[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 226.

Welp. Managed to split the most offensive part of the kernel into four parallel threads per hash, result is spectacularly unimpressive. The best I've come up with breaks even with the current single thread per hash implementation. Well, almost. It's actually a percent slower AND loses compute 2.0 compatibility due to using shuffle. On the other hands it performs a lot more reasonably with various launch configurations, 15 blocks of 32 threads works our equally well as the original 8x60 magic bullet for 750 Ti.

At this point I'm starting to think I'll just forget about that part and start looking if there's something else to be improved. I'm still curious as to how it runs on other hardware, so if a couple of gents on Win boxes with something else than a 750 Ti in would be willing to take it for a spin, I'd appreciate it. I've added the number for SMX/SMM/Whateverthingmabobs into the miner thread start-up info, you'll probably find your card performing best when the block count is a multiple of the SMX count and the number of threads a power of 2. 4/8/16/32/64 are the best bets.

https://github.com/tsiv/ccminer-cryptonight/releases/download/v0.15-rc1/ccminer-cryptonight_20140723_exp.zip

Also, any chances for this code to get released already? Or are you competing against Wolf0 Cheesy

It works like a charm, 220H/s for GTX760, before it was 190. GTX750TIs seem unchanged.

I get 270H(peaks of 297H with -l 8x50) with this release and a GTX 760 overclocked -->v0.15-rc1 ccminer-cryptonight_20140723

Thanks for that launch setting Cheesy

306H/s (MSI gaming, +180core, +500mem). Still have to test what's the most stable, but thanks for giving me a start Wink

Ooh damn, you've released that a looong time ago, tsiv. Should've noticed ^^"

EDIT: 320H/s with +222core, +666mem Tongue

I'm waiting anxiously for a driver crash Wink

ilovecudacompute

member

Activity: 112

Merit: 10

Quote from: bathrobehero on July 23, 2014, 03:04:28 PM

Welp. Managed to split the most offensive part of the kernel into four parallel threads per hash, result is spectacularly unimpressive. The best I've come up with breaks even with the current single thread per hash implementation. Well, almost. It's actually a percent slower AND loses compute 2.0 compatibility due to using shuffle. On the other hands it performs a lot more reasonably with various launch configurations, 15 blocks of 32 threads works our equally well as the original 8x60 magic bullet for 750 Ti.

At this point I'm starting to think I'll just forget about that part and start looking if there's something else to be improved. I'm still curious as to how it runs on other hardware, so if a couple of gents on Win boxes with something else than a 750 Ti in would be willing to take it for a spin, I'd appreciate it. I've added the number for SMX/SMM/Whateverthingmabobs into the miner thread start-up info, you'll probably find your card performing best when the block count is a multiple of the SMX count and the number of threads a power of 2. 4/8/16/32/64 are the best bets.

https://github.com/tsiv/ccminer-cryptonight/releases/download/v0.15-rc1/ccminer-cryptonight_20140723_exp.zip

Also, any chances for this code to get released already? Or are you competing against Wolf0 Cheesy

It works like a charm, 220H/s for GTX760, before it was 190. GTX750TIs seem unchanged.

I get 270H(peaks of 297H with -l 8x50) with this release and a GTX 760 overclocked -->v0.15-rc1 ccminer-cryptonight_20140723

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: djm34 on July 23, 2014, 01:58:14 PM

I updated, again, my github so it uses cuda 5.5 (I started the compilation to check it works correctly... )

I did the same as I only have 5.5.

no matter what, I don't understand why the net hashrate of that coin is 1.2GHash/s, there are 160ppl in buy order on coinswap and ~~500ppl~~ (that's the sell order) 5ppl on allcrypt (for 0.004btc...). Refrigeratorcoin from ocminer has a better chance of success... Grin

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Quote from: djm34 on July 23, 2014, 01:58:14 PM

I updated, again, my github so it uses cuda 5.5 (I started the compilation to check it works correctly... )

I did the same as I only have 5.5.

djm34

legendary

Activity: 1400

Merit: 1050

I have uploaded again to github, this time the files should show up in the right place in visual studio...

ilovecudacompute

member

Activity: 112

Merit: 10

I downloaded the new nvidia drivers djm posted with the new cuda software and his x17 binaries work fine
1600KH with asus 760

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: Amph on July 23, 2014, 02:10:15 PM

Quote from: djm34 on July 23, 2014, 01:47:42 PM

Quote from: bathrobehero on July 23, 2014, 12:27:38 PM

Thanks djm!

ccminer djm34 commit 54 x64 CUDA 5.5 binaries:
http://www.mediafire.com/download/9382f3s6q2nc62s/ccminer+djm34+c54+x64+CUDA+5.5.zip

X17 hashrate is very close to x15.
I don't mind doing slow compiles.

can you upload the source for this?

you don't need the entire new source, you just need to fix your ccminer.vxproj (because you messed with it sorry... I don't take responsibility for that Grin

)

i just dragged new .cu under x13 folder in visual studio, was that maybe lol

they should be already in x13 (in visual studio), I don't understand why they don't want to stick...

Amph

legendary

Activity: 3248

Merit: 1070

Quote from: djm34 on July 23, 2014, 01:47:42 PM

Quote from: bathrobehero on July 23, 2014, 12:27:38 PM

Thanks djm!

ccminer djm34 commit 54 x64 CUDA 5.5 binaries:
http://www.mediafire.com/download/9382f3s6q2nc62s/ccminer+djm34+c54+x64+CUDA+5.5.zip

X17 hashrate is very close to x15.
I don't mind doing slow compiles.

can you upload the source for this?

you don't need the entire new source, you just need to fix your ccminer.vxproj (because you messed with it sorry... I don't take responsibility for that Grin

)

i just dragged new .cu under x13 folder in visual studio, was that maybe lol

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: bathrobehero on July 23, 2014, 01:50:22 PM

Quote from: bathrobehero on July 23, 2014, 12:27:38 PM

Thanks djm!

ccminer djm34 commit 54 x64 CUDA 5.5 binaries:
http://www.mediafire.com/download/9382f3s6q2nc62s/ccminer+djm34+c54+x64+CUDA+5.5.zip

X17 hashrate is very close to x15.
I don't mind doing slow compiles.

can you upload the source for this?

Edit: It's from https://github.com/djm34/ccminer , I'm just sticking with cuda 5.5.

I updated, again, my github so it uses cuda 5.5 (I started the compilation to check it works correctly... )

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Quote from: bathrobehero on July 23, 2014, 12:27:38 PM

Thanks djm!

ccminer djm34 commit 54 x64 CUDA 5.5 binaries:
http://www.mediafire.com/download/9382f3s6q2nc62s/ccminer+djm34+c54+x64+CUDA+5.5.zip

X17 hashrate is very close to x15.
I don't mind doing slow compiles.

can you upload the source for this?

Edit: It's from https://github.com/djm34/ccminer , I'm just sticking with cuda 5.5.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: bathrobehero on July 23, 2014, 12:27:38 PM

Thanks djm!

ccminer djm34 commit 54 x64 CUDA 5.5 binaries:
http://www.mediafire.com/download/9382f3s6q2nc62s/ccminer+djm34+c54+x64+CUDA+5.5.zip

X17 hashrate is very close to x15.
I don't mind doing slow compiles.

can you upload the source for this?

you don't need the entire new source, you just need to fix your ccminer.vxproj (because you messed with it sorry... I don't take responsibility for that Grin

)

Equitum

newbie

Activity: 29

Merit: 0

Quote from: DrAlco on July 23, 2014, 09:52:59 AM

Quote from: tsiv on July 23, 2014, 07:19:07 AM

Quote from: DrAlco on July 23, 2014, 06:23:21 AM

Quote from: bathrobehero on July 23, 2014, 12:27:38 PM

At this point I'm starting to think I'll just forget about that part and start looking if there's something else to be improved. I'm still curious as to how it runs on other hardware, so if a couple of gents on Win boxes with something else than a 750 Ti in would be willing to take it for a spin, I'd appreciate it. I've added the number for SMX/SMM/Whateverthingmabobs into the miner thread start-up info, you'll probably find your card performing best when the block count is a multiple of the SMX count and the number of threads a power of 2. 4/8/16/32/64 are the best bets.

https://github.com/tsiv/ccminer-cryptonight/releases/download/v0.15-rc1/ccminer-cryptonight_20140723_exp.zip

Improved hashrate of about 70H/s on a 780ti. Up from 320 to about 390 (using 8x60). Also doesn't seem to hang and bring the system to it's knees when using all GFX cards.

Seems to be in line with the ~18% improvements I saw when benchmarking only the AES part of the kernel. Have you tried other configs? 390 is still pretty low for a 780 Ti, I think people were getting best results with 4x120 on the 780 Ti.

Getting about 420h/s with 4x120.

I'm getting 505-525 h/s (short run just for benchmark) with my 780 Ti Classified with 6x120 (up from ~475 h/s on the 6/30 release of tsiv's cryptonight ccminer).

Amph

legendary

Activity: 3248

Merit: 1070

Thanks djm!

ccminer djm34 commit 54 x64 CUDA 5.5 binaries:
http://www.mediafire.com/download/9382f3s6q2nc62s/ccminer+djm34+c54+x64+CUDA+5.5.zip

X17 hashrate is very close to x15.
I don't mind doing slow compiles.

can you upload the source for this?

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: bathrobehero on July 23, 2014, 12:27:38 PM

Thanks djm!

ccminer djm34 commit 54 x64 CUDA 5.5 binaries:
http://www.mediafire.com/download/9382f3s6q2nc62s/ccminer+djm34+c54+x64+CUDA+5.5.zip

X17 hashrate is very close to x15.
I don't mind doing slow compiles.

Thanks a lot.
yes sha512 and haval256 are very fast...

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Thanks djm!

ccminer djm34 commit 54 x64 CUDA 5.5 binaries:
http://www.mediafire.com/download/9382f3s6q2nc62s/ccminer+djm34+c54+x64+CUDA+5.5.zip

X17 hashrate is very close to x15.
I don't mind doing slow compiles.

Alpharisc

newbie

Activity: 47

Merit: 0

Anyone got a link to the cuda 6.5 linux release files?

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: bathrobehero on July 23, 2014, 11:26:10 AM

Error 169 error LNK2001: unresolved external symbol scanhash_x17 \cpu-miner.obj ccminer

forgot to update vxproj... can you retry now... (that's the problem with github... I don't want to get generated files into github, and I need to finalyze change on vxproj somewhere else...)

Bombadil

hero member

Activity: 644

Merit: 500

Quote from: zelante on July 23, 2014, 11:24:22 AM

Welp. Managed to split the most offensive part of the kernel into four parallel threads per hash, result is spectacularly unimpressive. The best I've come up with breaks even with the current single thread per hash implementation. Well, almost. It's actually a percent slower AND loses compute 2.0 compatibility due to using shuffle. On the other hands it performs a lot more reasonably with various launch configurations, 15 blocks of 32 threads works our equally well as the original 8x60 magic bullet for 750 Ti.

At this point I'm starting to think I'll just forget about that part and start looking if there's something else to be improved. I'm still curious as to how it runs on other hardware, so if a couple of gents on Win boxes with something else than a 750 Ti in would be willing to take it for a spin, I'd appreciate it. I've added the number for SMX/SMM/Whateverthingmabobs into the miner thread start-up info, you'll probably find your card performing best when the block count is a multiple of the SMX count and the number of threads a power of 2. 4/8/16/32/64 are the best bets.

https://github.com/tsiv/ccminer-cryptonight/releases/download/v0.15-rc1/ccminer-cryptonight_20140723_exp.zip

Also, any chances for this code to get released already? Or are you competing against Wolf0 Cheesy

It works like a charm, 220H/s for GTX760, before it was 190. GTX750TIs seem unchanged.

What speed for 750ti ?

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

Error 169 error LNK2001: unresolved external symbol scanhash_x17 \cpu-miner.obj ccminer

zelante

full member

Activity: 263

Merit: 100