CUDALucas (a.k.a. MaclucasFFTW/CUDA 2.3/CUFFTW) - Page 180

LaurV · 2013-11-15, 03:03

Well, if it gives me negative credit numbers when I report the results*, I won't object too much

-----
* like instead getting 30GHzDays when I report a DC, I would get a -30GHzDays, which in hex would be (use windows calculator) FFFF FFFF FFFF FFE2, or 18446744073709551586 GHzDays... In fact I would be ok even with a 32 bit value... (4294967266 GHzDays)

ET_ · 2013-11-15, 11:54

Quote:

Originally Posted by LaurV

Well, if it gives me negative credit numbers when I report the results*, I won't object too much

-----
* like instead getting 30GHzDays when I report a DC, I would get a -30GHzDays, which in hex would be (use windows calculator) FFFF FFFF FFFF FFE2, or 18446744073709551586 GHzDays... In fact I would be ok even with a 32 bit value... (4294967266 GHzDays)

2^32 - 6GB = -2Gb

Luigi

James Heinrich · 2013-11-15, 14:29

How much RAM does the card actually have? 2GB? 4GB?

owftheevil · 2013-11-15, 14:37

Quote:

Originally Posted by Antonio

CUDALucas seems to think my GT640 has a hole where it's memory should be, it says I have (minus)2GiB totalGlobalmem!
I find this strangely disquieting for software which is dealing with large numbers

This should be reported correctly in 2.05.

Antonio · 2013-11-15, 22:52

Quote:

Originally Posted by owftheevil

This should be reported correctly in 2.05.

The -2GiB is reported by 2.05-Beta-x64, downloaded today.
The card has +2GiB of memory installed.

kladner · 2013-11-16, 04:51

I am having a pretty rough time with 2.05 on the GTX 580.
'CUDALucas -cufftbench 1 8192 1' crashes on GTX 580, brings down graphic driver 327.23 (which restarts). 782 MHz core, 1600 VRAM

This is just the latest test of many. Occasionally, the test completes.

Rolling back the driver from 331.65 to 327.23 made no difference.

2.04-beta successfully completes
'CUDALucas -cufftbench 32768 3276800 32768' and has turned in good DCs at 830 MHz core, 1600 VRAM.

I haven't yet tried running a DC on 2.05-beta.

I have the card throttled back from where it normally runs mfaktc to stock: from 844 MHz to 782 MHz. The RAM is 400 MHz below stock.

Any suggestions would be appreciated.

EDIT: Tried running an exponent, 30651xxx on 2.05, 830 MHz core, 1600 VRAM. Started with 1728K, instead of stepping up to it from 1600K, as 2.04 did. Crashed a bit after the 40,000th iteration.

Code:

Iteration 10000 M( 30651671 )C, 0x6b79bd6d5adfb7de, n = 1728K, CUDALucas v2.05 Beta err = 0.05396 (0:26 real, 2.8857 ms/iter, ETA 24:33:42)
Iteration 20000 M( 30651671 )C, 0x53064732900985e9, n = 1728K, CUDALucas v2.05 Beta err = 0.06055 (0:26 real, 2.6133 ms/iter, ETA 22:14:09)
Iteration 30000 M( 30651671 )C, 0xe85abecfe0f40dce, n = 1728K, CUDALucas v2.05 Beta err = 0.05469 (0:26 real, 2.6123 ms/iter, ETA 22:13:12)
Iteration 40000 M( 30651671 )C, 0xa4208cf27dd73713, n = 1728K, CUDALucas v2.05 Beta err = 0.06250 (0:26 real, 2.6123 ms/iter, ETA 22:12:48)
CUDALucas.cu(310) : cudaSafeCall() Runtime API error 30: unknown error.

Antonio · 2013-11-16, 07:05

Quote:

Originally Posted by kladner

I am having a pretty rough time with 2.05 on the GTX 580.
'CUDALucas -cufftbench 1 8192 1' crashes on GTX 580, brings down graphic driver 327.23 (which restarts). 782 MHz core, 1600 VRAM

This is just the latest test of many. Occasionally, the test completes.

Rolling back the driver from 331.65 to 327.23 made no difference.

2.04-beta successfully completes
'CUDALucas -cufftbench 32768 3276800 32768' and has turned in good DCs at 830 MHz core, 1600 VRAM.

I haven't yet tried running a DC on 2.05-beta.

I have the card throttled back from where it normally runs mfaktc to stock: from 844 MHz to 782 MHz. The RAM is 400 MHz below stock.

Any suggestions would be appreciated.

EDIT: Tried running an exponent, 30651xxx on 2.05, 830 MHz core, 1600 VRAM. Started with 1728K, instead of stepping up to it from 1600K, as 2.04 did. Crashed a bit after the 40,000th iteration.

I tried your exponent on my GT 640, graphics driver 331.65, and it ran ok past your fail point (see below). So it looks like you may still have a hardware problem.

Code:

Iteration 10000 M( 30651671 )C, 0x6b79bd6d5adfb7de, n = 1728K, CUDALucas v2.05 Beta err = 0.05859 (2:21 real, 15.7012 ms/iter, ETA 133:38:30)
Iteration 20000 M( 30651671 )C, 0x53064732900985e9, n = 1728K, CUDALucas v2.05 Beta err = 0.05664 (2:37 real, 15.6991 ms/iter, ETA 133:34:50)
Iteration 30000 M( 30651671 )C, 0xe85abecfe0f40dce, n = 1728K, CUDALucas v2.05 Beta err = 0.05469 (2:37 real, 15.6950 ms/iter, ETA 133:30:06)
Iteration 40000 M( 30651671 )C, 0xa4208cf27dd73713, n = 1728K, CUDALucas v2.05 Beta err = 0.05640 (2:37 real, 15.6912 ms/iter, ETA 133:25:34)
Iteration 50000 M( 30651671 )C, 0x056716c2203b5c29, n = 1728K, CUDALucas v2.05 Beta err = 0.05859 (2:37 real, 15.6935 ms/iter, ETA 133:24:08)
Iteration 60000 M( 30651671 )C, 0x5da5d75c80f2587c, n = 1728K, CUDALucas v2.05 Beta err = 0.05469 (2:37 real, 15.6927 ms/iter, ETA 133:21:05)

( I know it's slow - but that DDR3 memory is incredibly reliable

)

kladner · 2013-11-16, 07:47

Of course, it could always be hardware, but I did subsequently run the exponent up to ~750K it. with more aggressive core clock (830 MHz). As mentioned previously, the VRAM at 1600 has performed well in the past with 2.04-beta. I have the libraries up through 5.5.

Still just tossing things out there.

Antonio · 2013-11-16, 08:20

Quote:

Originally Posted by kladner

Of course, it could always be hardware, but I did subsequently run the exponent up to ~750K it. with more aggressive core clock (830 MHz). As mentioned previously, the VRAM at 1600 has performed well in the past with 2.04-beta. I have the libraries up through 5.5.

Still just tossing things out there.

Just a thought: are you using this card to drive your display as well as run CUDALucas?
I've had problems in the past with (if my somewhat old and dimmed memory serves me correctly) very similar error reports, when I've tried using my display card for various CUDA work. Could it be a memory conflict when the screen is updated? (The GT 640 I'm using is not driving my display, I have a GTX 650 Ti for that).

kladner · 2013-11-16, 16:28

Quote:

Originally Posted by Antonio

Just a thought: are you using this card to drive your display as well as run CUDALucas?
I've had problems in the past with (if my somewhat old and dimmed memory serves me correctly) very similar error reports, when I've tried using my display card for various CUDA work. Could it be a memory conflict when the screen is updated? (The GT 640 I'm using is not driving my display, I have a GTX 650 Ti for that).

Good point! Yes, the 580 is driving the display. I'll have to have a look at that. I'll try switching the display to the GTX 570. For that matter, I'll have another shot at running CL 2.05-beta on the 570.

I'm currently back running the 331.65 driver, since the roll back didn't seem to make any difference.

Thanks for the suggestion! I needed a new lead to follow. I currently have both cards back to doing LL-TF, but I do like to figure out things which don't work as they should.

EDIT: Clarification of fuzzy thoughts from late night experiments:

Quote:

Originally Posted by kladner

Of course, it could always be hardware, but I did subsequently run the exponent up to ~750K it. with more aggressive core clock (830 MHz). As mentioned previously, the VRAM at 1600 has performed well in the past with 2.04-beta.

The above refers to running 2.04-beta on the 580 card.

EDIT2: I also tried running with threads at 512 instead of the default 256, but it did not seem to make any difference.

flashjh · 2013-11-16, 16:58

I just completed a DC on M57885161 with CUDALucas 2.05-Beta-x64. It completed without error and I even switched FFT sizes a few times. Since I have the full run of residues from the first time I ran it, I was able to check progress along the way.

The only issue I found so far was keyboard input. If Interactive=n is set to 1 in the .ini file then anytime I pressed a key the program would stop progress. GPU usage dropped to about 50% but ^c still stopped the run. I could restart with no problems. Anyone else seen this in Windows or Linux? Can some others test this to see if it's working or not in Windows and Linux?

I haven't run all the FFT benchmarks yet, I'll do that now. Anyone else having a problem with the amount of memory reported by CUDALucas?

2013-11-15, 14:29	#1972
James Heinrich "James Heinrich" May 2004 ex-Northern Ontario 23×149 Posts	How much RAM does the card actually have? 2GB? 4GB? Last fiddled with by James Heinrich on 2013-11-15 at 14:30

2013-11-16, 04:51	#1975
kladner "Kieren" Jul 2011 In My Own Galaxy! 2×3×1,693 Posts	I am having a pretty rough time with 2.05 on the GTX 580. 'CUDALucas -cufftbench 1 8192 1' crashes on GTX 580, brings down graphic driver 327.23 (which restarts). 782 MHz core, 1600 VRAM This is just the latest test of many. Occasionally, the test completes. Rolling back the driver from 331.65 to 327.23 made no difference. 2.04-beta successfully completes 'CUDALucas -cufftbench 32768 3276800 32768' and has turned in good DCs at 830 MHz core, 1600 VRAM. I haven't yet tried running a DC on 2.05-beta. I have the card throttled back from where it normally runs mfaktc to stock: from 844 MHz to 782 MHz. The RAM is 400 MHz below stock. Any suggestions would be appreciated. EDIT: Tried running an exponent, 30651xxx on 2.05, 830 MHz core, 1600 VRAM. Started with 1728K, instead of stepping up to it from 1600K, as 2.04 did. Crashed a bit after the 40,000th iteration. Code: Iteration 10000 M( 30651671 )C, 0x6b79bd6d5adfb7de, n = 1728K, CUDALucas v2.05 Beta err = 0.05396 (0:26 real, 2.8857 ms/iter, ETA 24:33:42) Iteration 20000 M( 30651671 )C, 0x53064732900985e9, n = 1728K, CUDALucas v2.05 Beta err = 0.06055 (0:26 real, 2.6133 ms/iter, ETA 22:14:09) Iteration 30000 M( 30651671 )C, 0xe85abecfe0f40dce, n = 1728K, CUDALucas v2.05 Beta err = 0.05469 (0:26 real, 2.6123 ms/iter, ETA 22:13:12) Iteration 40000 M( 30651671 )C, 0xa4208cf27dd73713, n = 1728K, CUDALucas v2.05 Beta err = 0.06250 (0:26 real, 2.6123 ms/iter, ETA 22:12:48) CUDALucas.cu(310) : cudaSafeCall() Runtime API error 30: unknown error. Last fiddled with by kladner on 2013-11-16 at 05:11 Reason: th

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Don't DC/LL them with CudaLucas	LaurV	Data	131	2017-05-02 18:41
CUDALucas / cuFFT Performance on CUDA 7 / 7.5 / 8	Brain	GPU Computing	13	2016-02-19 15:53
CUDALucas: which binary to use?	Karl M Johnson	GPU Computing	15	2015-10-13 04:44
settings for cudaLucas	fairsky	GPU Computing	11	2013-11-03 02:08
Trying to run CUDALucas on Windows 8 CP	Rodrigo	GPU Computing	12	2012-03-07 23:20

2013-11-15, 03:03	#1970
LaurV Romulan Interpreter Jun 2011 Thailand 3×3,221 Posts	Well, if it gives me negative credit numbers when I report the results, I won't object too much ----- like instead getting 30GHzDays when I report a DC, I would get a -30GHzDays, which in hex would be (use windows calculator) FFFF FFFF FFFF FFE2, or 18446744073709551586 GHzDays... In fact I would be ok even with a 32 bit value... (4294967266 GHzDays)

2013-11-16, 07:47	#1977
kladner "Kieren" Jul 2011 In My Own Galaxy! 2×3×1,693 Posts	Of course, it could always be hardware, but I did subsequently run the exponent up to ~750K it. with more aggressive core clock (830 MHz). As mentioned previously, the VRAM at 1600 has performed well in the past with 2.04-beta. I have the libraries up through 5.5. Still just tossing things out there.

2013-11-16, 16:58	#1980
flashjh "Jerry" Nov 2011 Vancouver, WA 1,123 Posts	I just completed a DC on M57885161 with CUDALucas 2.05-Beta-x64. It completed without error and I even switched FFT sizes a few times. Since I have the full run of residues from the first time I ran it, I was able to check progress along the way. The only issue I found so far was keyboard input. If Interactive=n is set to 1 in the .ini file then anytime I pressed a key the program would stop progress. GPU usage dropped to about 50% but ^c still stopped the run. I could restart with no problems. Anyone else seen this in Windows or Linux? Can some others test this to see if it's working or not in Windows and Linux? I haven't run all the FFT benchmarks yet, I'll do that now. Anyone else having a problem with the amount of memory reported by CUDALucas?