:02 pm »

Great, thanks. 84 views and not a bite. They are probably all views from the people on this forum

We'll see, the altcoin mining sub forum looks pretty dead (less than 1 post per hour atm) so may try something else later.

ScripterRon · « **Reply #100 on:** January 19, 2015, 03:27:30 pm »

Quote from: crimi on January 18, 2015, 09:39:46 pm

Ty works like a charm(EGOLD), if you need someone to test multi gpu send it over to me.

Multiple GPU support is next on my list. Since I have a single GPU, this means you will be an alpha tester

. Device range support is something that Aparapi recently added, so you might end up debugging their code as well.

The SCRYPT support is looking promising. I'm almost done with the HmacSha256 support which is required for Scrypt. Once that is working, Scrypt itself should be quick since it is just a loop with multiple HmacSha256 hash requests in each loop. The time-consuming part is testing each piece in isolation since there are no debug message while in kernel mode and there are a lot of subtle gotchas using multiple address spaces on the GPU (for global vs private data).

cryptoventurefund · « **Reply #101 on:** January 19, 2015, 07:00:19 pm »

ScripterRon i couldn't get the first version to work for some strange reason, but 1.1 did the trick.

It's amazing!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

You sir, rock!

ScripterRon · « **Reply #102 on:** January 19, 2015, 07:57:59 pm »

SCRYPT support looks promising. I need to go through the code and try to reduce the path length as much as possible. Then I'll let it mint for awhile to shake out any bugs. Note that you'll need a graphics card with a fair amount of memory since the hash data will take 130 KB per kernel group instance. So if your card supports the maximum of 256 instances per group, that would around 32 MB of data.

ScripterRon · « **Reply #103 on:** January 19, 2015, 11:08:54 pm »

Quote from: ScripterRon on January 19, 2015, 07:57:59 pm

SCRYPT support looks promising. I need to go through the code and try to reduce the path length as much as possible. Then I'll let it mint for awhile to shake out any bugs. Note that you'll need a graphics card with a fair amount of memory since the hash data will take 130 KB per kernel group instance. So if your card supports the maximum of 256 instances per group, that would around 32 MB of data.

Well, SCRYPT mining is working and the NRS server accepts the minting transactions. But it brings my NVIDIA card to its knees and isn't much faster than just using the CPU. Looking at some comparison charts, AMD is much better than NVIDIA for SCRYPT mining. If somebody wants to try this on their system, let me know and I'll send them the jar.

I'm not sure that I can improve performance because SCRYPT has very large hash loops (especially the salsa rounds).

KrusherPT · « **Reply #104 on:** January 19, 2015, 11:37:11 pm »

You can send it to me I can test, I'm a bit lame so anything else I need to do, or just paste on the 1.1 folder and change the exec command...

ScripterRon · « **Reply #105 on:** January 20, 2015, 02:45:41 am »

I was getting a JVM crash attempting to use local memory on the graphics card. So I updated my NVIDIA driver to see if that would fix the problem. Nope - just made it worse since now I can't use private memory either. The only thing that works is global memory. Reverting to the old NVIDIA driver would get me running again but I don't want to tell people they have to downlevel their systems to get NxtMint to work. So I'm going to stay with the current NVIDIA driver (the Keccak25 and SHA-256 GPU code still works since they use just global memory).

This all started as I tried to fix the SCRYPT performance problem. The problem was caused by using private memory for the huge digest array. Changing that to global memory gave a 10x performance boost. But global memory is shared by all kernel instances, so that isn't a solution to instance-dependent data unless you have a small number of instances.

The answer is probably to access the GPU from native C/C++ code instead of using Aparapi. But that leads to platform dependencies when building and testing the code. I'll look around the web and see if anything is available.

Wakasaki808 · « **Reply #106 on:** January 20, 2015, 06:24:03 am »

Not sure if anybody else has had a problem with this...

But I can't get the NxtMint to actually use my Nvidia GPU...using my laptop it always goes to my Intel HD Graphics and not my Nvidia GPU. (It shows a 'GPU' is being used in the cmd prompt but just not the correct GPU) I can't load it as a program because its not an .exe file and when I set .jar to run using the dedicated card through control settings it doesn't do anything...

Guarantee its because of my 'Nvidia Control Panel' and the way Nvidia specifies what programs to use what GPU. But still haven't been able to get the NxtMint to actually use the Nvidia after playing with settings for a few hours. There is likely a way to fix this, can't figure it out though.

ScripterRon · « **Reply #107 on:** January 20, 2015, 06:29:19 am »

Quote from: Wakasaki808 on January 20, 2015, 06:24:03 am

But I can't get the NxtMint to actually use my Nvidia GPU...using my laptop it always goes to my Intel HD Graphics and not my Nvidia 860. (It shows a 'GPU' is being used in the cmd prompt but just not the correct GPU) I can't load it as a program because its not an .exe file and when I set .jar to run using the dedicated card through control settings it doesn't do anything...

Aparapi uses the first GPU that it finds. I'm going to see if I can let you specify which GPU to use. If that is successful, then I should also be able to execute an OpenCL kernel on more than one GPU.

lopalcar · « **Reply #108 on:** January 20, 2015, 08:16:52 am »

Quote from: Wakasaki808 on January 20, 2015, 06:24:03 am

Not sure if anybody else has had a problem with this...

But I can't get the NxtMint to actually use my Nvidia GPU...using my laptop it always goes to my Intel HD Graphics and not my Nvidia GPU. (It shows a 'GPU' is being used in the cmd prompt but just not the correct GPU) I can't load it as a program because its not an .exe file and when I set .jar to run using the dedicated card through control settings it doesn't do anything...

Guarantee its because of my 'Nvidia Control Panel' and the way Nvidia specifies what programs to use what GPU. But still haven't been able to get the NxtMint to actually use the Nvidia after playing with settings for a few hours. There is likely a way to fix this, can't figure it out though.

Try to disable the integrated GPU from the BIOS

794259332 · « **Reply #109 on:** January 20, 2015, 03:21:36 pm »

only one gpu can mining? how about setting for multi gpu cards

ScripterRon · « **Reply #110 on:** January 20, 2015, 04:42:26 pm »

I got local memory working on my NVidia GT 720 and moved the Scrypt salsa array from global memory to local memory. This gave a 5x performance improvement. If you want to try a pre-release, you can download an updated NxtMint-1.1.0.jar from https://drive.google.com/open?id=0B1312_6UqRHPQmhXWWdSdzNzWUk&authuser=0.

KrusherPT · « **Reply #111 on:** January 20, 2015, 05:32:46 pm »

I'm getting 0.0038 ~ 0.0045 MH/s on a AMD 7770 at I=125 and GPU 99%, desktop lags alot.
And 1 core from my amd fx 6300 its 0.0021, so still slow.
But good job anyway, you are a hero for doing the miner

ScripterRon · « **Reply #112 on:** January 20, 2015, 06:34:05 pm »

Quote from: KrusherPT on January 20, 2015, 05:32:46 pm

I'm getting 0.0038 ~ 0.0045 MH/s on a AMD 7770 at I=125 and GPU 99%, desktop lags alot.
And 1 core from my amd fx 6300 its 0.0021, so still slow.
But good job anyway, you are a hero for doing the miner

Scrypt has a very long instruction path length compared to other algorithms (you can see this by comparing the CPU hash rates for the various algorithms). This leads to the sluggish desktop problem because the GPU is busy for an extended period for each kernel execution.

I'm sure the bottleneck is the salsa permutations (which consist of a bunch of shifts, xor and swaps). I get around 0.0080 MH/s on an NVIDIA GT 720, which still isn't very good.

By the way, Aparapi provides a little 'info' utility in their samples. You can run it to display your device configuration by running 'info.bat' or 'info.sh'

My configuration:

Machine contains 1 OpenCL platforms
Platform 0{
Name : "NVIDIA CUDA"
Vendor : "NVIDIA Corporation"
Version : "OpenCL 1.1 CUDA 7.0.18"
Platform contains 1 OpenCL devices
Device 0{
Type : GPU
GlobalMemSize : 1073741824
LocalMemSize : 49151
MaxComputeUnits : 2
MaxWorkGroupSizes : 1024
MaxWorkItemDimensions : 3
}
}

ScripterRon · « **Reply #113 on:** January 20, 2015, 07:42:26 pm »

Scrypt is getting better. Up to 0.0095 MH/s (Worker 4 is my GPU worker). The desktop is more responsive (I can actually enter data while the minter is running now). New test version out on Google Drive.

Code: [Select]

02:36:49 INFO Main.main: Java Nxt Mint Version 1.1.0
02:36:49 INFO Main.main: Application data path: C:\Users\Ronald\Appdata\Roaming\NxtMint
02:36:49 INFO Main.main: Using Nxt node at scripterron.dyndns.biz:7876
02:36:49 INFO Main.main: Minting 5.000000 units of ELEMS for account NXT-XM86-4ZNA-65L5-CD
WUE: 4 CPU threads, 10 GPU intensity
02:36:49 INFO Nxt.init: API node=scripterron.dyndns.biz, API port=7876
02:36:51 FINE GpuScrypt.<init>: SCRYPT GPU local size 256, group size 512
02:36:51 FINE MintWorker.run: Worker 2 starting on counter 10
02:36:51 FINE MintWorker.run: Worker 1 starting on counter 10
02:36:51 FINE MintWorker.run: Worker 0 starting on counter 10
02:36:51 FINE MintWorker.run: Worker 3 starting on counter 10
02:36:51 FINE MintWorker.run: Worker 4 starting on counter 10
02:37:51 FINE MintWorker.run: Worker 2: 0.15 MHash, 0.0025 MHash/s
02:37:51 FINE MintWorker.run: Worker 0: 0.15 MHash, 0.0025 MHash/s
02:37:51 FINE MintWorker.run: Worker 3: 0.15 MHash, 0.0025 MHash/s
02:37:51 FINE MintWorker.run: Worker 1: 0.15 MHash, 0.0025 MHash/s
02:37:52 FINE MintWorker.run: Worker 4: 0.57 MHash, 0.0096 MHash/s

Wakasaki808 · « **Reply #114 on:** January 20, 2015, 07:44:07 pm »

Quote from: ScripterRon on January 20, 2015, 07:42:26 pm

Scrypt is getting better. Up to 0.95 MH/s (Worker 4 is my GPU worker). The desktop is more responsive (I can actually enter data while the minter is running now). New version out on Google Drive.
Code: [Select]
02:36:49 INFO Main.main: Java Nxt Mint Version 1.1.0 02:36:49 INFO Main.main: Application data path: C:\Users\Ronald\Appdata\Roaming\NxtMint 02:36:49 INFO Main.main: Using Nxt node at scripterron.dyndns.biz:7876 02:36:49 INFO Main.main: Minting 5.000000 units of ELEMS for account NXT-XM86-4ZNA-65L5-CD WUE: 4 CPU threads, 10 GPU intensity 02:36:49 INFO Nxt.init: API node=scripterron.dyndns.biz, API port=7876 02:36:51 FINE GpuScrypt.<init>: SCRYPT GPU local size 256, group size 512 02:36:51 FINE MintWorker.run: Worker 2 starting on counter 10 02:36:51 FINE MintWorker.run: Worker 1 starting on counter 10 02:36:51 FINE MintWorker.run: Worker 0 starting on counter 10 02:36:51 FINE MintWorker.run: Worker 3 starting on counter 10 02:36:51 FINE MintWorker.run: Worker 4 starting on counter 10 02:37:51 FINE MintWorker.run: Worker 2: 0.15 MHash, 0.0025 MHash/s 02:37:51 FINE MintWorker.run: Worker 0: 0.15 MHash, 0.0025 MHash/s 02:37:51 FINE MintWorker.run: Worker 3: 0.15 MHash, 0.0025 MHash/s 02:37:51 FINE MintWorker.run: Worker 1: 0.15 MHash, 0.0025 MHash/s 02:37:52 FINE MintWorker.run: Worker 4: 0.57 MHash, 0.0096 MHash/s

Nice to see this progressing

ScripterRon · « **Reply #115 on:** January 21, 2015, 01:45:42 am »

A new test version of NxtMint is available at https://drive.google.com/open?id=0B1312_6UqRHPQmhXWWdSdzNzWUk&authuser=0.

I have added support for specifying one or more GPU devices to be used for minting. Since I have just one graphics card, I'm unable to test this. OpenCL does not provide a method to tie an OpenCL device to a physical adapter (each card maker has its own OpenCL plugins but these are not available to my Java program). Even the OpenCL device identifier changes each time you start your program. So you will have to identify the adapters to be used based on the enumeration order (first adapter listed is 0, second adapter is 1, and so forth). To help you identify what is available. NxtMint will list all available GPU devices when it is started with a non-zero value for gpuIntensity. Hopefully the order will remain the same unless you add/remove cards from your system.

The gpuDevice configuration parameter specifies the device to be used and can be repeated multiple times to use more than one adapter.

Jimmy2011 · « **Reply #116 on:** January 21, 2015, 01:58:44 am »

I'm getting the following results on my GTX 580 Nvidia GPU with ScryptGPUV1 and ScryptGPUV2(the latest version):

Code: [Select]

cpuThreads = 0
GPU local size 256, group size 1024
ScryptGPUV1 = 0.0035 Mhash/s

Code: [Select]

cpuThreads = 0
GPU local size 256, group size 512
ScryptGPUV2 = 0.0027 Mhash/s

Code: [Select]

cpuThreads = 4
GPU local size 256, group size 512
CPUWorker = 0.0014 Mhash/s for each CPU worker
ScryptGPUV1 = 0.0027 Mhash/s
totalHashRate ~ 0.0083 MH/s

Code: [Select]

cpuThreads = 4
GPU local size 256, group size 512
CPUWorker = 0.0014 Mhash/s for each CPU worker
ScryptGPUV2 = 0.0027 Mhash/s
totalHashRate ~ 0.0083 MH/s

Code: [Select]

cpuThreads = 6
GPU = 0
CPUWorker = 0.0014 Mhash/s for each CPU worker
totalHashRate ~ 0.0084 MH/s

Note that GPU temperature is only ~66 dc for all the above cases with GPULoad =99%, however, it is more than 70 dc when mining EGOLD. I have two processors with total 8 threads in my machine.

ScripterRon · « **Reply #117 on:** January 21, 2015, 04:36:38 am »

Quote from: Jimmy2011 on January 21, 2015, 01:58:44 am

I'm getting the following results on my GTX 580 Nvidia GPU with ScryptGPUV1 and ScryptGPUV2(the latest version):

Note that GPU temperature is only ~66 dc for all the above cases with GPULoad =99%, however, it is more than 70 dc when mining EGOLD. I have two processors with total 8 threads in my machine.

It looks like your GPU has more than 2 compute units since the hash rate went down when I reduced the group count from 4 to 2 (1024 global instances to 512 global instances with 256 instances per compute unit). The code I just uploaded to Google Drive will display GPU information during NxtMint initialization, so you can see how many compute units are available on your card. Each compute unit has multiple cores and its own local memory. For example, my card has 384 CUDA cores and 2 compute units, which means that each compute unit has 192 cores.

I'll change the code to dynamically allocate the global size based on the number of compute units (right now it is a fixed initialization). Still, I wonder why you are only getting 0.0027MH/s. Perhaps your local memory is too small (that will also be displayed when NxtMint starts up). I need 128 bytes of local memory for each instance or 32KB for 256 instances (my card has 48KB of local memory per compute unit). So I might need to adjust the local size as well if enough memory isn't available.

Aparapi has an upper limit of 256 instances in a work group. So if a compute unit has more than 256 cores, the extra cores won't be used.

Jimmy2011 · « **Reply #118 on:** January 21, 2015, 09:22:51 am »

Quote from: ScripterRon on January 21, 2015, 04:36:38 am

Quote from: Jimmy2011 on January 21, 2015, 01:58:44 am
I'm getting the following results on my GTX 580 Nvidia GPU with ScryptGPUV1 and ScryptGPUV2(the latest version):

Note that GPU temperature is only ~66 dc for all the above cases with GPULoad =99%, however, it is more than 70 dc when mining EGOLD. I have two processors with total 8 threads in my machine.
It looks like your GPU has more than 2 compute units since the hash rate went down when I reduced the group count from 4 to 2 (1024 global instances to 512 global instances with 256 instances per compute unit). The code I just uploaded to Google Drive will display GPU information during NxtMint initialization, so you can see how many compute units are available on your card. Each compute unit has multiple cores and its own local memory. For example, my card has 384 CUDA cores and 2 compute units, which means that each compute unit has 192 cores.

I'll change the code to dynamically allocate the global size based on the number of compute units (right now it is a fixed initialization). Still, I wonder why you are only getting 0.0027MH/s. Perhaps your local memory is too small (that will also be displayed when NxtMint starts up). I need 128 bytes of local memory for each instance or 32KB for 256 instances (my card has 48KB of local memory per compute unit). So I might need to adjust the local size as well if enough memory isn't available.

Aparapi has an upper limit of 256 instances in a work group. So if a compute unit has more than 256 cores, the extra cores won't be used.

I have 16 compute units as following. BTW, it seems that the latest version has poor performance as minting EGOLD compared with the public released 1.1.0 version.

Code: [Select]

05:17:34 INFO Main.buildGpuList: GPU device 0: NVIDIA CUDA, 3072MB global memory
, 48KB local memory, 16 compute units

794259332 · « **Reply #119 on:** January 21, 2015, 10:08:34 am »

how to set up for 2 gpu cards

z38630610 · « **Reply #120 on:** January 21, 2015, 11:27:22 am »

Quote from: 794259332 on January 21, 2015, 10:08:34 am

how to set up for 2 gpu cards

-d 0,1

lopalcar · « **Reply #121 on:** January 21, 2015, 11:41:37 am »

Quote from: 794259332 on January 21, 2015, 10:08:34 am

how to set up for 2 gpu cards

Using the latest NxtMint-1.1.0.jar hosted in google drive, edit NxtMint.conf adding:
gpuDevice=X

where X is a number betwen 0 and (NumberOfGPUs-1)
You need to write that line as many times as devices you want to run specifying the X for each GPU

lopalcar · « **Reply #122 on:** January 21, 2015, 11:44:40 am »

Quote from: Jimmy2011 on January 21, 2015, 09:22:51 am

I have 16 compute units as following. BTW, it seems that the latest version has poor performance as minting EGOLD compared with the public released 1.1.0 version.

Code: [Select]
05:17:34 INFO Main.buildGpuList: GPU device 0: NVIDIA CUDA, 3072MB global memory , 48KB local memory, 16 compute units

I also have poor performance, in a 6990, each GPU now gives me 45mhash and before it was giving me 100mhash, so I don't gain anything using the whole card
Each GPU has 1536 cores and 24 compute units

crimi · « **Reply #123 on:** January 21, 2015, 02:55:03 pm »

Got 2 cards running now works fine so far.

So if minting is like a big sudoku looking for a solution, would it make sense just to stop the puzzle and start a new one if you think it going to take to long.
Sometimes i need 1 mio hashes sometimes only 30k hashes atm to find 200 units egold.

ScripterRon · « **Reply #124 on:** January 21, 2015, 03:09:59 pm »

Quote from: Jimmy2011 on January 21, 2015, 09:22:51 am

I have 16 compute units as following. BTW, it seems that the latest version has poor performance as minting EGOLD compared with the public released 1.1.0 version.

Code: [Select]
05:17:34 INFO Main.buildGpuList: GPU device 0: NVIDIA CUDA, 3072MB global memory , 48KB local memory, 16 compute units

Hmm - I didn't change the Keccak25 or Sha256 code. I'll rerun some Keccak25 and Sha256 minting and see how it looks.

16 compute units is outstanding (wish I had one

. So even the old code which used 4 compute units wasn't stressing your card. The change to configure dynamically should take care of that problem. However, you will probably have to increase the Java heap since each instance requires 128KB of global memory (which is mapped to Java memory). Each compute unit would run 256 instances, so 16*256*128KB=512MB.

ScripterRon · « **Reply #125 on:** January 21, 2015, 03:11:40 pm »

Quote from: 794259332 on January 21, 2015, 10:08:34 am

how to set up for 2 gpu cards

Add a gpuDevice parameter to NxtMint.conf. The default is gpuDevice=0 if it isn't specified (that is, use the first GPU found)

For example, to use the first two GPU devices, add:
gpuDevice=0
gpuDevice=1

ScripterRon · « **Reply #126 on:** January 21, 2015, 03:18:11 pm »

Quote from: lopalcar on January 21, 2015, 11:44:40 am

I also have poor performance, in a 6990, each GPU now gives me 45mhash and before it was giving me 100mhash, so I don't gain anything using the whole card
Each GPU has 1536 cores and 24 compute units

OK - I'm guessing it isn't using all cores in a compute unit. That could be a result of the changes for Scrypt. I'll look into it.

24 compute units will be even worse then 16 as far as Java memory is concerned. I wish I could find out how many cores are available but Aparapi doesn't return that information. So I have to assume it is 256 per compute unit but in your case it would be 64 (1536/24). I guess I can add it as a configuration parameter and assume 256 if it is not specified. That should improve performance further and reduce memory requirements (both in Java and on the graphics card).

lopalcar · « **Reply #127 on:** January 21, 2015, 03:45:53 pm »

Quote from: ScripterRon on January 21, 2015, 03:18:11 pm

Quote from: lopalcar on January 21, 2015, 11:44:40 am
I also have poor performance, in a 6990, each GPU now gives me 45mhash and before it was giving me 100mhash, so I don't gain anything using the whole card
Each GPU has 1536 cores and 24 compute units
OK - I'm guessing it isn't using all cores in a compute unit. That could be a result of the changes for Scrypt. I'll look into it.

24 compute units will be even worse then 16 as far as Java memory is concerned. I wish I could find out how many cores are available but Aparapi doesn't return that information. So I have to assume it is 256 per compute unit but in your case it would be 64 (1536/24). I guess I can add it as a configuration parameter and assume 256 if it is not specified. That should improve performance further and reduce memory requirements (both in Java and on the graphics card).

Ok thank you

I don't know what memory are you refearing to... I ran the jar file with -Xmx1024m, either the RAM memory nor the GPU memory are stressed... I also know that the GPU is at 99% of use, but makes less noise than before, so I suppose isn't using all the available cores... "I don't really have idea how the GPU usage is measured then :S "
The GPU memory load is 346MB, same as IDLE, and the RAM from the jar file 184MB
Hope this info helps you in something
Ah, I'm mining EGOLD also, so I don't understand how the scrypt changes are influencing here

Jimmy2011 · « **Reply #128 on:** January 21, 2015, 04:05:23 pm »

I think GTX 580 has 16 compute units with total 512 cores, 32 cores for each unit, right?

Jimmy2011 · « **Reply #129 on:** January 21, 2015, 04:06:15 pm »

Quote from: ScripterRon on January 21, 2015, 03:18:11 pm

I guess I can add it as a configuration parameter and assume 256 if it is not specified. That should improve performance further and reduce memory requirements (both in Java and on the graphics card).

It will be great.

ScripterRon · « **Reply #130 on:** January 21, 2015, 05:07:30 pm »

Quote from: lopalcar on January 21, 2015, 03:45:53 pm

Ok thank you
I don't know what memory are you refearing to... I ran the jar file with -Xmx1024m, either the RAM memory nor the GPU memory are stressed... I also know that the GPU is at 99% of use, but makes less noise than before, so I suppose isn't using all the available cores... "I don't really have idea how the GPU usage is measured then :S "
The GPU memory load is 346MB, same as IDLE, and the RAM from the jar file 184MB
Hope this info helps you in something
Ah, I'm mining EGOLD also, so I don't understand how the scrypt changes are influencing here

I was referring to the local memory on the card. Each compute unit has its own high-speed local memory. But you have plenty, so that isn't a problem.

I changed the mint worker to calculate the workgroup range before executing the kernel. This is done for all algorithms, so it would affect Keccak25 and Sha256. I think the range call is causing the Aparapi kernel code to reset cached data. I'll change this to calculate the range during initialization and save the result. That should get us back to where we were before the Scrypt changes.

ScripterRon · « **Reply #131 on:** January 21, 2015, 05:12:00 pm »

Quote from: Jimmy2011 on January 21, 2015, 04:05:23 pm

I think GTX 580 has 16 compute units with total 512 cores, 32 cores for each unit, right?

The spec sheet says 512 but you should check the device information for your card since card manufacturers can improve the base specs from NVIDIA (overclocking, extra compute units, etc). For example, the spec sheet for the GT 720 has 192 cores. But the card manufacturer placed two compute units on the card, giving me 384 cores and 2 compute units.

On Windows, the NVIDIA Control Panel will show the number of cores in the System Information section.

ScripterRon · « **Reply #132 on:** January 21, 2015, 07:54:37 pm »

OK - new test version on Google Drive. Hopefully, Keccak25 and Sha256 hash rates will be back to where they were.

The gpuDevice configuration parameter now has 2 fields - the first one is the device number (0, 1, etc) and the second one is the total number of processing cores on the graphics card. For example, if GPU 0 has 384 cores, you would specify
gpuDevice=0,384

For Scrypt, the memory allocation is approximately 128KB per processing core. So a system with 512 cores will require 64MB of Java heap and 64MB of adapter memory. If you run out of Java heap memory, increase the value for the -Xmx parameter on the Java command line.

ScripterRon · « **Reply #133 on:** January 21, 2015, 08:56:37 pm »

I did some testing and performance is actually better if you don't specify the number of kernel instances per compute unit. I think this happens because OpenCL will execute all of the requested kernel instances before returning control to Aparapi, which is faster than having Aparapi initiate additional executions.

So, just specify 'gpuDevice=n' and NxtMint will use all of the compute units on the card with 256 kernel instances per compute unit. If this causes a resource shortage on the card, then you can reduce the number of kernel instances per compute unit by specifying the number of processor cores as 'gpuDevice=n,cores'. The kernel instances per compute unit will then be calculated as (the number of processing cores) / (number of compute units).

Note that SCRYPT requires 128KB per kernel instance. The total number of instances is (instances per compute unit) x (number of compute units). So 24 compute units with 256 instances per compute unit would require 24 * 256 * 128KB or 768MB, in both the Java heap and adapter memory.

Jimmy2011 · « **Reply #134 on:** January 22, 2015, 02:44:45 am »

The latest version is so great. 10X speed up compared with last versio!

I confirmed my GPU has 512 cores.

The following results show that the number of instances affect the performance a lot. The hash rate is so low if each core runs only 1 instances, however it is ~ 8X speed up if each core runs the default 8 instances ( 32 cores a unit, total default 32*8=256 instances a unit, that is total 16*32*8=4096 instances). On the other hand, it doesn't support "gpuDevice = 0, 8192" when I tried to set 16 instances for each core. BTW, what does "gpuIntensity" mean if gpuDevice parameter is used?

Code: [Select]

gupDevice=0,512
HR = 0.0025MH/s

Code: [Select]

#gupDevice=0,512
HR = 0.019MH/s

ScripterRon · « **Reply #135 on:** January 22, 2015, 04:02:10 am »

Quote from: Jimmy2011 on January 22, 2015, 02:44:45 am

The latest version is so great. 10X speed up compared with last versio!

I confirmed my GPU has 512 cores.

The following results show that the number of instances affect the performance a lot. The hash rate is so low if each core runs only 1 instances, however it is ~ 8X speed up if each core runs the default 8 instances ( 32 cores a unit, total default 32*8=256 instances a unit, that is total 16*32*8=4096 instances). On the other hand, it doesn't support "gpuDevice = 0, 8192" when I tried to set 16 instances for each core. BTW, what does "gpuIntensity" mean if gpuDevice parameter is used?

Code: [Select]
gupDevice=0,512 HR = 0.0025MH/s
Code: [Select]
#gupDevice=0,512 HR = 0.019MH/s

The device reports maximum sizes to Aparapi, which verifies the requested execution range is valid. I'll add the maximum item size and maximum group size to the GPU information that NxtMint displays when it starts up.

gpuIntensity determines the number of kernel executions before control is returned to Java. As you increase the intensity, the card is busier but display performance will begin to suffer since display updates get queued behind the minting kernel executions.

I think we both arrived at the same conclusion on number of work items per work group. My performance increased if I had 8 or 16 work items per work group. The performance decreased if the number went below 4 or above 16. I think the problem results from memory cache contention. Scrypt maintains 128 bytes of state information for each instance and this data is read/written continuously during the hashing.

Try the NxtMint-1.1.0.jar that I just placed on Google Drive. It sets the local size (number of work items) to 8 and sets the global size (number of work groups per execution) to 256 x (number of compute units). See if that gives you better performance.

ScripterRon · « **Reply #136 on:** January 22, 2015, 04:04:47 am »

Here is the current hash rate on my GT 720 (384 cores, 2 compute units). This is minting ELEMS and worker 4 is the GPU worker.

Code: [Select]

10:59:35 FINE MintWorker.run: Worker 4: 21.65 MHash, 0.0104 MHash/s
10:59:56 FINE MintWorker.run: Worker 1: 5.40 MHash, 0.0026 MHash/s
10:59:56 FINE MintWorker.run: Worker 2: 5.39 MHash, 0.0026 MHash/s
10:59:56 FINE MintWorker.run: Worker 3: 5.32 MHash, 0.0025 MHash/s
10:59:56 FINE MintWorker.run: Worker 0: 5.39 MHash, 0.0026 MHash/s
11:00:04 INFO MintWorker.run: Worker 4 found solution for counter 26
11:00:04 FINE Mint.mint: Solution for counter 26 added to pending queue
11:00:04 FINE MintWorker.run: Worker 1 abandoning counter 26
11:00:04 FINE MintWorker.run: Worker 2 abandoning counter 26
11:00:04 FINE MintWorker.run: Worker 3 abandoning counter 26
11:00:04 FINE MintWorker.run: Worker 2 starting on counter 27
11:00:04 FINE MintWorker.run: Worker 3 starting on counter 27
11:00:04 FINE MintWorker.run: Worker 4 starting on counter 27
11:00:04 FINE MintWorker.run: Worker 0 abandoning counter 26
11:00:04 FINE MintWorker.run: Worker 1 starting on counter 27
11:00:04 FINE MintWorker.run: Worker 0 starting on counter 27
11:00:04 INFO Mint.mint: Solution for counter 26 submitted

Jimmy2011 · « **Reply #137 on:** January 22, 2015, 04:14:52 am »

Quote from: ScripterRon on January 22, 2015, 04:02:10 am

Try the NxtMint-1.1.0.jar that I just placed on Google Drive. It sets the local size (number of work items) to 8 and sets the global size (number of work groups per execution) to 256 x (number of compute units). See if that gives you better performance.

Is the above default setting?

How to set the number of work groups per execution? It seems that the GPU here can support max 1024 work groups.

Code: [Select]

12:26:38 INFO Main.lambda$null$2: GPU device 0: NVIDIA CUDA
  3072MB global memory, 48KB local memory, 16 compute units, Max work group size
 1024

ScripterRon · « **Reply #138 on:** January 22, 2015, 04:33:38 am »

Quote from: Jimmy2011 on January 22, 2015, 04:14:52 am

Is the above default setting? What is "number of work groups per execution" ?

Scrypt now hardcodes the work group size to 8. Specifying the number of cores will not change this value (it will still affect the other algorithms).

The way Aparapi works is as follows:

Data is transferred from host memory to adapter memory
The OpenCL program is executed and Aparapi waits for it to complete
Data is transferred from adapter memory to host memory

OpenCL handles scheduling and dispatching work groups. Each work group contains one or more work items (instances of the kernel program). OpenCL terminates when all of the work groups have been processed. The gpuIntensity value determines the number of work groups to be scheduled. So, the higher the intensity, the longer OpenCL runs before completing.

Multiple work groups can be dispatched concurrently, so Scrypt should still keep the adapter busy as long as the work group size divides evenly into the number of cores on the compute unit. I chose 8 instead of 16 to handle cases like 48 cores.

ScripterRon · « **Reply #139 on:** January 22, 2015, 04:40:08 am »

Quote from: Jimmy2011 on January 22, 2015, 04:14:52 am

How to set the number of work groups per execution? It seems that the GPU here can support max 1024 work groups.
Code: [Select]
12:26:38 INFO Main.lambda$null$2: GPU device 0: NVIDIA CUDA 3072MB global memory, 48KB local memory, 16 compute units, Max work group size 1024

No, it means a work group can have a maximum of 1024 work items. I don't think there is a limit on the number of work groups other than limits imposed by the operating system or device driver (I know the NVIDIA device driver will fail and restart if the OpenCL program runs too long -- had that happen many a time when the code I wrote went into a loop or hung on recursive calls -- at least things are much better than 10 years ago when you would get a blue screen of death).

794259332 · « **Reply #140 on:** January 22, 2015, 08:35:08 am »

error:

exception in thread "main" java.lang.OutOfMemoryError: Java heap space

how to set up?

i have used -Xmx1024m

coretechs · « **Reply #141 on:** January 22, 2015, 01:50:44 pm »

Quote from: 794259332 on January 22, 2015, 08:35:08 am

error:

exception in thread "main" java.lang.OutOfMemoryError: Java heap space

how to set up?

i have used -Xmx1024m

Is this with scrypt hashing? I tried minting some ELEMS on an r290x and had the same error with 2048m. If I increase it to 4096m it is able to start hashing, but it crashes within a few seconds every time. I think the card may have been overclocked at the time and I haven't had a chance to tinker with it any further yet so YMMV.

794259332 · « **Reply #142 on:** January 22, 2015, 02:19:07 pm »

Quote from: coretechs on January 22, 2015, 01:50:44 pm

Quote from: 794259332 on January 22, 2015, 08:35:08 am
error:

exception in thread "main" java.lang.OutOfMemoryError: Java heap space

how to set up?

i have used -Xmx1024m

Is this with scrypt hashing? I tried minting some ELEMS on an r290x and had the same error with 2048m. If I increase it to 4096m it is able to start hashing, but it crashes within a few seconds every time. I think the card may have been overclocked at the time and I haven't had a chance to tinker with it any further yet so YMMV.

yes，ELEMS， i use two r280x cards

ScripterRon · « **Reply #143 on:** January 22, 2015, 03:16:59 pm »

Quote from: 794259332 on January 22, 2015, 08:35:08 am

error:

exception in thread "main" java.lang.OutOfMemoryError: Java heap space

how to set up?

i have used -Xmx1024m

Please post the exception stack trace so I can see where it happened. Also post the startup messages which show information about the graphics card and the calculated local and global sizes.

A 1GB heap should be plenty. Did you specify -Xmx1024m before -jar?

coretechs · « **Reply #144 on:** January 22, 2015, 05:08:29 pm »

I spent some time trying to get the r290x working with scrypt, but the display driver crashes every time with no java error thrown. I tried stock settings as well as lower cpu/memory clocks, intensity 0-XXX, but it always crashes after about 10-20 seconds of hashing. KNV25 is hashing fine even with overclocking (310Mh/s !) so I don't think its an issue with the card. For ref. this is on Windows 7 x64 and the card is reporting 4096m global, 32k local, 44 compute units.

ScripterRon · « **Reply #145 on:** January 22, 2015, 05:22:14 pm »

Quote from: coretechs on January 22, 2015, 05:08:29 pm

I spent some time trying to get the r290x working with scrypt, but the display driver crashes every time with no java error thrown. I tried stock settings as well as lower cpu/memory clocks, intensity 0-XXX, but it always crashes after about 10-20 seconds of hashing. KNV25 is hashing fine even with overclocking (310Mh/s !) so I don't think its an issue with the card. For ref. this is on Windows 7 x64 and the card is reporting 4096m global, 32k local, 44 compute units.

I just put Version 1.2.0 out on github. Give it a try and set the work group size to 8 (gpuDevice=0,8).
The problem might be the 32K of local memory since the default work group size of 256 will require 32K of local memory, leaving nothing for other uses.

ScripterRon · « **Reply #146 on:** January 22, 2015, 05:26:16 pm »

I put Version 1.2.0 out on github https://github.com/ScripterRon/NxtMint/releases.

The gpuDevice configuration parameter has changed from the test version of NxtMint. The second parameter now specifies the work group size instead of the number of cores on the card. You can vary the work group size to reduce local memory contention. I got the best Scrypt results on my card with a work group size of 8. For Keccak25 and Sha256, I would suggest setting the work group size to the number of cores in a compute unit (cores on the card divided by the number of compute units)

cbw · « **Reply #147 on:** January 22, 2015, 06:51:52 pm »

Video cards - 3xr9270x. On version 1.2.0 also tried start with 3gpuDevice (out of memory) and with only 1gpuDevice. With only 1gpuDevice tried set "gpuDevice=0", "gpuDevice=0,8", "gpuDevice=0,4", "gpuDevice=0,2" and "gpuDevice=0,1" - not working, videodriver fails.

coretechs · « **Reply #148 on:** January 22, 2015, 07:23:01 pm »

Same result here, tried with gpuDevice=0,8 / 0,4 / 0,1 and intensity 1,10, video driver crashes. This is only with the AMD, I tested an Nvidia card and it hashes with no problems.

Code: [Select]

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x0000000002a94a2b, pid=2620, tid=404
#
# JRE version: Java(TM) SE Runtime Environment (8.0_31-b13) (build 1.8.0_31-b13)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (25.31-b07 mixed mode windows-amd64 compressed oops)
# Problematic frame:
# J 1238 C1 java.util.ArrayList.toArray([Ljava/lang/Object;)[Ljava/lang/Object; (60 bytes) @ 0x0000000002a94a2b [0x0000000002a945e0+0x44b]
...

crimi · « **Reply #149 on:** January 22, 2015, 07:45:40 pm »

Quote from: coretechs on January 22, 2015, 05:08:29 pm

I spent some time trying to get the r290x working with scrypt, but the display driver crashes every time with no java error thrown. I tried stock settings as well as lower cpu/memory clocks, intensity 0-XXX, but it always crashes after about 10-20 seconds of hashing. KNV25 is hashing fine even with overclocking (310Mh/s !) so I don't think its an issue with the card. For ref. this is on Windows 7 x64 and the card is reporting 4096m global, 32k local, 44 compute units.

No problems here 2 x 290x, intensity 512. But seems a little bit one of the workers is a lot slower.

ScripterRon · « **Reply #150 on:** January 22, 2015, 08:14:45 pm »

Quote from: coretechs on January 22, 2015, 07:23:01 pm

Same result here, tried with gpuDevice=0,8 / 0,4 / 0,1 and intensity 1,10, video driver crashes. This is only with the AMD, I tested an Nvidia card and it hashes with no problems.

Code: [Select]
# # A fatal error has been detected by the Java Runtime Environment: # # EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x0000000002a94a2b, pid=2620, tid=404 # # JRE version: Java(TM) SE Runtime Environment (8.0_31-b13) (build 1.8.0_31-b13) # Java VM: Java HotSpot(TM) 64-Bit Server VM (25.31-b07 mixed mode windows-amd64 compressed oops) # Problematic frame: # J 1238 C1 java.util.ArrayList.toArray([Ljava/lang/Object;)[Ljava/lang/Object; (60 bytes) @ 0x0000000002a94a2b [0x0000000002a945e0+0x44b] ...

That looks like an error in the Aparapi runtime library. I've seen that kind of crash when there is a problem accessing adapter local memory. SCRYPT is the only algorithm that uses local memory, KECCAK25 and SHA256 use just global memory.

Have you checked your device driver and OpenCL levels?

ScripterRon · « **Reply #151 on:** January 22, 2015, 08:16:22 pm »

Quote from: cbw on January 22, 2015, 06:51:52 pm

Video cards - 3xr9270x. On version 1.2.0 also tried start with 3gpuDevice (out of memory) and with only 1gpuDevice. With only 1gpuDevice tried set "gpuDevice=0", "gpuDevice=0,8", "gpuDevice=0,4", "gpuDevice=0,2" and "gpuDevice=0,1" - not working, videodriver fails.

What is the GPU information displayed by NxtMint when it starts?

ScripterRon · « **Reply #152 on:** January 22, 2015, 08:19:19 pm »

Quote from: crimi on January 22, 2015, 07:45:40 pm

Quote from: coretechs on January 22, 2015, 05:08:29 pm
I spent some time trying to get the r290x working with scrypt, but the display driver crashes every time with no java error thrown. I tried stock settings as well as lower cpu/memory clocks, intensity 0-XXX, but it always crashes after about 10-20 seconds of hashing. KNV25 is hashing fine even with overclocking (310Mh/s !) so I don't think its an issue with the card. For ref. this is on Windows 7 x64 and the card is reporting 4096m global, 32k local, 44 compute units.
No problems here 2 x 290x, intensity 512. But seems a little bit one of the workers is a lot slower.

Try reducing the work group size: gpuDevice=0,8 for example. This improved performance for me because it reduces the local memory contention.

cbw · « **Reply #153 on:** January 22, 2015, 09:17:37 pm »

Quote from: ScripterRon on January 22, 2015, 08:16:22 pm

Quote from: cbw on January 22, 2015, 06:51:52 pm
Video cards - 3xr9270x. On version 1.2.0 also tried start with 3gpuDevice (out of memory) and with only 1gpuDevice. With only 1gpuDevice tried set "gpuDevice=0", "gpuDevice=0,8", "gpuDevice=0,4", "gpuDevice=0,2" and "gpuDevice=0,1" - not working, videodriver fails.
What is the GPU information displayed by NxtMint when it starts?

11:16:47 INFO Main.lambda$null$2: GPU device 0: AMD Accelerated Parallel Processing 2048MB global memory, 32KB local memory, 20 compute units, Max work group size 256
11:16:47 INFO Main.lambda$null$2: GPU device 1: AMD Accelerated Parallel Processing 2048MB global memory, 32KB local memory, 20 compute units, Max work group size 256
11:16:47 INFO Main.lambda$null$2: GPU device 2: AMD Accelerated Parallel Processing 2048MB global memory, 32KB local memory, 20 compute units, Max work group size 256

If starts with gpuDevice=0,8 gpuDevice=1,8 gpuDevice=2,8

Quote

11:19:40 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:19:40 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:19:40 INFO MintWorker.run: GPU worker 0 starting on GPU 0
11:19:41 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:19:41 INFO MintWorker.run: GPU worker 1 starting on GPU 1
Exception in thread "main" java.lang.OutOfMemoryError: Java heap space
at org.ScripterRon.NxtMint.GpuScrypt.<init>(GpuScrypt.java:207)
at org.ScripterRon.NxtMint.GpuFunction.factory(GpuFunction.java:59)
at org.ScripterRon.NxtMint.MintWorker.<init>(MintWorker.java:86)
at org.ScripterRon.NxtMint.Mint.mint(Mint.java:93)
at org.ScripterRon.NxtMint.Main.main(Main.java:306)

If starts with gpuDevice=0,8 gpuDevice=1,8

Quote

11:21:32 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:21:32 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:21:32 INFO MintWorker.run: GPU worker 0 starting on GPU 0
11:21:33 INFO MintWorker.run: GPU worker 1 starting on GPU 1
11:21:33 FINE MintWorker.run: Worker 1 starting on counter 1
11:21:33 FINE MintWorker.run: Worker 0 starting on counter 1
#
# A fatal error has been detected by the Java Runtime Environment:
#
# EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x000000006c6e7957, pid=6980, tid=5196
#
# JRE version: Java(TM) SE Runtime Environment (8.0_40-b21) (build 1.8.0_40-ea-b21)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (25.40-b24 mixed mode windows-amd64 compressed oops)
# Problematic frame:
# V [jvm.dll+0x187957]
#
# Failed to write core dump. Minidumps are not enabled by default on client versions of Windows
#
# An error report file with more information is saved as:
# C:\nxtmint\hs_err_pid6980.log

Logfile: pastebin.com/sfWAfa7U

If starts with gpuDevice=0,8 - for 15-20 seconds I see load of my 1 GPU and after 15-20 seconds I got system message, video driver crashed. GPU load 0%. But in NxtMnt cmd window no errors:

Quote

11:28:26 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:28:26 INFO MintWorker.run: GPU worker 0 starting on GPU 0
11:28:26 FINE MintWorker.run: Worker 0 starting on counter 1

Changes in gpuIntensity given nothing, try set 1 and 10.

Start command:
java -Djava.library.path=C:\nxtmint -Xmx2048m -Dnxt.datadir=C:\nxtmint -jar C:\nxtmint\NxtMint-1.2.0.jar
In c:\nxtmint I'm copy all files and folders from aparapi archive and all files and folders from AMD APP SDK x64 - C:\Program Files (x86)\AMD APP SDK\3.0-0-Beta\bin\x86_64 (in previous I'm copy only OpenCL.dll from AMD APP SDK folder). Drivers - AMD Catalyst Omega 14.12.

cbw · « **Reply #154 on:** January 22, 2015, 10:33:42 pm »

Quote from: crimi on January 22, 2015, 07:45:40 pm

No problems here 2 x 290x, intensity 512. But seems a little bit one of the workers is a lot slower.

4 CPU threads, 0 GPU intensity
12:23:25 INFO Nxt.init: API node=localhost, API port=7876
12:23:27 INFO MintWorker.run: CPU worker 0 starting
12:23:27 INFO MintWorker.run: CPU worker 1 starting
12:23:27 INFO MintWorker.run: CPU worker 2 starting
12:23:27 INFO MintWorker.run: CPU worker 3 starting
12:24:38 FINE MintWorker.run: Worker 0: 0,11 MHash, 0,0018 MHash/s
12:24:38 FINE MintWorker.run: Worker 3: 0,11 MHash, 0,0018 MHash/s
12:24:38 FINE MintWorker.run: Worker 2: 0,11 MHash, 0,0018 MHash/s
12:24:38 FINE MintWorker.run: Worker 1: 0,11 MHash, 0,0018 MHash/s

Also tried to set gpuDevice=0,512 and gpuIntensity=512 - not working.
My Athlon X4 750k (weak CPU) doing 0,0072MHash/s and you 2x 290x doing around 0,0088MHash/s at SCRYPT. You say "no problems", really?

crimi · « **Reply #155 on:** January 22, 2015, 11:00:07 pm »

Quote from: cbw on January 22, 2015, 10:33:42 pm

Quote from: crimi on January 22, 2015, 07:45:40 pm
No problems here 2 x 290x, intensity 512. But seems a little bit one of the workers is a lot slower.

4 CPU threads, 0 GPU intensity
12:23:25 INFO Nxt.init: API node=localhost, API port=7876
12:23:27 INFO MintWorker.run: CPU worker 0 starting
12:23:27 INFO MintWorker.run: CPU worker 1 starting
12:23:27 INFO MintWorker.run: CPU worker 2 starting
12:23:27 INFO MintWorker.run: CPU worker 3 starting
12:24:38 FINE MintWorker.run: Worker 0: 0,11 MHash, 0,0018 MHash/s
12:24:38 FINE MintWorker.run: Worker 3: 0,11 MHash, 0,0018 MHash/s
12:24:38 FINE MintWorker.run: Worker 2: 0,11 MHash, 0,0018 MHash/s
12:24:38 FINE MintWorker.run: Worker 1: 0,11 MHash, 0,0018 MHash/s

Also tried to set gpuDevice=0,512 and gpuIntensity=512 - not working.
My Athlon X4 750k (weak CPU) doing 0,0072MHash/s and you 2x 290x doing around 0,0088MHash/s at SCRYPT. You say "no problems", really?

Good point this is really slow. But i dont have any errors like you and did not say it is fast either. Both gpus are loaded 99%.

@ScripterRon

gpuDevice=0,8 etc doesnt even start the minting process, always get out of memory(current java memory usage 2,7 gb). local size: 256, global size 11264 hmm,

ScripterRon · « **Reply #156 on:** January 22, 2015, 11:44:52 pm »

Quote from: cbw on January 22, 2015, 09:17:37 pm

11:16:47 INFO Main.lambda$null$2: GPU device 0: AMD Accelerated Parallel Processing 2048MB global memory, 32KB local memory, 20 compute units, Max work group size 256
11:16:47 INFO Main.lambda$null$2: GPU device 1: AMD Accelerated Parallel Processing 2048MB global memory, 32KB local memory, 20 compute units, Max work group size 256
11:16:47 INFO Main.lambda$null$2: GPU device 2: AMD Accelerated Parallel Processing 2048MB global memory, 32KB local memory, 20 compute units, Max work group size 256

If starts with gpuDevice=0,8 gpuDevice=1,8 gpuDevice=2,8
Quote
11:19:40 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:19:40 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:19:40 INFO MintWorker.run: GPU worker 0 starting on GPU 0
11:19:41 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:19:41 INFO MintWorker.run: GPU worker 1 starting on GPU 1
Exception in thread "main" java.lang.OutOfMemoryError: Java heap space
at org.ScripterRon.NxtMint.GpuScrypt.<init>(GpuScrypt.java:207)
at org.ScripterRon.NxtMint.GpuFunction.factory(GpuFunction.java:59)
at org.ScripterRon.NxtMint.MintWorker.<init>(MintWorker.java:86)
at org.ScripterRon.NxtMint.Mint.mint(Mint.java:93)
at org.ScripterRon.NxtMint.Main.main(Main.java:306)

If starts with gpuDevice=0,8 gpuDevice=1,8
Quote
11:21:32 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:21:32 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:21:32 INFO MintWorker.run: GPU worker 0 starting on GPU 0
11:21:33 INFO MintWorker.run: GPU worker 1 starting on GPU 1
11:21:33 FINE MintWorker.run: Worker 1 starting on counter 1
11:21:33 FINE MintWorker.run: Worker 0 starting on counter 1
#
# A fatal error has been detected by the Java Runtime Environment:
#
# EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x000000006c6e7957, pid=6980, tid=5196
#
# JRE version: Java(TM) SE Runtime Environment (8.0_40-b21) (build 1.8.0_40-ea-b21)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (25.40-b24 mixed mode windows-amd64 compressed oops)
# Problematic frame:
# V [jvm.dll+0x187957]
#
# Failed to write core dump. Minidumps are not enabled by default on client versions of Windows
#
# An error report file with more information is saved as:
# C:\nxtmint\hs_err_pid6980.log
Logfile: pastebin.com/sfWAfa7U

If starts with gpuDevice=0,8 - for 15-20 seconds I see load of my 1 GPU and after 15-20 seconds I got system message, video driver crashed. GPU load 0%. But in NxtMnt cmd window no errors:
Quote
11:28:26 FINE GpuScrypt.<init>: GPU local size 8, global size 5120
11:28:26 INFO MintWorker.run: GPU worker 0 starting on GPU 0
11:28:26 FINE MintWorker.run: Worker 0 starting on counter 1

Changes in gpuIntensity given nothing, try set 1 and 10.

Start command:
java -Djava.library.path=C:\nxtmint -Xmx2048m -Dnxt.datadir=C:\nxtmint -jar C:\nxtmint\NxtMint-1.2.0.jar
In c:\nxtmint I'm copy all files and folders from aparapi archive and all files and folders from AMD APP SDK x64 - C:\Program Files (x86)\AMD APP SDK\3.0-0-Beta\bin\x86_64 (in previous I'm copy only OpenCL.dll from AMD APP SDK folder). Drivers - AMD Catalyst Omega 14.12.

Each work item takes 128KB. With 5120 work items (256*20), that is 660MB. If you start three cards, that means you need a Java heap size of 3*660MB or 1920MB. I can change the global size to be (number of compute units) x (local size) instead of always using 256. This would reduce the storage requirement in your example to (8*20) work items or 20MB. Running 3 cards would then require just 60MB.

The JVM crash might be related to the large data allocation as well. So the above change might fix that problem.

The drawback is that the hash rate will decrease since you won't be keeping all of the cores busy if you set the local size less than the number of cores per compute unit. So it is a trade-off between local memory contention and huge data allocations. I think we need another parameter on gpuDevice to specify the global size so you can vary the both local and global sizes. That way, if you have 32 cores per compute unit and 20 compute units, you could specify gpuDevice=0,8,640. You Java heap size would be just 80MB and all of the cores would be busy.

I'll put up a test version once I have made the change.

ScripterRon · « **Reply #157 on:** January 22, 2015, 11:48:40 pm »

Quote from: crimi on January 22, 2015, 11:00:07 pm

gpuDevice=0,8 etc doesnt even start the minting process, always get out of memory(current java memory usage 2,7 gb). local size: 256, global size 11264 hmm,

Yep - I'm going to have to change that. Multiplying the compute unit number by 256 leads to outrageous group sizes and global storage requirements. I'm going to change gpuDevice to have 3 parameters: device number, local size, global size.

cbw · « **Reply #158 on:** January 23, 2015, 12:01:47 am »

Quote from: ScripterRon on January 22, 2015, 11:44:52 pm

Each work item takes 128KB. With 5120 work items (256*20), that is 660MB. If you start three cards, that means you need a Java heap size of 3*660MB or 1920MB. I can change the global size to be (number of compute units) x (local size) instead of always using 256. This would reduce the storage requirement in your example to (8*20) work items or 20MB. Running 3 cards would then require just 60MB.

If I understand correctly - it have setted up 2048MB of RAM in config and each GPU have 2048MB - 2048MB>1920MB, it couldn't be a problem.

Quote from: ScripterRon on January 22, 2015, 11:44:52 pm

The JVM crash might be related to the large data allocation as well. So the above change might fix that problem.

OK, waiting for fix and test again.

ScripterRon · « **Reply #159 on:** January 23, 2015, 12:58:32 am »

Quote from: cbw on January 23, 2015, 12:01:47 am

If I understand correctly - it have setted up 2048MB of RAM in config and each GPU have 2048MB - 2048MB>1920MB, it couldn't be a problem.

Java needs heap space for itself as well. Try 1920+192=-2122MB or greater. But that is still a lot of storage to allocate and might be causing PCI-E bus problems. The fact that you get a JVM crash for a small allocation size indicates there is still a problem.

I'll post a test version of NxtMint which will allow you to set the global size. I'm also going to look into directly running native OpenCL programs. This would eliminate the Java heap usage since storage will be allocated only on the adapter.

ScripterRon · « **Reply #160 on:** January 23, 2015, 02:21:53 am »

Test version at https://drive.google.com/open?id=0B1312_6UqRHPWUVoZjUxR1ZvazA&authuser=0

gpuDevice now has three parameters: device number, work group size, work group count.

If you specify just the device number (ie, gpuDevice=0), the work group size will be 256 and the work group count will be determined by gpuIntensity.

For my GT 720, I got the best Scrypt results with gpuDevice=0,16,64. Larger work group sizes reduced the hash rate and larger work group counts didn't increase the hash rate.

Code: [Select]

09:19:12 INFO Main.main: Java Nxt Mint Version 1.2.0
09:19:12 INFO Main.main: Application data path: C:\Users\Ronald\Appdata\Roaming\NxtMint
09:19:12 INFO Main.main: Using Nxt node at scripterron.dyndns.biz:7876
09:19:12 INFO Main.main: Minting 5.000000 units of ELEMS for account NXT-XM86-4ZNA-65L5-CDWUE: 1 CPU threads, 30 GPU intensity
09:19:12 INFO Nxt.init: API node=scripterron.dyndns.biz, API port=7876
09:19:13 INFO Main.lambda$null$2: GPU device 0: NVIDIA CUDA
  1024MB global memory, 47KB local memory, 2 compute units, Max work group size 1024
09:19:14 INFO MintWorker.run: CPU worker 0 starting
09:19:14 FINE GpuScrypt.<init>: GPU local size 16, global size 1024
09:19:14 INFO MintWorker.run: GPU worker 1 starting on GPU 0
09:19:14 FINE MintWorker.run: Worker 1 starting on counter 33
09:19:14 FINE MintWorker.run: Worker 0 starting on counter 33
09:20:14 FINE MintWorker.run: Worker 0: 0.20 MHash, 0.0033 MHash/s
09:20:16 FINE MintWorker.run: Worker 1: 0.68 MHash, 0.0111 MHash/s
09:21:14 FINE MintWorker.run: Worker 0: 0.40 MHash, 0.0033 MHash/s
09:21:17 FINE MintWorker.run: Worker 1: 1.35 MHash, 0.0110 MHash/s

794259332 · « **Reply #161 on:** January 23, 2015, 05:58:55 am »

cbw · « **Reply #162 on:** January 23, 2015, 06:37:44 am »

Quote from: ScripterRon on January 23, 2015, 12:58:32 am

Java needs heap space for itself as well. Try 1920+192=-2122MB or greater. But that is still a lot of storage to allocate and might be causing PCI-E bus problems. The fact that you get a JVM crash for a small allocation size indicates there is still a problem.

I'm set 2500MB (can't set 3000MB, free mem around 2,7GB) - no changes. And I never see, that process use over 1GB of mem.

cbw · « **Reply #163 on:** January 23, 2015, 06:46:47 am »

Quote from: ScripterRon on January 23, 2015, 02:21:53 am

Test version at drive.google.com/open?id=0B1312_6UqRHPWUVoZjUxR1ZvazA&authuser=0

Tried gpuDevice=0,16,64, gpuDevice=0,8,32, gpuDevice=0,4,16, gpuDevice=0,2,8... And tried starts with one or two or three video cards. Results: 1) video driver crashes, 2) see fatal error in console.

ScripterRon · « **Reply #164 on:** January 23, 2015, 12:37:46 pm »

Quote from: 794259332 on January 23, 2015, 05:58:55 am

Try the test version on Google Drive. It changes gpuDevice to specify device number, work group size and work group count. See my append right before yours.

ScripterRon · « **Reply #165 on:** January 23, 2015, 12:42:15 pm »

Quote from: cbw on January 23, 2015, 06:46:47 am

Tried gpuDevice=0,16,64, gpuDevice=0,8,32, gpuDevice=0,4,16, gpuDevice=0,2,8... And tried starts with one or two or three video cards. Results: 1) video driver crashes, 2) see fatal error in console.

I don't know what else to try. The failures are in the runtime libraries or the device driver and I'm not able to debug them.

I'm looking into writing native C routines to do CPU hashing. That might result in improve CPU hash performance. The GPU Scrypt support isn't much faster than the current CPU support, so that might be the way to go.

Another possibility is a native OpenCL Scrypt implementation based on something like CGMiner. But I don't think that is something I want to do (at least not now). If somebody wants to tackle it, I'm willing to include it (JOCL will let me invoke the OpenCL program from Java). The catch is the digest generated by the native program must be identical to the digest generated by the NRS server code. Otherwise, the mint transaction will fail when it is submitted to the server.

ScripterRon · « **Reply #166 on:** January 23, 2015, 09:42:58 pm »

I did some testing using native C versus Java for the Keccak25 hash algorithm.

Code: [Select]

Ronald@Hoffman-Desktop /c/temp
$ java JniTest
Running on 64-bit JVM
Performing 16,777,216 hashes
Native hash: Elapsed time 6.75 sec, Rate 2,485.51 KH/sec
Java hash: Elapsed time 9.83 sec, Rate 1,706.74 KH/sec
Hash digests are equal

This is a 45% improvement. So I'm going to modify NxtMint to use native routines for CPU hashing. Since I'm on Windows 8.1, I can provide native routines for Windows (32-bit and 64-bit). If this is desired for other platforms, somebody else will need to build and test the code on those platforms since I don't have Mac or Linux systems.

crimi · « **Reply #167 on:** January 24, 2015, 12:05:47 pm »

Quote from: ScripterRon on January 23, 2015, 09:42:58 pm

This is a 45% improvement.

Waiting for this

ScripterRon · « **Reply #168 on:** January 24, 2015, 01:36:22 pm »

Quote from: crimi on January 24, 2015, 12:05:47 pm

Waiting for this

I've finished the Keccak25 support. I was getting 1.3MH/s using the Java hash routine, so this is a 45% improvement. This is on Windows 8.1 64-bit. The 32-bit library should work but I don't have a 32-bit system to test it on. Note that the Java library path now specifies both the Aparapi and NxtMint runtime libraries (both directories will be packaged in the NxtMint installation archive).

I'll work on Scrypt next. Sha-256 and Sha-3 don't need a native hash method since they are implemented in the Java runtine where native routines are used already.

Code: [Select]

C:\Work>java -Xmx256m -Djava.library.path="\GitHub\NxtMint\target\aparapi;\GitHub\NxtMint\target\jni" -jar "\GitHub\NxtMint\target\NxtMint-1.2.0.jar"
08:25:10 INFO Main.main: Java Nxt Mint Version 1.2.0
08:25:10 INFO Main.main: Application data path: C:\Users\Ronald\Appdata\Roaming\NxtMint
08:25:10 INFO Main.main: Using Nxt node at scripterron.dyndns.biz:7876
08:25:10 INFO Main.main: Minting 5.000000 units of EGOLD for account NXT-XM86-4ZNA-65L5-CDWUE: 4 CPU threads, 0 GPU intensity
08:25:10 INFO Nxt.init: API node=scripterron.dyndns.biz, API port=7876
08:25:11 INFO HashFunction.<init>: JNI library NxtMint_x86_64 loaded - using native hash routines
08:25:11 INFO MintWorker.run: CPU worker 2 starting
08:25:11 INFO MintWorker.run: CPU worker 1 starting
08:25:11 INFO MintWorker.run: CPU worker 3 starting
08:25:11 INFO MintWorker.run: CPU worker 0 starting
08:25:11 FINE MintWorker.run: Worker 2 starting on counter 222
08:25:11 FINE MintWorker.run: Worker 1 starting on counter 222
08:25:11 FINE MintWorker.run: Worker 0 starting on counter 222
08:25:11 FINE MintWorker.run: Worker 3 starting on counter 222
08:26:11 FINE MintWorker.run: Worker 1: 114.82 MHash, 1.9137 MHash/s
08:26:11 FINE MintWorker.run: Worker 3: 113.25 MHash, 1.8874 MHash/s
08:26:11 FINE MintWorker.run: Worker 2: 113.77 MHash, 1.8962 MHash/s
08:26:11 FINE MintWorker.run: Worker 0: 117.44 MHash, 1.9573 MHash/s
08:27:11 FINE MintWorker.run: Worker 0: 231.21 MHash, 1.9268 MHash/s
08:27:11 FINE MintWorker.run: Worker 1: 229.64 MHash, 1.9137 MHash/s
08:27:11 FINE MintWorker.run: Worker 3: 226.49 MHash, 1.8874 MHash/s
08:27:12 FINE MintWorker.run: Worker 2: 229.11 MHash, 1.9093 MHash/s

ScripterRon · « **Reply #169 on:** January 25, 2015, 02:25:47 am »

A test version of NxtMint is available at https://drive.google.com/open?id=0B1312_6UqRHPX2p3RWZDM0VpaGs&authuser=0. It has Windows support for native Keccak25 and Scrypt hashing. The package contains the Aparapi runtime and the NxtMint native library. You can use the sample .bat or .sh to launch NxtMint. If you have your own start script, you need to add
-Djava.library.path="aparapi;jni"
before the -jar option to use the native support. If you don't want to use the native support on Windows, remove the 'jni' directory from the library path.

crimi · « **Reply #170 on:** January 25, 2015, 01:05:44 pm »

Quote from: ScripterRon on January 25, 2015, 02:25:47 am

A test version of NxtMint is available at https://drive.google.com/open?id=0B1312_6UqRHPX2p3RWZDM0VpaGs&authuser=0. It has Windows support for native Keccak25 and Scrypt hashing. The package contains the Aparapi runtime and the NxtMint native library. You can use the sample .bat or .sh to launch NxtMint. If you have your own start script, you need to add
-Djava.library.path="aparapi;jni"
before the -jar option to use the native support. If you don't want to use the native support on Windows, remove the 'jni' directory from the library path.

Nice.

Both cards have a increase(EGOLD) 1 x 40MH/s & 1 x 10MH/s dont know if that settings are optimal.

ScripterRon · « **Reply #171 on:** January 25, 2015, 03:02:27 pm »

Quote from: crimi on January 25, 2015, 01:05:44 pm

Nice.

Both cards have a increase(EGOLD) 1 x 40MH/s & 1 x 10MH/s dont know if that settings are optimal.

The native routines are for CPU hashing (that is, the hash algorithm is implemented in C instead of Java). If somebody has 32-bit Windows, I'd like to hear if it works for them since I'm compiling on Windows 64-bit and using 32-bit link libraries. But there could be execution-time problems since I have the 64-bit JDK.

GPU hashing is implemented in OpenCL and the program runs on the graphics card. There isn't much more that I can do there if I continue to use Aparapi. Aparapi translates Java code to OpenCL code, which means you don't have access to OpenCL mechanisms that aren't embodied in Java (such as register shuffle, vectorization, etc). I'm going to look at writing a native OpenCL program and passing it directly to the adapter for execution (Aparapi would no longer be used).

Hachoir · « **Reply #172 on:** January 25, 2015, 03:51:29 pm »

Tried GPU minting again with version 1.2.0, I get this message now :

Quote

04:46:33 INFO Nxt.init: API node=localhost, API port=7876
04:46:33 SEVERE Main.main: Exception during program initialization
java.lang.IllegalArgumentException: GPU device 0 is not available
at org.ScripterRon.NxtMint.Main.main(Main.java:282)

ScripterRon · « **Reply #173 on:** January 25, 2015, 07:29:04 pm »

Quote from: Hachoir on January 25, 2015, 03:51:29 pm

Tried GPU minting again with version 1.2.0, I get this message now :

Quote
04:46:33 INFO Nxt.init: API node=localhost, API port=7876
04:46:33 SEVERE Main.main: Exception during program initialization
java.lang.IllegalArgumentException: GPU device 0 is not available
at org.ScripterRon.NxtMint.Main.main(Main.java:282)

NxtMint didn't find any GPU devices and gpuIntensity is not 0. Aparapi detects GPU devices based on the information returned by OpenCL. It lists the available GPU devices during initialization and they would appear after the Nxt.init message.

Did something change in your system setup? Or did you have gpuIntensity=0 before?

ScripterRon · « **Reply #174 on:** January 25, 2015, 07:38:43 pm »

I used VirtualBox to set up 64-bit Debian Linux on my Windows 8 system. So I'll look into generating native libaries for Linux as well as Windows.

Code: [Select]

rwh@debian:~/NxtMint$ Starting NxtMint
02:20:34 INFO Main.main: Java Nxt Mint Version 1.2.0
02:20:34 INFO Main.main: Application data path: /home/rwh/.NxtMint
02:20:34 INFO Main.main: Using Nxt node at scripterron.dyndns.biz:7876
02:20:34 INFO Main.main: Minting 5.000000 units of EGOLD for account NXT-XM86-4ZNA-65L5-CDWUE: 2 CPU threads, 0 GPU intensity
02:20:34 INFO Nxt.init: API node=scripterron.dyndns.biz, API port=7876
02:20:38 INFO MintWorker.run: CPU worker 0 starting
02:20:38 INFO MintWorker.run: CPU worker 1 starting
02:20:38 FINE MintWorker.run: Worker 0 starting on counter 229
02:20:38 FINE MintWorker.run: Worker 1 starting on counter 229
02:21:38 FINE MintWorker.run: Worker 0: 94.37 MHash, 1.5729 MHash/s
02:21:39 FINE MintWorker.run: Worker 1: 94.37 MHash, 1.5729 MHash/s
02:22:38 FINE MintWorker.run: Worker 0: 188.74 MHash, 1.5729 MHash/s
02:22:39 FINE MintWorker.run: Worker 1: 188.74 MHash, 1.5599 MHash/s

ScripterRon · « **Reply #175 on:** January 25, 2015, 09:29:19 pm »

NxtMint native support is working on Debian 64-bit. The Keccak25 hash rate improved from 1.5MH/s to 2.2MH/s (46%). This is similar to the improvement under Windows.

Code: [Select]

rwh@debian:~/NxtMint$ Starting NxtMint
04:25:10 INFO Main.main: Java Nxt Mint Version 1.2.0
04:25:10 INFO Main.main: Application data path: /home/rwh/.NxtMint
04:25:10 INFO Main.main: Using Nxt node at scripterron.dyndns.biz:7876
04:25:10 INFO Main.main: Minting 5.000000 units of EGOLD for account NXT-XM86-4ZNA-65L5-CDWUE: 2 CPU threads, 0 GPU intensity
04:25:10 INFO Nxt.init: API node=scripterron.dyndns.biz, API port=7876
04:25:12 INFO HashFunction.<init>: JNI library NxtMint_x86_64 loaded - using native hash routines
04:25:12 INFO MintWorker.run: CPU worker 0 starting
04:25:12 INFO MintWorker.run: CPU worker 1 starting
04:25:12 FINE MintWorker.run: Worker 0 starting on counter 229
04:25:12 FINE MintWorker.run: Worker 1 starting on counter 229
04:26:12 FINE MintWorker.run: Worker 0: 132.12 MHash, 2.2020 MHash/s
04:26:12 FINE MintWorker.run: Worker 1: 132.12 MHash, 2.2020 MHash/s

qwep4ik · « **Reply #176 on:** January 26, 2015, 07:28:33 am »

but you can run around the line for miner GPU

Ziploc · « **Reply #177 on:** January 26, 2015, 12:09:15 pm »

Hi,

any suggestions for settings and intensity for r9 290 ?

Thx

Tempest · « **Reply #178 on:** January 26, 2015, 03:31:13 pm »

Hi,

CPU minting with this Java minter is fine, however, I have an issue when I enable GPU minting.

I am constantly presented with this Java error, can anybody help?

Thanks

ScripterRon · « **Reply #179 on:** January 26, 2015, 04:25:31 pm »

Quote from: Tempest on January 26, 2015, 03:31:13 pm

Hi,

CPU minting with this Java minter is fine, however, I have an issue when I enable GPU minting.

I am constantly presented with this Java error, can anybody help?

Thanks

That's a bug. For now, specify gpuDevice with at least 2 parameters. For example, gpuDevice=0,256

Wolf0 · « **Reply #180 on:** January 27, 2015, 11:46:01 am »

Started mining with this - wow, has it been a while since I've done Java...

ScripterRon · « **Reply #181 on:** January 27, 2015, 12:07:13 pm »

Quote from: Wolf0 on January 27, 2015, 11:46:01 am

Started mining with this - wow, has it been a while since I've done Java...

It has become a mix of Java, C and OpenCL. I'm in the process of replacing Aparapi and writing the hash routines in native OpenCL. This will let me access the GPU vector facilities which I can't do with Aparapi.

As for performance, modern Java is nearly equal to C++ and I much prefer it for the language and the extensive support routines that are bundled with the runtime. For CPU-intensive work, C is better (and assembler is even better).

Wolf0 · « **Reply #182 on:** January 27, 2015, 12:14:29 pm »

Quote from: ScripterRon on January 27, 2015, 12:07:13 pm

Quote from: Wolf0 on January 27, 2015, 11:46:01 am
Started mining with this - wow, has it been a while since I've done Java...
It has become a mix of Java, C and OpenCL. I'm in the process of replacing Aparapi and writing the hash routines in native OpenCL. This will let me access the GPU vector facilities which I can't do with Aparapi.

As for performance, modern Java is nearly equal to C++ and I much prefer it for the language and the extensive support routines that are bundled with the runtime. For CPU-intensive work, C is better (and assembler is even better).

Aparapi is TERRIBLE at getting Java into OpenCL; I'm working on the same thing.

But first, I'm going to be lazy and see what I can do with this Java.

794259332 · « **Reply #183 on:** January 28, 2015, 03:15:27 am »

1.2.0 ver,

error:secret phrase not specified!

but i have specified the phrase in the nxtmint.conf file

ScripterRon · « **Reply #184 on:** January 28, 2015, 03:46:51 am »

Quote from: 794259332 on January 28, 2015, 03:15:27 am

1.2.0 ver,

error:secret phrase not specified!

but i have specified the phrase in the nxtmint.conf file

What directory did you use for NxtMint.conf? For Windows, it goes in
\Users\<your-user-name>\AppData\Roaming\NxtMint\NxtMint.conf

Taxidermista · « **Reply #185 on:** January 28, 2015, 03:00:24 pm »

I can't mint with gpus, only with cpu. This is the error I get:

This is my conf file:

Some help, please.

qwep4ik · « **Reply #186 on:** January 28, 2015, 03:04:35 pm »

Quote from: Taxidermista on January 28, 2015, 03:00:24 pm

I can't mint with gpus, only with cpu. This is the error I get:

gpuDevice=0,8

Taxidermista · « **Reply #187 on:** January 28, 2015, 03:12:00 pm »

Quote from: qwep4ik on January 28, 2015, 03:04:35 pm

Quote from: Taxidermista on January 28, 2015, 03:00:24 pm
I can't mint with gpus, only with cpu. This is the error I get:
gpuDevice=0,8

Sorry, I've just seen previous ScripterRon message about this bug. Thank you for your answer.

qwep4ik · « **Reply #188 on:** January 28, 2015, 04:09:13 pm »

Help me

ScripterRon · « **Reply #189 on:** January 28, 2015, 09:36:02 pm »

Quote from: qwep4ik on January 28, 2015, 04:09:13 pm

Help me

Scrypt requires 128KB per work item. So 13312 work items (your global size) requires 13312*128KB or 1664MB in the Java heap. Aparapi requires all buffers to be allocated in the Java heap even if a particular buffer is an adapter work area which is never accessed by the host.

The requirement will go away once I finish rewriting everything to use native OpenCL. Until then, set your work group count (third parameter for gpuDevice) to a smaller value. For example, if you want to use a Java Heap of 512MB (-Xmx512) with a local size of 8, then set your group count to 512MB/(8*128KB) = 512. So gpuDevice=0,8,512.

Note that if you use multiple GPUs, the total Java heap requirement will be the sum of the heap requirement for each adapter.

qwep4ik · « **Reply #190 on:** January 28, 2015, 10:30:02 pm »

Quote from: ScripterRon

The requirement will go away once I finish rewriting everything to use native OpenCL. Until then, set your work group count (third parameter for gpuDevice) to a smaller value. For example, if you want to use a Java Heap of 512MB (-Xmx512) with a local size of 8, then set your group count to 512MB/(8*128KB) = 512. So gpuDevice=0,8,512.

So gpuDevice=0,8,512.and less to do and no results no, not worked

ScripterRon · « **Reply #191 on:** January 29, 2015, 12:56:54 am »

Quote from: qwep4ik on January 28, 2015, 10:30:02 pm

So gpuDevice=0,8,512.and less to do and no results no, not worked

What error did you get? Did you set the Java heap size? The example I gave requires a 512MB heap.

ScripterRon · « **Reply #192 on:** January 29, 2015, 05:41:32 am »

I finished writing C99 OpenCL routines for the GPU hash algorithms (SHA-256, Keccak25 and Scrypt). NxtMint no longer uses Aparapi and instead uses JOCL for the OpenCL Java bindings. The performance is pretty much unchanged, which is to be expected since the bulk of the work is being done on the graphics card and my initial pass at the OpenCL routines just ported the C code used for CPU hashing (which is almost identical to the C99 code generated by Aparapi).

I then reworked part of the Scrypt algorithm to use the vector facility to do 16 concurrent XOR and ADD operations. The Scrypt performance for my GT 720 went from 0.0114 MH/s to 0.0275 MH/s.

I'll need to look for more places where concurrent vector operations can be done, since that is really the strength of the graphics card (the scalar performance of a single processing unit is generally worse than a CPU).

Code: [Select]

C:\Work>java -Xmx256m -Djava.library.path="\GitHub\NxtMint\target\jni;\GitHub\NxtMint\package\jni" -jar "\GitHub\NxtMint
\target\NxtMint-1.3.0.jar"
12:25:23 INFO Main.main: Java Nxt Mint Version 1.3.0
12:25:23 INFO Main.main: Application data path: C:\Users\Ronald\Appdata\Roaming\NxtMint
12:25:23 INFO Main.main: Using Nxt node at scripterron.dyndns.biz:7876
12:25:23 INFO Main.main: Minting 5.000000 units of ELEMS for account NXT-XM86-4ZNA-65L5-CDWUE: 0 CPU threads, 10 GPU int
ensity
12:25:23 INFO Nxt.init: API node=scripterron.dyndns.biz, API port=7876
12:25:24 INFO Main.buildGpuList: GPU device 0: GeForce GT 720
  1024MB global memory, 47KB local memory, 2 compute units, Max work group size 1024
12:25:25 INFO HashFunction.<init>: JNI library NxtMint_x86_64 loaded - using native CPU hash routines
12:25:25 INFO GpuScrypt.<init>: GPU 0: Preferred work group size multiple is 32
12:25:25 FINE GpuScrypt.<init>: GPU 0: Local size 16, Global size 2048, Passes 5
12:25:25 INFO MintWorker.run: GPU worker 0 starting on GPU 0
12:25:25 FINE MintWorker.run: Worker 0 starting on counter 54
12:26:26 FINE MintWorker.run: Worker 0: 1.34 MHash, 0.0224 MHash/s
12:27:26 FINE MintWorker.run: Worker 0: 2.68 MHash, 0.0224 MHash/s

Wolf0 · « **Reply #193 on:** January 29, 2015, 06:41:48 am »

Quote from: ScripterRon on January 29, 2015, 05:41:32 am

I finished writing C99 OpenCL routines for the GPU hash algorithms (SHA-256, Keccak25 and Scrypt). NxtMint no longer uses Aparapi and instead uses JOCL for the OpenCL Java bindings. The performance is pretty much unchanged, which is to be expected since the bulk of the work is being done on the graphics card and my initial pass at the OpenCL routines just ported the C code used for CPU hashing (which is almost identical to the C99 code generated by Aparapi).

I wouldn't bother with Keccak - the only one that matters to me. Also, protip: GCN (7xxx and up) do NOT have hardware vectors.