Gimps

[glennpat]

GIMPS (http://www.mersenne.org/) was added to the Vault today and we are currently in place 126 ot 128.

When I first started PRIME95 I let it pick what it wanted to send an it looks like the WU I got will take a few days to complete. I now have "type of work" set to "Trail factoring to low limits" to get short work units. I set this in the PRIME95 application. Its under Test - Work Windows, but when you 1st run it I believe it was also in a window you get when it's 1st run. You can also set defaults in your account but the descriptions are different.

[glennpat]

GIMPS (http://www.mersenne.org/) was added to the Vault today and we are currently in place 126 ot 128.

When I first started PRIME95 I let it pick what it wanted to send an it looks like the WU I got will take a few days to complete. I now have "type of work" set to "Trail factoring to low limits" to get short work units. I set this in the PRIME95 application. Its under Test - Work Windows, but when you 1st run it I believe it was also in a window you get when it's 1st run. You can also set defaults in your account but the descriptions are different.

I am changing my type of work to just use "trial factoring". It looks like that takes about 25 minutes per work unit.

[Paladin]

Yeah, the two sets of descriptions are confusing since on the web account there are at least two "trial" options. I took a best guess there and then also set it in the application. It seems to be taking digestible sized WU (about an hour each) on my Q6600.

Onward & upward! :up:

[[XC] riptide]

GIMPS... they need to change the name. :D:D Also its interesting that tons of folks around here run Prime 95 a lot. Little do thay know that they could be using their stress testing efforts for GIMPS.

[Paladin]

I didn't even know there was a Win64 client until this project got Vaulted. Don't know if this project favors 64 or 32 bit, Win or Linux.

[glennpat]

I didn't even know there was a Win64 client until this project got Vaulted. Don't know if this project favors 64 or 32 bit, Win or Linux.

I just ran the benchmark prime95/mprime on Win64 versus Linux 64 and the numbers were about the same. For points on GIMPS it looks like it's just based on the Giga Hertz the processor is clocked at. Where they have data for the CPU they will have something like:


3.049 GHz (5.224 GHz P4 effective equivalent)

The above was from a Q6600. From the results I have seen it seems the GHz awarded are not based on the P4 equivalent but just the raw clock number.

[Paladin]

By raw clock #, does that mean a Prescott @3.2ghz would score the same as an i7-920 @3.2ghz per core? Not that it matters much since the i7 would finish each WU in a faction of the time.

[glennpat]

By raw clock #, does that mean a Prescott @3.2ghz would score the same as an i7-920 @3.2ghz per core? Not that it matters much since the i7 would finish each WU in a faction of the time.

That is what I think, but I have not read anything on how they do it. I have the following "2.394 GHz (4.163 GHz P4 effective equivalent)" on this for about 3.3 days and have gotten 7.561 GHz-days for it.

Not sure if you get anything extra for the work units that take 3 weeks or so to complete. I have a couple of work units that will complete in 10 days and I will put the results in this tread what I get. They are of the work type "D" defined as "LL double-check". I also have a couple of type "LL" defined as "LL first test". Those take longer. Maybe I'll find a giant prime :) with those.

[[XC] riptide]

By raw clock #, does that mean a Prescott @3.2ghz would score the same as an i7-920 @3.2ghz per core? Not that it matters much since the i7 would finish each WU in a faction of the time.

That would be a flaw. Take for example a 2Ghz p4 vs a 4Ghz P4. The 4ghz does x2 the work units... and going by whats been said, ALSO score x2 from the weighted point per WU on the 4Ghz. Thats x2x2 = x4 times the points. :shrug:

[glennpat]

riptide;3662683']That would be a flaw. Take for example a 2Ghz p4 vs a 4Ghz P4. The 4ghz does x2 the work units... and going by whats been said, ALSO score x2 from the weighted point per WU on the 4Ghz. Thats x2x2 = x4 times the points. :shrug:

Forget some of the stuff I said before. They calculate the Ghz awarded for WU by a formula based on parameters of the work unit. There is a thread here that talks about it: //http://www.mersenneforum.org/showthread.php?t=10937 (http://www.mersenneforum.org/showthread.php?t=10937)

The last entry in the thread has a link to a caculator. As an example I put in for a type TF for M61945931 factoring from 2^65 to 2^66. It came out with an answer of 0.120633 GHz-days. That is also the credit given to me in my log for the work unit.

For the one that takes 3 weeks: Type LL for M38622823 it calculates 50.853 Ghz-days.

So there is nothing in the formula that is based cpu clock speed.

[[XC] riptide]

Alright. It sounds equitable then.

[Paladin]

Has anyone found a site that shows Bok/Dyyryath style stats? (FreeDC stats don't include GIMPS)
GIMPS' site only shows a team summary for the last year and ARS only shows their own team.

[[XC] riptide]

I had a quick look but didn't find anything.

[glennpat]

:welcome:Our team is growing on here:welcome:

We are currently at 4,626 Vault points so we have a ways to go, but we are climbing. Free-DC has 76 points and as a result we passed them a day or so ago in the vault. Won't last for long, but we're in 6th place in the vault for now :clap:

[[XC] riptide]

Ya. I saw that. Plenty of kills ahead.

OT@glennpat: Your quads need to be clocked more. I see Q6600's at ~3Ghz. Not acceptable. :stick:

[glennpat]

riptide;3665016']Ya. I saw that. Plenty of kills ahead.

OT@glennpat: Your quads need to be clocked more. I see Q6600's at ~3Ghz. Not acceptable. :stick:

Yep it's a little sad. I was reminded it was a little (maybe lot) low when I saw the nice clock on Paladins. Then of course you are at 4.3Ghz on a E8500. I was trying to keep the wattage down some plus I was buying cheaper motherboards with onboard video, but it could be my skills:).

[[XC] riptide]

Although it is a POS and it takes 1.59vcore to keep it there. Its a C0 stepping. Early adopters get PWNED. :rolleyes:

Good news is, I plan on going i7 after March. :)

[UCmajewski]

I am on board with one machine that I can muster !

[Paladin]

New features in Version 25.9 of prime95.exe
-------------------------------------------

1) Performance increases for the 64-bit version by taking advantage of the
eight extra SSE2 registers. The 32-bit version may also be a bit faster
by taking advantage of the Core 2 architecture. The FFT code was
originally optimized for the Pentium 4 where instructions like
"movapd reg,reg" take 6 clocks and should be avoided, whereas on Core 2
the same instruction uses just 1 clock cycle and improves scheduling.
The 32-bit version may be slightly faster or slower on the Pentium 4, but
not enough to worry about. Preliminary benchmarks indicate:
32-bit Pentium 4: between 2% slower and 1% faster
32-bit Core 2: between 2% and 6% faster
64-bit Pentium 4: between 2% and 4% faster
64-bit Core 2: between 10% and 13% faster
64-bit Core i7: between 7% and 12% faster
64-bit Phenom: between 10% and 13% faster
AMD64: between 1% slower and 1% faster
2) Intermediate save files were overhauled. Trial factoring, P-1, and
ECM now support multiple save files. The backup save file uses a .bu
extension. While a new save file is being written a .write extension
is used and renamed when successfully written (thus, briefly there are
3 save files). NOTE #1: LL save files no longer use an initial letter
of q and r. The program will automatically use these obsolete files if
necessary and delete them when no longer needed. NOTE #2: ECM and P-1
save files on k*2^n+1 no longer use file names beginning with d and l.
The -An command line argument no longer appends an extension to save
file names. NOTE #3: The client now supports keeping three save files.
3) Miscellaneous minor bug fixes.

[[XC] riptide]

Thanks Paladin. I'm going to cross post this in the Software section.

[vaio]

Shame there isn't a decent stats page for this project.