ECM for CUDA GPUs in latest GMP-ECM ? - Page 18

wombatman · 2013-08-28, 13:13

I'm pretty sure there's one in here somewhere, though I don't remember exactly where.

wombatman · 2013-08-30, 19:14

Is there a simple way to only run, say, stage 1 of GPU-ECM? I'd like to run the stage 1 on my desktop's GPU and then run stage 2 on my laptop (64-bit processor). I can churn through 480 stage 1 curves at a time on my GPU, but then my (relatively) ancient processor takes an inordinately long amount of time to do the stage 2 work before the residues show up in the save file. So if at all possible, I'd like to cut out stage 2 and simply use "-resume" with the residues file.

Edit: And just like that, I seem to have answered my own question. Setting B2=0 appears to have done it.

skan · 2013-10-08, 16:45

Hello

Can this gpu-ecm.exe executable be used with yafu by just renaming it?
Or it's not compatible?

wombatman · 2013-10-08, 16:48

I imagine it would work fine by directing yafu to it, but you wouldn't be using the GPU to actually run the curves. As such, it would be simpler to use the standard GMP-ECM. In addition, there's a bug (at least in the versions I've compiled) where factors that should be found in Stage 2 are not found by the GPU version. So you'd likely miss factors if you tried to use the GPU-enabled version.

frmky · 2013-10-10, 00:55

Has anyone gotten save/resume to work? Here's what I get:

Code:

./ecm -gpu -save c192.ecm 4e4 0 < C192
GMP-ECM 7.0-dev [configured with GMP 5.1.3, --enable-asm-redc, --enable-gpu, --enable-assert, --enable-openmp] [ECM]
Input number is 361867761627749619574674677769104106609222942556126375131641520128131226876942053491915389383216114202550615939995196059592933366284139274937165773815556682024285008268664983092158298987980517 (192 digits)
Using B1=40000, B2=0, sigma=3:3642364246-3:3642365077 (832 curves)
Computing 832 Step 1 took 472ms of CPU time / 21426ms of GPU time

This gives a save file like

Code:

METHOD=ECM; PARAM=3; SIGMA=3642365077; B1=1; N=361867761627749619574674677769104106609222942556126375131641520128131226876942053491915389383216114202550615939995196059592933366284139274937165773815556682024285008268664983092158298987980517; X=0x7cb7309ea2ec4ee5d88962fa7516f43385ad69f16e28b04b8beee48d15091cf009cd4119d57a505d6c062036b1d8a784d11d4d87873df12d62bc03b9f10e7a1671e666030d895d55d22d5da9743992e; CHECKSUM=3136759905; PROGRAM=GMP-ECM 7.0-dev; X0=0x0; Y0=0x0; WHO=cluster@compute-0-1.local; TIME=Wed Oct  9 17:49:08 2013;

Then on resume:

Code:

./ecm -resume c192.ecm 4e4
GMP-ECM 7.0-dev [configured with GMP 5.1.3, --enable-asm-redc, --enable-gpu, --enable-assert, --enable-openmp] [ECM]
Resuming ECM residue saved by cluster@compute-0-1.local with GMP-ECM 7.0-dev on Wed Oct  9 17:49:08 2013 
Input number is 361867761627749619574674677769104106609222942556126375131641520128131226876942053491915389383216114202550615939995196059592933366284139274937165773815556682024285008268664983092158298987980517 (192 digits)
Error, x0 should be equal to 2 with this parametrization
Please report internal errors at <ecm-discuss@lists.gforge.inria.fr>.

Does this work for anyone? Thanks!

wombatman · 2013-10-10, 01:15

Sorry to have edited so many times, but you're using the right command--it looks like the issue is with the B1 being wrong in the save file. I just ran with the GPU using the input number, and I have no issues using -resume. I don't remember what was done to fix the B1 save file issue though.

wombatman · 2013-10-10, 02:20

Here's a post that might have a fix for the B1=1 bug:

http://mersenneforum.org/showpost.ph...&postcount=174

Also, my version is 32-bit. Are you using a 32 or 64 bit build? I went ahead and ran 480 curves of your C192 at 4e4--no factors found.

frmky · 2013-10-10, 05:53

Quote:

Originally Posted by wombatman

Here's a post that might have a fix for the B1=1 bug:

http://mersenneforum.org/showpost.ph...&postcount=174

Also, my version is 32-bit. Are you using a 32 or 64 bit build? I went ahead and ran 480 curves of your C192 at 4e4--no factors found.

Thanks for the link. That workaround seems to fix it. I'm using the 64-bit version.

xilman · 2013-10-10, 06:59

Quote:

Originally Posted by frmky

Has anyone gotten save/resume to work? Here's what I get:

Code:

./ecm -gpu -save c192.ecm 4e4 0 < C192
GMP-ECM 7.0-dev [configured with GMP 5.1.3, --enable-asm-redc, --enable-gpu, --enable-assert, --enable-openmp] [ECM]
Input number is 361867761627749619574674677769104106609222942556126375131641520128131226876942053491915389383216114202550615939995196059592933366284139274937165773815556682024285008268664983092158298987980517 (192 digits)
Using B1=40000, B2=0, sigma=3:3642364246-3:3642365077 (832 curves)
Computing 832 Step 1 took 472ms of CPU time / 21426ms of GPU time

This gives a save file like

Code:

METHOD=ECM; PARAM=3; SIGMA=3642365077; B1=1; N=361867761627749619574674677769104106609222942556126375131641520128131226876942053491915389383216114202550615939995196059592933366284139274937165773815556682024285008268664983092158298987980517; X=0x7cb7309ea2ec4ee5d88962fa7516f43385ad69f16e28b04b8beee48d15091cf009cd4119d57a505d6c062036b1d8a784d11d4d87873df12d62bc03b9f10e7a1671e666030d895d55d22d5da9743992e; CHECKSUM=3136759905; PROGRAM=GMP-ECM 7.0-dev; X0=0x0; Y0=0x0; WHO=cluster@compute-0-1.local; TIME=Wed Oct  9 17:49:08 2013;

Then on resume:

Code:

./ecm -resume c192.ecm 4e4
GMP-ECM 7.0-dev [configured with GMP 5.1.3, --enable-asm-redc, --enable-gpu, --enable-assert, --enable-openmp] [ECM]
Resuming ECM residue saved by cluster@compute-0-1.local with GMP-ECM 7.0-dev on Wed Oct  9 17:49:08 2013 
Input number is 361867761627749619574674677769104106609222942556126375131641520128131226876942053491915389383216114202550615939995196059592933366284139274937165773815556682024285008268664983092158298987980517 (192 digits)
Error, x0 should be equal to 2 with this parametrization
Please report internal errors at <ecm-discuss@lists.gforge.inria.fr>.

Does this work for anyone? Thanks!

Bizarre! I've never seen the B1 bug which is why I've been plugging the workaround.

Has this been reported to the dev team?

wombatman · 2013-10-10, 13:21

No idea. I think the save/resume bug is known to at least one developer (Cyril?), but I don't know about the B1=1 bug.

xilman · 2013-10-10, 15:17

Quote:

Originally Posted by wombatman

No idea. I think the save/resume bug is known to at least one developer (Cyril?), but I don't know about the B1=1 bug.

Ho hum. Some times my style is just too oblique and a hint has to be hammered in with a length of two by four.

To state the bleedin' obvious: the bug should be reported to the GMP-ECM team, whether or not they've already heard of it.

2013-08-30, 19:14	#189
wombatman I moo ablest echo power! May 2013 11011001101₂ Posts	Is there a simple way to only run, say, stage 1 of GPU-ECM? I'd like to run the stage 1 on my desktop's GPU and then run stage 2 on my laptop (64-bit processor). I can churn through 480 stage 1 curves at a time on my GPU, but then my (relatively) ancient processor takes an inordinately long amount of time to do the stage 2 work before the residues show up in the save file. So if at all possible, I'd like to cut out stage 2 and simply use "-resume" with the residues file. Edit: And just like that, I seem to have answered my own question. Setting B2=0 appears to have done it. Last fiddled with by wombatman on 2013-08-30 at 19:16

2013-10-10, 01:15	#193
wombatman I moo ablest echo power! May 2013 6CD₁₆ Posts	Sorry to have edited so many times, but you're using the right command--it looks like the issue is with the B1 being wrong in the save file. I just ran with the GPU using the input number, and I have no issues using -resume. I don't remember what was done to fix the B1 save file issue though. Last fiddled with by wombatman on 2013-10-10 at 01:26

2013-10-10, 02:20	#194
wombatman I moo ablest echo power! May 2013 1,741 Posts	Here's a post that might have a fix for the B1=1 bug: http://mersenneforum.org/showpost.ph...&postcount=174 Also, my version is 32-bit. Are you using a 32 or 64 bit build? I went ahead and ran 480 curves of your C192 at 4e4--no factors found. Last fiddled with by wombatman on 2013-10-10 at 02:21

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Running CUDA on non-Nvidia GPUs	Rodrigo	GPU Computing	3	2016-05-17 05:43
Error in GMP-ECM 6.4.3 and latest svn	ATH	GMP-ECM	10	2012-07-29 17:15
latest SVN 1677	ATH	GMP-ECM	7	2012-01-07 18:34
Has anyone seen my latest treatise?	davieddy	Lounge	0	2011-01-21 19:29
Latest version?	[CZ]Pegas	Software	3	2002-08-23 17:05

2013-08-28, 13:13	#188
wombatman I moo ablest echo power! May 2013 1,741 Posts	I'm pretty sure there's one in here somewhere, though I don't remember exactly where.

2013-10-08, 16:45	#190
skan Apr 2012 2·47 Posts	Hello Can this gpu-ecm.exe executable be used with yafu by just renaming it? Or it's not compatible?

2013-10-08, 16:48	#191
wombatman I moo ablest echo power! May 2013 3315₈ Posts	I imagine it would work fine by directing yafu to it, but you wouldn't be using the GPU to actually run the curves. As such, it would be simpler to use the standard GMP-ECM. In addition, there's a bug (at least in the versions I've compiled) where factors that should be found in Stage 2 are not found by the GPU version. So you'd likely miss factors if you tried to use the GPU-enabled version.

2013-10-10, 13:21	#197
wombatman I moo ablest echo power! May 2013 1,741 Posts	No idea. I think the save/resume bug is known to at least one developer (Cyril?), but I don't know about the B1=1 bug.