CUDALucas (a.k.a. MaclucasFFTW/CUDA 2.3/CUFFTW)

[msft]

Hi, frmky

Originally Posted by frmky

Another 3 of 4 verified. Again, I anticipate that the fourth is fine as well since it's a different GPU than last time.

Do you save M23102129 and M22094129 10000iterations Res64 Dump ?

[msft]

Hi, henryzz

Originally Posted by henryzz

i have a few questions about mlucas-cuda on 64-bit windows 7
does it write checkpoint files often so that progress can be resumed?

When receive kill signal MacLucasFFTW make checkpointfile.

$ ./MacLucasFFTW 216091 &
[3] 8183
$ Iteration 10000 M( 216091 )C, 0x30247786758b8792, n = 2097152, MacLucasFFTW v8.1 Ballester
kill 8183
$
[3]+ Done ./MacLucasFFTW 216091
$ ls -l c216091
-rw-r--r-- 1 msft msft 16777240 2009-11-20 10:48 c216091
$ ./MacLucasFFTW c216091
caso 2
Iteration 20000 M( 216091 )C, 0x13e968bf40fda4d7, n = 2097152, MacLucasFFTW v8.1 Ballester

And -cN option Create a new checkpoint after every N iterations.

$ ./MacLucasFFTW -c10000 216091 &
[3] 8371
Iteration 10000 M( 216091 )C, 0x30247786758b8792, n = 2097152, MacLucasFFTW v8.1 Ballester
$ ls -l c216091
-rw-r--r-- 1 msft msft 16777240 2009-11-20 10:55 c216091
Iteration 20000 M( 216091 )C, 0x13e968bf40fda4d7, n = 2097152, MacLucasFFTW v8.1 Ballester
$ ls -l c216091
-rw-r--r-- 1 msft msft 16777240 2009-11-20 10:57 c216091

I hope TEST on Windows.

[frmky]

Originally Posted by msft

Hi, frmky

Do you save M23102129 and M22094129 10000iterations Res64 Dump ?

For M23102129, I think this is the 10000th iteration:
M( 23102129 )C, 0xbbbbecc86d34055f, n = 2097152, MacLucasFFTW v8.1 Ballester

I don't have the one for M22094129. Hopefully the second double check will come back soon and conclusively determine if the CUDA residue is correct.

[msft]

Hi, frmky

Originally Posted by frmky

Hopefully the second double check will come back soon and conclusively determine if the CUDA residue is correct.

Now I start M23102129 with GTX260 and M22094129 with Q8400.

[msft]

Hi,
Version "C" at .0121 sec/iter for the 2048K FFT and .0229 sec/iter for the 4096K FFT on GTX260.
Thank you,

[ET_]

Did anybody test the following with GPU/CUDA?

1 - A known Mersenne Prime
2 - A Mersenne number when CPU cores are busy with Prime95/mprime ?

Luigi

[lycorn]

Originally Posted by ET_

Did anybody test the following with GPU/CUDA?

1 - A known Mersenne Prime

Yes. M216091 and M24036583 have been successfully verified.

As for the 2nd part of your question, I really don´t know.

[msft]

Hi, ET_

Originally Posted by ET_

2 - A Mersenne number when CPU cores are busy with Prime95/mprime ?

Now testing M23102129 with GTX260(GPU/CUDA) and M22094129 with Q8400(Prime95), It is same machine ;-p

[bayanne]

I have a GTX 285 that I will be installing in my Mac Pro today.

I'm happy to run doublechecks, so which version of MacLucasFFTW is likely to give best results ....

[msft]

Hi, bayanne

Originally Posted by bayanne

I have a GTX 285 that I will be installing in my Mac Pro today.

I'm happy to run doublechecks, so which version of MacLucasFFTW is likely to give best results ....

you can download MacLucasFFTW.cuda.C.tar.gz,
and test performance no change version and delete -DTESRA from Makefile version.
Thank you,

[msft]

Originally Posted by frmky

M23102129

Hi,
My GTX260 result,

0x88f56f1c11c729__