News for hackers. Stuff that matters.
Re:48th Mersenne prime found by sp332 on Tuesday February 05, 2013 @10:17AM (#5171030)
This part confuses me: "a 32-core server in 6 days" and "i7 CPU in 4.5 days" to do the same verification? Is the MLucas code really that much slower than GIMPS, or did the verification only use one core on the 32-core machine or what?
Re:48th Mersenne prime found by Dubslow on Tuesday February 05, 2013 @12:10PM (#5171767)
It's a myriad of answers. The Mlucas run used a larger FFT than was necessary, partly due to multithreading reasons. The Mlucas code also only uses SSE2 instructions, where the overclocked i7-2600K using GIMPS' Prime95 program is fully AVX capable. I would also expect that the i7 runs at a higher frequency, with a higher IPC, than the server. The "main" CUDALucas run took around 90 hours on a GTX 580. We didn't even hear about the 560 Ti run by Gilchrist, that was on the side.

Here are some links: http://www.mersenneforum.org/showpost.php?p=325878&postc... http://www.mersenneforum.org/showthread.php?p=325951#post325... http://www.mersenneforum.org/showpost.php?p=325995&postc... http://www.mersenneforum.org/showpost.php?p=326020&postc...

Re:48th Mersenne prime found by Dubslow on Tuesday February 05, 2013 @12:13PM (#5171786)
One last clarification: the reason to use Mlucas, and not Prime95 since it's so much faster, is for independent-code verification of the prime. (CUDALucas also counts in that regard, but the more independent triple and quadruple checks, the merrier!)
Re:48th Mersenne prime found by mkl on Tuesday February 05, 2013 @11:03AM (#5171325)
Each of these verifications would have been using different code and algorithms, otherwise it's not really verification. So different speeds are entirely expected.
Re:48th Mersenne prime found by masklinn on Tuesday February 05, 2013 @10:24AM (#5171072)
Yes it seems very odd, especially as they also mention CUDALucas running in 3.7 and 7.7 days on two different NVidia GPUs (one a 560, the other one unnamed — maybe a Fermi board?)
Re:48th Mersenne prime found by sp332 on Tuesday February 05, 2013 @10:40AM (#5171181)
I have the 560 Ti mentioned in the article. It's a mid-to-high grade card, better than most $200 cards IMHO but if you're willing to spend the cash, it's not at all difficult to find a card with more horsepower. Dropping a cool kilodollar will net you a GTX 690 which seems to be the right relative speed. http://www.hwcompare.com/12683/geforce-gtx-560-ti-vs-geforce...

Edit: since Dubslow's reply, here's the comparison for the GTX 580: http://www.hwcompare.com/9133/geforce-gtx-560-ti-vs-geforce-... It's also about $400.

Re:48th Mersenne prime found by frozenport on Tuesday February 05, 2013 @02:25PM (#5172659)
Sounds like 2 AMD Opteron 6200 Interlagos chips as found on many clusters. Each has 16 cores, but they have only 8 floating point units and a smaller clock then their Intel counterparts.

For some codes each core gives 60% of an Intel core for many others it is 40%.

The Bulldozers and the Interlogos go down in my mind as the chip that destroyed the AMD. It late, was hard to use, and underperformed.

Re:48th Mersenne prime found by shocks on Tuesday February 05, 2013 @10:22AM (#5171063)
The article does not mention how powerful these 32 server cores are.