mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   GPU Computing (https://www.mersenneforum.org/forumdisplay.php?f=92)
-   -   CUDALucas (a.k.a. MaclucasFFTW/CUDA 2.3/CUFFTW) (https://www.mersenneforum.org/showthread.php?t=12576)

wavelet3000 2011-01-23 03:14

This is awesome. How long did prime95 run?

I also did [URL="http://www.mersenne.org/report_exponent/?exp_lo=88880009&exp_hi=88880009&B1=Get+status"]M88880009[/URL] and [URL="http://www.mersenne.org/report_exponent/?exp_lo=88888883&exp_hi=88888883&B1=Get+status"]M88888883[/URL]
(both 8M fft). The first took ~45 days on GTX280. The second took 13.5 days on GTX580 OC. Could have done more but I was distracted by LL-D and TF ...

Thanks you, msft

msft 2011-01-23 13:15

[QUOTE=wavelet3000;248605]This is awesome. How long did prime95 run?
[/QUOTE]
Almost 40 days.
[QUOTE]
I also did [URL="http://www.mersenne.org/report_exponent/?exp_lo=88880009&exp_hi=88880009&B1=Get+status"]M88880009[/URL] and [URL="http://www.mersenne.org/report_exponent/?exp_lo=88888883&exp_hi=88888883&B1=Get+status"]M88888883[/URL]
(both 8M fft). The first took ~45 days on GTX280. The second took 13.5 days on GTX580 OC. Could have done more but I was distracted by LL-D and TF ...
[/QUOTE]
I try LL-D.

ixfd64 2011-01-23 16:14

Have you considered using IBDWT? It should make the squaring a lot faster.

msft 2011-01-24 11:38

Hi, ixfd64
[QUOTE=ixfd64;248726]Have you considered using IBDWT? It should make the squaring a lot faster.[/QUOTE]
This program use IBDWT.

ET_ 2011-01-29 12:39

New record, I guess...
 
M( 123999973 )C, 0xa387606bed18abXX, n = 8388608, MacLucasFFTW v8.1 Ballester

Thank you msft! :smile:

Luigi

msft 2011-01-30 03:56

Hi ,ET_
Thank you,

kjaget 2011-02-01 03:00

Windows x64 port!
 
1 Attachment(s)
I've done some minor porting to get this to build and run under Windows using MSVC 2008. There was nothing major that needed to change, just some fixes to get the code to use prototypes when building, use binary mode when opening checkpoint files, and a few #defines in the makefile.

So far I've run a number of double checks and all except one were ok - the remaining test (26396473) has 3 runs each with a different result, so I'm not too worried about it. The complete list also has some first time checks in the 39,xxx,xxx to 46,xxx,xxx range. The residues seem reasonable but if anyone wants to jump in and double-check them feel free.

I've included both the source code and a 64-bit windows binary. I had no luck getting a 32-bit binary working, but I didn't try that hard since my system is 64-bit. My hope is that I didn't break anything for Linux builds, but without a system to test on I'm not sure how I was. In any case, there weren't that many changes so I don't think I did too much damage :)

[ATTACH]6166[/ATTACH]

I don't have any specific benchmarks, but a 26,xxx,xxx doublecheck takes about 2.5 days on a GTX 275. I think that's in line with Linux performance, more or less. If you can give me instructions on how to get per-iteration timing I'll try it out.

Enjoy,
Kevin

Exponents tested with this code, by date :

[url]http://www.mersenne.org/report_exponent/?exp_lo=46536901&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24428827&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=26838041&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25747723&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24703337&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24703291&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24703333&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=26580581&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=39912157&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=39721139&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=26396473&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=26159473&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=26112841&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=26066839&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24256069&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24207529&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=23990413&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=23402713&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=23354899&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25218751&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25103329&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=26112893&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25103233&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25103191&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25096873&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=39207293&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=38692109&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25816907&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25776701&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25766989&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=25766953&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24614437&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24975571&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24509567&exp_hi=&B1=Get+status[/url]
[url]http://www.mersenne.org/report_exponent/?exp_lo=24495193&exp_hi=&B1=Get+status[/url]

^^^ Ironically, the first exponent I tested turned out to be one which mismatched. Scared me until I started getting more good results and eventually a matching triple-check.

msft 2011-02-01 13:22

Hi ,kjaget
Thank you,
Today is CUDALucas's second birthday!:lol:
[QUOTE=kjaget;250727]I don't have any specific benchmarks, but a 26,xxx,xxx doublecheck takes about 2.5 days on a GTX 275. I think that's in line with Linux performance, more or less. If you can give me instructions on how to get per-iteration timing I'll try it out.
[/QUOTE]
$ ./CUDALucas -c10000 33333333 &
$ ls -l c33333333 t33333333
-rw-r--r-- 1 msft msft 16777256 2011-02-01 08:05 c33333333
-rw-r--r-- 1 msft msft 16777256 2011-02-01 08:03 t33333333

2m*60sec/10000=12msec/iter

Prime95 2011-02-01 15:44

[QUOTE=kjaget;250727]I've done some minor porting to get this to build and run under Windows using MSVC 2008.[/QUOTE]

Thanks!

But where do I get cufft64_31_9.dll from?

Karl M Johnson 2011-02-01 16:28

[QUOTE=Prime95;250804]Thanks!

But where do I get cufft64_31_9.dll from?[/QUOTE]

CUDA 3.1 toolkit.

Karl M Johnson 2011-02-01 17:33

By the way, is there a way to restart a certain exponent from a certain iteration ?
Or, at least, from a checkpoint ?

I got a certain exponent to 90k iterations, then got a reboot(I had -c25k flag), and when I started it again, it went from the very beginning.


All times are UTC. The time now is 22:55.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.