Author

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 754. (Read 3426976 times)

legendary
Activity: 1400
Merit: 1050
I did a small check on the high N optimization (however I didn't run on yac since quite some time).
I also stopped the autotune at the 17th line (it really take a long long time...).
The hashrate on this test has increased until -L12 and yields to 4.7khah/s on windows (didn't tried linux, I assume the real gain should be on linux). Running on a pool (one share every 5 minutes... now I remember why I gave up on yac) this gave me around 4.5khash/s.
But as I said I didn't perform the autotune on the full table, so there could be better value...

On MRC, it gives me around 5~10 more khash/s (at L2)
newbie
Activity: 21
Merit: 0
Quote from: WattBurner
Here's the result when I use autotune...

[2014-02-27 15:57:33] GPU #3: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaDeviceSynchronize()' (D:/Christian/Documents/Visual Studio 2010
/Projects/CudaMiner/salsa_kernel.cu line 644)

"launch timed out" is unusual for autotuning an scrypt coin.

use -l T14x24 instead of -l T14x32. This should fit your cards.

the problem arose when you updated your cudaminer release from a December 2013 to a February 2014 release. Kernels were swapped out (nVidia's submission proved to be a bit faster than David Andersen's kernels).


T14x24 launches and hashes around 7-8 kh/s on each card for a few seconds...then I get the same launch error on GPU #3

Should I try to go back to the December version? Am I missing a prerequisite software to make this work with the latest version?
hero member
Activity: 756
Merit: 502
Compiling windows binaries is SO complicated Sad

Why? I hit "build" and it's done ;-)

Christian
hero member
Activity: 812
Merit: 1000
Compiling windows binaries is SO complicated Sad
I wanna try the new nvidia n factor 14 kernel.
Page 130 still works Wink

Gonna try it tonight...  Cool
hero member
Activity: 756
Merit: 502
Quote from: WattBurner
Here's the result when I use autotune...

[2014-02-27 15:57:33] GPU #3: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaDeviceSynchronize()' (D:/Christian/Documents/Visual Studio 2010
/Projects/CudaMiner/salsa_kernel.cu line 644)

"launch timed out" is unusual for autotuning an scrypt coin.

use -l T14x24 instead of -l T14x32. This should fit your cards.

the problem arose when you updated your cudaminer release from a December 2013 to a February 2014 release. Kernels were swapped out (nVidia's submission proved to be a bit faster than David Andersen's kernels).



legendary
Activity: 1400
Merit: 1050

[2014-02-27 15:21:35] GPU #2: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #2: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #2: GeForce GTX TITAN, 3581 khash/s
[2014-02-27 15:21:35] GPU #0: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #0: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #0: GeForce GTX TITAN, 477.41 khash/s
[2014-02-27 15:21:35] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=1899, s=0)!
[2014-02-27 15:21:35] GPU #1: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #1: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #3: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
A little bit of reading the output can not hurt, it's there for a reason, basically your specified numbers are too high, try autotune Smiley


Here's the result when I use autotune...

[2014-02-27 15:57:33] GPU #3: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaDeviceSynchronize()' (D:/Christian/Documents/Visual Studio 2010
/Projects/CudaMiner/salsa_kernel.cu line 644)
On T kernel you cannot go higher than x24, so x32 won't validate and now you need to reboot to reset the card
newbie
Activity: 21
Merit: 0

[2014-02-27 15:21:35] GPU #2: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #2: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #2: GeForce GTX TITAN, 3581 khash/s
[2014-02-27 15:21:35] GPU #0: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #0: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #0: GeForce GTX TITAN, 477.41 khash/s
[2014-02-27 15:21:35] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=1899, s=0)!
[2014-02-27 15:21:35] GPU #1: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #1: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #3: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
A little bit of reading the output can not hurt, it's there for a reason, basically your specified numbers are too high, try autotune Smiley


Here's the result when I use autotune...

[2014-02-27 15:57:33] GPU #3: cudaError 6 (the launch timed out and was terminat
ed) calling 'cudaDeviceSynchronize()' (D:/Christian/Documents/Visual Studio 2010
/Projects/CudaMiner/salsa_kernel.cu line 644)
full member
Activity: 182
Merit: 100
Compiling windows binaries is SO complicated Sad
I wanna try the new nvidia n factor 14 kernel.
Page 130 still works Wink
hero member
Activity: 812
Merit: 1000
Compiling windows binaries is SO complicated Sad
I wanna try the new nvidia n factor 14 kernel.
hero member
Activity: 676
Merit: 500
Well most of us ,when chrome opened and running flash game for example, have a boost of 20-30khs.
full member
Activity: 182
Merit: 100

[2014-02-27 15:21:35] GPU #2: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #2: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #2: GeForce GTX TITAN, 3581 khash/s
[2014-02-27 15:21:35] GPU #0: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #0: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #0: GeForce GTX TITAN, 477.41 khash/s
[2014-02-27 15:21:35] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=1899, s=0)!
[2014-02-27 15:21:35] GPU #1: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #1: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #3: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
A little bit of reading the output can not hurt, it's there for a reason, basically your specified numbers are too high, try autotune Smiley
sr. member
Activity: 490
Merit: 254
Run it and you will see  Grin Grin Grin Grin Grin a booost..

I am running it and if anything I think I am down about 5 kHash each card... Need to do anything special, or just have it open?
newbie
Activity: 21
Merit: 0
Hi Guys! I need your help to fix my rig.

Just did a format-clean-re-install of windows 8.1 and cudaminer won't hash now properly. Any insight how to fix this would be much appreciated!

System Specs:
Windows 8.1
Nvidia driver 334.89
1 Quadro 4000 slot 1 on Supermicro chassis
4 GTX Titans GPUs in Cubix Xpander Desktop Elite (I'm trying to hash with just the Titan's)
Supermicro 7047grTFT
2 x E5-2660 XEON CPUs
128 GB system ram


C:\cm>c:\cm\x64\cudaminer -a scrypt -d 0,1,2,3 -l T14x32,T14x32,T14x32,T14x32 -i
 0,0,0,0 -o stratum+tcp://pool1.us.multipool.us:7777 -O dm.titan:x
           *** CudaMiner for nVidia GPUs by Christian Buchner ***
                     This is version 2014-02-18 (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
[2014-02-27 15:21:30] 4 miner threads started, using 'scrypt' algorithm.
[2014-02-27 15:21:30] Starting Stratum on stratum+tcp://pool1.us.multipool.us:77
77
[2014-02-27 15:21:31] Stratum detected new block
[2014-02-27 15:21:33] GPU #0: GeForce GTX TITAN with compute capability 3.5
[2014-02-27 15:21:33] GPU #0: interactive: 0, tex-cache: 0 , single-alloc: 0
[2014-02-27 15:21:33] GPU #0: 32 hashes / 4.0 MB per warp.
[2014-02-27 15:21:33] GPU #2: GeForce GTX TITAN with compute capability 3.5
[2014-02-27 15:21:33] GPU #2: interactive: 0, tex-cache: 0 , single-alloc: 0
[2014-02-27 15:21:33] GPU #2: 32 hashes / 4.0 MB per warp.
[2014-02-27 15:21:33] GPU #3: GeForce GTX TITAN with compute capability 3.5
[2014-02-27 15:21:33] GPU #3: interactive: 0, tex-cache: 0 , single-alloc: 0
[2014-02-27 15:21:33] GPU #3: 32 hashes / 4.0 MB per warp.
[2014-02-27 15:21:33] GPU #1: GeForce GTX TITAN with compute capability 3.5
[2014-02-27 15:21:33] GPU #1: interactive: 0, tex-cache: 0 , single-alloc: 0
[2014-02-27 15:21:33] GPU #1: 32 hashes / 4.0 MB per warp.
[2014-02-27 15:21:35] GPU #2: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #2: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #2: GeForce GTX TITAN, 3581 khash/s
[2014-02-27 15:21:35] GPU #0: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #0: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #0: GeForce GTX TITAN, 477.41 khash/s
[2014-02-27 15:21:35] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=1899, s=0)!
[2014-02-27 15:21:35] GPU #1: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #1: using launch configuration T14x32
[2014-02-27 15:21:35] GPU #3: Given launch config 'T14x32' exceeds warp limit fo
r 'T' kernel.
[2014-02-27 15:21:35] GPU #3: using launch configuration T14x32
[2014-02-27 15:21:36] GPU #1: GeForce GTX TITAN, 3580 khash/s
[2014-02-27 15:21:36] GPU #3: GeForce GTX TITAN, 2863 khash/s
[2014-02-27 15:21:36] GPU #0: GeForce GTX TITAN result does not validate on CPU
(i=3332, s=0)!
[2014-02-27 15:21:37] GPU #3: GeForce GTX TITAN result does not validate on CPU
(i=846, s=0)!
[2014-02-27 15:21:37] GPU #1: GeForce GTX TITAN result does not validate on CPU
(i=5892, s=0)!
[2014-02-27 15:21:37] Stratum detected new block
[2014-02-27 15:21:37] GPU #0: GeForce GTX TITAN result does not validate on CPU
(i=5958, s=0)!
[2014-02-27 15:21:38] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=9095, s=0)!
[2014-02-27 15:21:41] GPU #0: GeForce GTX TITAN, 4334 khash/s
[2014-02-27 15:21:41] GPU #3: GeForce GTX TITAN, 3085 khash/s
[2014-02-27 15:21:41] GPU #1: GeForce GTX TITAN, 5549 khash/s
[2014-02-27 15:21:41] GPU #2: GeForce GTX TITAN, 6212 khash/s
[2014-02-27 15:21:41] GPU #3: GeForce GTX TITAN result does not validate on CPU
(i=11930, s=1)!
[2014-02-27 15:21:41] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=532, s=1)!
[2014-02-27 15:21:41] GPU #1: GeForce GTX TITAN result does not validate on CPU
(i=7292, s=0)!
[2014-02-27 15:21:41] GPU #1: GeForce GTX TITAN result does not validate on CPU
(i=11114, s=0)!
[2014-02-27 15:21:41] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=13855, s=0)!
[2014-02-27 15:21:42] GPU #1: GeForce GTX TITAN result does not validate on CPU
(i=13375, s=1)!
[2014-02-27 15:21:42] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=7333, s=0)!
[2014-02-27 15:21:42] GPU #0: GeForce GTX TITAN result does not validate on CPU
(i=10406, s=0)!
[2014-02-27 15:21:42] GPU #2: GeForce GTX TITAN result does not validate on CPU
(i=12933, s=1)!
[2014-02-27 15:21:42] GPU #1: GeForce GTX TITAN result does not validate on CPU
(i=4438, s=0)!
legendary
Activity: 1154
Merit: 1001
Christian, did you give a second thought to MYR? This is the coin with multiple algorithms running at the same time. A standalone miner w/source code was just posted for the one algorithm that is currently CPU only (groestl). Me smells an opportunity to sneak in with a CUDA blast! Smiley

does it mean all of these algorithms have to run on CUDA, or any one algorithm can be run on CUDA?
I am not too fond of porting over a bunch of algorithms to CUDA.

Anyway, I want to try m myself at blake256 first as this shouldn't be any harder than Keccak, which I did in 24 hours.

Christian


Each of the algorithms runs separate difficulties & miners, so it's as if it was not 1 coin, but 5 different coins all sharing the same rewards.
In other words, you'd need to add cuda support for 'groestl' only, and we'd have the 1st & only gpu miner for that algorithm.

On a completely unrelated topic: do your 'pals' over at nvidia give you any hints on launch dates for the high end 800 series? I'm greatly impressed by the 750's & 750 TI's, can't wait for the high performance (and super efficient) siblings to come about.

Thanks for the feedback though! Cheers!
Myagui
sr. member
Activity: 350
Merit: 250
christian is it basic --algo=keccak i want to run for helixcoin?

im not sure if keccak has N factors?? yehhh can tell i havent been very active lately
hero member
Activity: 676
Merit: 500
Run it and you will see  Grin Grin Grin Grin Grin a booost..
sr. member
Activity: 490
Merit: 254
Ok, got my parts in and my initial 2 GIGABYTE GV-N75TOC-2GI GTX 750 Ti's in, and have everything up and running. So far I am only managing to get ~280 kHash out of each one. Running Windows 7 64 with latest Nvidia drivers and latest cudaminer.

My launch config is -m0 -l T5x24 -I 0 -C 1 -H 1

Tried different values for m, C and H and no difference. I am also running a slight +50 core and +300 memory overclock. I was hoping to see the 320 kHash everyone else seems to be getting but running out of ideas. Any insight or ideas of things to try would be appreciated.
Opened Chrome?

No I do not have Chrome running, does it cause slowdowns?
sr. member
Activity: 490
Merit: 254
My launch config is -m0 -l T5x24 -I 0 -C 1 -H 1

upper case I is not legal, or not the intended option. -C 1 is ignored by the T kernel.

does it crash when going above +50 MHz overclock?

Christian


Thanks, I'll fix those settings. No it does not crash, and it really seems to have little to no impact on my hash rate. Using Precision-X and I see the frequencies go up, but the hash rates seem to stay consistent at 280 kHash even when going to +100 on core and +500 on memory.
sr. member
Activity: 350
Merit: 250
You mean Helix or Exe?

Helix
i will use my maxcoin launch config, but need to change some things to suit the coin i believe?

cudaminer.exe --algo=keccak -d gtx780  -i 0 -m 1 -l K2304x32 -L 1024 -o 127.0.0.1 -O user:pass -D

for anyone wondering why helix coin is being talked about. its a pokemon based coin, for anyone that didnt know, twitch is having huge issues right now due to pokemon tournaments causing huge server load i believe it is?
member
Activity: 69
Merit: 10
anyone with a good .bat for microcoin for gtx 780 3GD? I used to mine with 400 each card but after the N factor went up i get less than 100 with both cards...

EDIT with autotune I get around 185 with each...
Jump to: