Switch to ArdorForum

Then for no reason at all I lost 100 MHs/s back down to about 120 MHs/s

Quote from: jshatlee on February 01, 2015, 11:19:14 pm

Then for no reason at all I lost 100 MHs/s back down to about 120 MHs/s

Told you - tempermental.

Lol, after 3 restarts... nothing... Same speeds |Facepalm|

HI,

I use a Radeon HD6990 but only hit about 90MH/s (43MH/s per GPU Unit) with 1.3
With 1.4 same settings only 2/3 MH/S
......

Same here
5970 2x30+ MH/sm 1.3
6970 46 MH/s with 1.3
It looks like not optimized for old cards with VLIW architecture.
7770 36 MH/s

HI,

I use a Radeon HD6990 but only hit about 90MH/s (43MH/s per GPU Unit) with 1.3
With 1.4 same settings only 2/3 MH/S
Running Ubuntu 12.04 .

Settings :

# Set the Nxt node host as a DNS name or an IP address
connect=localhost

# Set the API port (the default is 7876)
apiPort=7876

# Set the secret phrase for the minting account (required field)
secretPhrase=****************

# Set the 3-5 character currency code (required field)
currency=EGOLD

# Set the number of units to generate for each hash round (the default is 1).
units=40

# Set the CPU thread count (the default is 1).  CPU minting will not be done
# if cpuThreads=0.
cpuThreads=1

# Set the GPU intensity (the default is 0).  GPU minting will not be done
# if gpuIntensity=0.
gpuIntensity=10

# Set 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.
gpuDevice=0,256,44000
gpuDevice=1,256,44000
# Enable/Disable the GUI (the default is true)
enableGUI=false

I have a few questions.
I'm using a windows computer and I don't see those settings ?

I see these :

nxt.mint.serverAddress=localhost
nxt.mint.currencyCode=URANM
nxt.mint.secretPhrase=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
nxt.mint.unitsPerMint=50             
nxt.mint.threadPoolSize=1
nxt.mint.initialNonce=0
nxt.mint.isSubmitted=true


Can I add a line like this : nxt.mint.gpuIntensity=10   ?

Also how or where do I look to see what speed I'm hashing at ?

Thanks
Mark

Note to developper??


Same here
5970 2x30+ MH/s
6970 40+ MH/s with 1.2
It looks like not optimized for old cards with VLIW architecture.

The OpenCL code isn't optimized for any particular GPU.  If anybody has optimized OpenCL programs for either Scrypt or Keccak25, I'll take a look at incorporating the changes into NxtMint.  I have a single card (NVIDIA 720), so I'm not able to test different configurations myself.  Note that the generated hash must match the hash generated by the NRS reference code in order for the NRS server to accept the currency mint transaction.

The third parameter in gpuDevice defaults to 32 in 1.4.0 instead of 0 as in 1.3.0.  If you want the group count to be based on gpuIntensity, then explicitly specify 0 for the third parameter (for example, gpuDevice=0,256,0).  I'll change it to default to 0 once more in the next release.

I'm also going to add a check for a non-zero account balance when NxtMint starts.  This should catch situations where the passphrase is incorrect since it is unlikely that the resulting account will be funded.

I have a Problem.
I minting, and all its okey but i become an error, called...

Error 4 returned for broadcastTransaction request: Double spenden or insufficient Balance.

That's an error returned by the NRS server for the currency mint transaction.  Most likely cause is the account has 0 Nxt (the transaction fee is 1 Nxt for each currency transaction).

When NxtMint starts, it displays your account number.  Is it the correct one?  Mistyping the secret phrase will result in a different account number.

Yes, its the right account number and i have nxt Balance in my wallet, but only 0.6 nxt, is that okey?

It isn't enough, you need 1 NXT per mint.

but another question , what is the best to minting , i mintet URANM , but now i have no NXT coins in my wallet , all it goes for the fee and nothing is incoming ?

Minting does not create Nxt, it creates coins for whatever currency you specify.  In order to exchange the coins for Nxt, you need to find a buy offer on the Monetary System exchange for that currency.

Each minting transaction will cost you 1 Nxt.  And each buy request that is satisfied will cost you 1 Nxt.  So your goal is to set the number of units to mint high enough to make a profit.  Taking EGOLD as an example, it is currently selling for around 0.05 Nxt.  So you need to sell at least 40 EGOLD to make a profit of 1 Nxt (40*0.05=2 Nxt minus 1 Nxt transaction fee).  However, it also costs 1 Nxt for each minting transaction, so you need to mint 60 EGOLD at a time (60*0.05=3 Nxt minus 2 Nxt transaction fees) to make a 1 Nxt profit.  The more units you mint at once, the greater your profit.  However, if you set the units too high, you might not mint anything (the difficulty increases as you increase the number of units).

I have been running into this error a few times today and I am hoping you might have some feedback.  You made comment about a similar issue, however this is with EGOLD and I can't figure out what is the cause (too much intensity, more mv+ on the gpu, etc).  The below issue happened on v1.3 and with VNC open at the time

12:53:05 SEVERE Nxt.issueRequest: Error 4 returned for broadcastTransaction requ
est: nxt.NxtException$NotCurrentlyValidException: Hash doesn't meet target {"cur
rency":"5775213290661997199","units":5000000,"counter":181,"version.CurrencyMint
ing":1,"nonce":-9191694230504894365}
12:53:05 SEVERE Mint.mint: Server rejected 'currencyMint' transaction - discardi
ng

Also, I'm noticing with 1.4, that instead of 270+mh on r9290 card, I get like 15 to 20mh

I haven't changed the algorithms in 1.4.  Keccak25 isn't memory-intensive, it is a 25-pass loop consisting of a long series of XOR, AND and NOT operations.  What did you specify for gpuDevice?  Note that the group count defaults to 32 in 1.4, so that could be the cause of the drop if you didn't specify it.  Specify 0 to get the 1.3 behavior (ex, gpuDevice=0,256,0).  It will default to 0 once more in 1.5 (I was trying to work around some Scrypt problems and didn't think about the effect for Keccak25).

To see if it is a GPU load problem, try minting with a single GPU work item (gpuDevice=0,1,1).  Use a small units value (such as units=0.0001) so you get a fairly quick result.  If that works, then I think the problem is with your card since there is a single Keccak25 kernel which is executed multiple times based on the local and global sizes.

The algorithm is the same for CPU and GPU.  So you can also try minting using just the CPU and see if the transaction is accepted by the server. 

Depending on how many registers are available on your card, some of the register data could be spilling over to global memory.  This could introduce memory latency problems which could cause corrupted data in some edge cases.  I'd like to reduce the register usage, but Keccak25 requires a 5x5 state matrix, which is accessed and modified for each pass.

For some reason 1.4 only yields about 2 MH/s.
I tried 0,256,0 with 1024 intensity on 1.4 vs 0,256,44000 intensity 10 on 1.3
Note the different GPU devicename.

11:29:23 INFO Main.buildGpuList: GPU device 0: Cayman, Driver 1573.4 (VM)
  1904MB global memory, 32KB local memory, 24 compute units, Max work group size 256
11:29:23 INFO Main.buildGpuList: GPU device 1: Cayman, Driver 1573.4 (VM)
  1953MB global memory, 32KB local memory, 24 compute units, Max work group size 256
11:29:23 INFO HashFunction.<init>: JNI library NxtMint_x86 loaded - using native CPU hash routines
11:29:23 INFO MintWorker.run: CPU worker 0 starting
11:29:23 FINE GpuKnv25.<init>: GPU 0: Local size 256, Global size 1048576, Passes 1
11:29:23 INFO MintWorker.run: GPU worker 1 starting on GPU 0
11:29:23 FINE GpuKnv25.<init>: GPU 1: Local size 256, Global size 1048576, Passes 1
11:29:23 INFO MintWorker.run: GPU worker 2 starting on GPU 1
11:29:23 FINE MintWorker.run: Worker 0 starting on counter 20
11:29:23 FINE MintWorker.run: Worker 1 starting on counter 20
11:29:23 FINE MintWorker.run: Worker 2 starting on counter 20
11:30:23 FINE MintWorker.run: Worker 2: 154.14 MHash, 2.5690 MHash/s
11:30:23 FINE MintWorker.run: Worker 1: 154.14 MHash, 2.5690 MHash/s
11:30:26 FINE MintWorker.run: Worker 0: 23.07 MHash, 0.3721 MHash/s
11:31:23 FINE MintWorker.run: Worker 2: 308.28 MHash, 2.5690 MHash/s
11:31:24 FINE MintWorker.run: Worker 1: 308.28 MHash, 2.5690 MHash/s
11:31:28 FINE MintWorker.run: Worker 0: 46.14 MHash, 0.3691 MHash/s
11:32:23 FINE MintWorker.run: Worker 2: 462.42 MHash, 2.5690 MHash/s
11:32:24 FINE MintWorker.run: Worker 1: 462.42 MHash, 2.5690 MHash/s
11:32:31 FINE MintWorker.run: Worker 0: 69.21 MHash, 0.3701 MHash/s

1.3 Gives 43MH/S


11:33:09 INFO Main.lambda$null$2: GPU device 0: AMD Accelerated Parallel Processing
  1904MB global memory, 32KB local memory, 24 compute units, Max work group size 256
11:33:09 INFO Main.lambda$null$2: GPU device 1: AMD Accelerated Parallel Processing
  1953MB global memory, 32KB local memory, 24 compute units, Max work group size 256
11:33:09 INFO HashFunction.<init>: JNI library NxtMint_x86 loaded - using native hash routines
11:33:09 INFO MintWorker.run: CPU worker 0 starting
11:33:09 FINE GpuKnv25.<init>: GPU local size 256, global size 11264000
11:33:09 FINE GpuKnv25.<init>: GPU local size 256, global size 11264000
11:33:09 INFO MintWorker.run: GPU worker 1 starting on GPU 0
11:33:09 INFO MintWorker.run: GPU worker 2 starting on GPU 1
11:33:09 FINE MintWorker.run: Worker 1 starting on counter 20
11:33:09 FINE MintWorker.run: Worker 2 starting on counter 20
11:33:09 FINE MintWorker.run: Worker 0 starting on counter 20
11:34:09 FINE MintWorker.run: Worker 1: 2,601.98 MHash, 43.3664 MHash/s
11:34:09 FINE MintWorker.run: Worker 2: 2,601.98 MHash, 43.3664 MHash/s
11:34:11 FINE MintWorker.run: Worker 0: 23.07 MHash, 0.3721 MHash/s
11:35:09 FINE MintWorker.run: Worker 1: 5,226.50 MHash, 43.5541 MHash/s
11:35:09 FINE MintWorker.run: Worker 2: 5,226.50 MHash, 43.5541 MHash/s
11:35:14 FINE MintWorker.run: Worker 0: 46.14 MHash, 0.3691 MHash/s
11:36:09 FINE MintWorker.run: Worker 1: 7,851.01 MHash, 43.6167 MHash/s
11:36:09 FINE MintWorker.run: Worker 2: 7,851.01 MHash, 43.6167 MHash/s