Author

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 666. (Read 3426922 times)

hero member
Activity: 756
Merit: 502
ccMiner v0.2 now supports pools with fuguecoin (like dwarfpool) and has a nice boost for Fermi and Maxwell owners.

https://github.com/cbuchner1/ccminer/releases/tag/v0.2

Christian
hero member
Activity: 756
Merit: 502
I tried getting source code from git and compiling cudaminer myself to see how blake works there Smiley
I got it running after some time but I was surprised to get only around 42 MH/s on my GF GTX670M.
cgminer is much faster (250 MH/s) but that is still very low hashrate compared to the AMD cards which are equally powerful.
I just wanted to ask - is blake in some very early stage and will the performance be improved so we get more than using cgminer?

in case you didn't get the memo: nVidia sucks at compute bound hashing algorithms. Wink
Factor 2 slower than AMD or so...

Now about the 42 MH/s vs 250 MH/s... I don't know how you compiled it. Custom builds
break the warranty ;-)

Christian

legendary
Activity: 1400
Merit: 1050
Doesn't cost much, to do the launch and grab a few blocks. Just in case.
What would be the scrypt-n parameters though?
I have never been on a N-factor coin launch Tongue
you haven't been on any vertcoin, exe, gpuc ? 
Correct Cheesy
cudaminer.exe --algo=scrypt:2048  -q -m1 -d gtx780 -b32768 -C2 -lt15x32 -i 0 -H 2
(try autotune first, I think for some cards the T kernel is better)
full member
Activity: 182
Merit: 100
Doesn't cost much, to do the launch and grab a few blocks. Just in case.
What would be the scrypt-n parameters though?
I have never been on a N-factor coin launch Tongue
you haven't been on any vertcoin, exe, gpuc ? 
Correct Cheesy
legendary
Activity: 1400
Merit: 1050
Doesn't cost much, to do the launch and grab a few blocks. Just in case.
What would be the scrypt-n parameters though?
I have never been on a N-factor coin launch Tongue
you haven't been on any vertcoin, exe, gpuc ?  
cudaminer.exe --algo=scrypt:2048  -q -m1 -d gtx780 -b32768 -C2 -lt15x32 -i 0 -H 2

you might have to change the config though...
full member
Activity: 182
Merit: 100
Doesn't cost much, to do the launch and grab a few blocks. Just in case.
What would be the scrypt-n parameters though?
I have never been on a N-factor coin launch Tongue
legendary
Activity: 1400
Merit: 1050
Doesn't cost much, to do the launch and grab a few blocks. Just in case.
hero member
Activity: 812
Merit: 1000
+1 to roto...instamine won't last long.
full member
Activity: 182
Merit: 100
any thought on spartancoin ? Grin
Just solo mine 2 blocks

I'm probably going to try out Rotocoin. Launch is in about 3-4 hours.

https://bitcointalksearch.org/topic/ann-rt2-rotocoinnow-on-cryptsybet-with-your-rotos-android-trader-app-516138
2% premine, no exchange, no required data for N-factor coins, good luck to you!
full member
Activity: 144
Merit: 100
Mek
jr. member
Activity: 72
Merit: 6
mtc.mekweb.eu - mega transistor clock
I tried getting source code from git and compiling cudaminer myself to see how blake works there Smiley
I got it running after some time but I was surprised to get only around 42 MH/s on my GF GTX670M.
cgminer is much faster (250 MH/s) but that is still very low hashrate compared to the AMD cards which are equally powerful.
I just wanted to ask - is blake in some very early stage and will the performance be improved so we get more than using cgminer?
legendary
Activity: 1400
Merit: 1050
 any thought on spartancoin ? Grin
Just solo mine 2 blocks
full member
Activity: 145
Merit: 101
So the hashrate velocity reach the 20,14 khash/s (that is very near to the banchmark value that was 24 khash/s)

Some ideas about how reduce more?
Try reduce kernel load more.
hero member
Activity: 812
Merit: 1000
You won't get anything decent out of that laptop gpu...maybe try yacoin?
How much ram does it have?
member
Activity: 112
Merit: 10

Code:
C:\Users\Andrea\Documents\crypt\cudaminer-2014-02-28\x64>cudaminer.exe --benchma
rk
           *** CudaMiner for nVidia GPUs by Christian Buchner ***
                     This is version 2014-02-28 (beta)
        based on pooler-cpuminer 2.3.2 (c) 2010 Jeff Garzik, 2012 pooler
            Cuda additions Copyright 2013,2014 Christian Buchner
          LTC donation address: LKS1WDKGED647msBQfLBHV3Ls8sveGncnm
          BTC donation address: 16hJF5mceSojnTD3ZTUDqdRhDyPJzoRakM
          YAC donation address: Y87sptDEcpLkLeAuex6qZioDbvy1qXZEj4
[2014-03-21 18:57:38] 1 miner threads started, using 'scrypt' algorithm.
[2014-03-21 18:57:38] GPU #0: GeForce GT 720M with compute capability 2.1
[2014-03-21 18:57:38] GPU #0: interactive: 1, tex-cache: 0 , single-alloc: 0
[2014-03-21 18:57:38] GPU #0: 32 hashes / 4.0 MB per warp.
[2014-03-21 18:57:38] GPU #0: Performing auto-tuning (Patience...)
[2014-03-21 18:57:38] GPU #0: maximum total warps (BxW): 436
[2014-03-21 18:57:45] GPU #0: 26213.73 hash/s with configuration F4x16
[2014-03-21 18:57:45] GPU #0: using launch configuration F4x16
[2014-03-21 18:57:45] GPU #0: GeForce GT 720M, 23.83 khash/s
[2014-03-21 18:57:45] Total: 23.83 khash/s
[2014-03-21 18:57:51] GPU #0: GeForce GT 720M, 23.58 khash/s
[2014-03-21 18:57:51] Total: 23.58 khash/s

So I think that my hasching speed is arount 24 khash/s (is it so low as value to do something concrete?) and, at first stage I have to change it in 12 khash/s

But what exactly have I to do to perform this operation? You say to me that I haveto reduce the -l...ok...but if execute:

cudaminer.exe --help for the -l paramether it say to me:

Code:
 -l, --launch-config   gives the launch configuration for each kernel
                       in a comma separated list, one per device.

So it seems to me that it is not related to the hashrate speed...

What am I missing?

Tnx
See this [2014-03-21 18:57:45] GPU #0: using launch configuration F4x16. Try run with -l F2x8

Later I will try but exactly what means F4x16 and F2x8
What these value means?

I have read the documentation about the Cuda Kernel but seems don't work because after that I execute my CudaMiner application using this statment:

C:\Users\Andrea\Documents\crypt\cudaminer-2014-02-28\x64>cudaminer.exe -o stratum+tcp://stratum.hasher.ca:3373 -O D4rkAng3l.Andrea1:MyPassword -l F2x8

I obtain the following output:

Code:
           *** CudaMiner for nVidia GPUs by Christian Buchner ***
                     This is version 2014-02-28 (beta)
        based on pooler-cpuminer 2.3.2 (c) 2010 Jeff Garzik, 2012 pooler
            Cuda additions Copyright 2013,2014 Christian Buchner
          LTC donation address: LKS1WDKGED647msBQfLBHV3Ls8sveGncnm
          BTC donation address: 16hJF5mceSojnTD3ZTUDqdRhDyPJzoRakM
          YAC donation address: Y87sptDEcpLkLeAuex6qZioDbvy1qXZEj4
[2014-03-21 20:31:12] 1 miner threads started, using 'scrypt' algorithm.
[2014-03-21 20:31:12] Starting Stratum on stratum+tcp://stratum.hasher.ca:3373
[2014-03-21 20:31:13] Stratum detected new block
[2014-03-21 20:31:14] GPU #0: GeForce GT 720M with compute capability 2.1
[2014-03-21 20:31:14] GPU #0: interactive: 1, tex-cache: 0 , single-alloc: 0
[2014-03-21 20:31:14] GPU #0: 32 hashes / 4.0 MB per warp.
[2014-03-21 20:31:14] GPU #0: using launch configuration F2x8
[2014-03-21 20:31:14] GPU #0: GeForce GT 720M, 16.38 khash/s
[2014-03-21 20:31:18] Stratum detected new block
[2014-03-21 20:31:18] GPU #0: GeForce GT 720M, 16.15 khash/s
[2014-03-21 20:32:11] GPU #0: GeForce GT 720M, 18.15 khash/s
[2014-03-21 20:32:13] GPU #0: GeForce GT 720M, 19.68 khash/s
[2014-03-21 20:32:23] accepted: 1/1 (100.00%), 19.68 khash/s (yay!!!)
[2014-03-21 20:32:26] GPU #0: GeForce GT 720M, 20.12 khash/s
[2014-03-21 20:32:27] accepted: 2/2 (100.00%), 20.12 khash/s (yay!!!)
[2014-03-21 20:32:58] GPU #0: GeForce GT 720M, 20.14 khash/s
[2014-03-21 20:32:58] accepted: 3/3 (100.00%), 20.14 khash/s (yay!!!)
[2014-03-21 20:33:58] GPU #0: GeForce GT 720M, 20.14 khash/s
[2014-03-21 20:34:58] GPU #0: GeForce GT 720M, 20.14 khash/s


So the hashrate velocity reach the 20,14 khash/s (that is very near to the banchmark value that was 24 khash/s)

Some ideas about how reduce more?
full member
Activity: 168
Merit: 100
Hi Cristian, how about new release?

tonight then. After the update performance will be approximately like this.

6x 750 Ti= 510 MHash/s = 3 x 780Ti   (all not overclocked, on Linux)

previously it would take you 12-15 750Ti's to get this hash Wink


Is the new release going to bump the performance of 780's and 780 Ti's?

I'm currently getting about 180 mh/s on my 780 and 250mh/s on my 780 Ti. Even then I'm not using 100% of my graphics cards. They're hovering around 90% to 95% usage with or without Chrome or flash open.
full member
Activity: 145
Merit: 101
Later I will try but exactly what means F4x16 and F2x8
What these value means?
Read >>> About CUDA Kernels <<< in README file
member
Activity: 68
Merit: 10
What is the average speed difference for scrypt mining between the GTX 750 and GTX 750 ti? Guessing 260 to 300 KH/s @ $110 to $155? Does the extra 1GB memory make that much of difference?
may-be on yaccoin

Right, so-what's the actual KH/s for a GTX 750? Couldn't find on spreadsheet.
I got about $220 to spend. One GTX 750 ti or two GTX 750's. ;-)
Opinions?
legendary
Activity: 1400
Merit: 1050
What is the average speed difference for scrypt mining between the GTX 750 and GTX 750 ti? Guessing 260 to 300 KH/s @ $110 to $155? Does the extra 1GB memory make that much of difference?
may-be on yaccoin
member
Activity: 68
Merit: 10
What is the average speed difference for scrypt mining between the GTX 750 and GTX 750 ti? Guessing 260 to 300 KH/s @ $110 to $155? Does the extra 1GB memory make that much of difference?
Jump to: