[QUOTE=ET_;218813]If no one replies I will open new CPU-Only ranges next weekend. :smile:
[/QUOTE]

Do you really think we should? Given the performance difference, wouldn't it be more responsible to warn people away from using CPUs as a waste of computer resources?

Then again, maybe it's moronic, in addition to oxymoronic, to urge efficiency in a whimsical project.

Does the GPU also deliver huge performance at current GIMPS levels? Perhaps somebody should be organizing a GPU-TF project for GIMPS.

William

[QUOTE=wblipp;218823]Do you really think we should? Given the performance difference, wouldn't it be more responsible to warn people away from using CPUs as a waste of computer resources?

Then again, maybe it's moronic, in addition to oxymoronic, to urge efficiency in a whimsical project.

Does the GPU also deliver huge performance at current GIMPS levels? Perhaps somebody should be organizing a GPU-TF project for GIMPS.

William[/QUOTE]

We are also trying GPUs on some 100mdpp exponents. The limit is well under GIMPS project, and AFAIK Uncwilly is managing the project.

Indeed, we could focus our GPU resources on GIMPS-LMH and leave OBD to people with older PCs.

I asked about extending our exponents' range because I had similar requests in the past. It seems that people likes OBD. :smile: Anyway I was thinking about a really small range (about 50 exponents, many of them already factored) just to play a little, to search for new factors.

Finally, "someone" thinks PC power should only be used for "interesting" factorizations. I think that people should use their PC power to do the search they like. Or even play games. I'm not recruiting GIMPSters, I'm just offering another opportunity to play with math and computing programs to newbies, developing my skills of co-managing a distributed project.

And offering a sort of gymnasium to those interested in GIMPS but with low constance.

I would like to hear from other readers, maybe you could create a poll asking for advice.

Luigi

[QUOTE=wblipp;218823]Do you really think we should? Given the performance difference, wouldn't it be more responsible to warn people away from using CPUs as a waste of computer resources?[/QUOTE]

Define "waste of computer resources". Aren't most distributed computing projects a waste of energy? The benefits from those projects are for Joe Average very low.


[QUOTE=wblipp;218823]Does the GPU also deliver huge performance at current GIMPS levels? Perhaps somebody should be organizing a GPU-TF project for GIMPS.[/QUOTE]

Yes, my code works fine at current GIMPS levels, too. It puts a little bit more pressure on the sieve (CPU) but the difference is not that big.
A few weeks ago I took all unfactored exponents in the range M50.00x.xxx to 2^71 (most where at 2^68 before) => 200 exponents, 14 factors found!
~16 days on a GTX 480 using the older "71bit kernel", with the new "75bit kernel" (which runs faster on Fermi) it would need ~11 days.

William: if your curious (and have some time) take a look at the thread "Trial division with CUDA" in the hardware forum.

Oliver

[CODE]
no factor for M3321928219 from 2^76 to 2^77 [mfaktc 0.08 95bit_mul32]
no factor for M3321928219 from 2^77 to 2^78 [mfaktc 0.08 95bit_mul32]
no factor for M3321928219 from 2^78 to 2^79 [mfaktc 0.08 95bit_mul32]
M3321928219 has a factor: 670475084138875461500081
found 1 factor(s) for M3321928219 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]
[/CODE]

Sorry Merfighters, you have to edit the mersennewiki again. This beats the old record by 3 bits! It is 2[SUP]2.775[/SUP] times bigger than the old record. :smile:

Oliver

[quote=TheJudger;219081][code]
no factor for M3321928219 from 2^76 to 2^77 [mfaktc 0.08 95bit_mul32]
no factor for M3321928219 from 2^77 to 2^78 [mfaktc 0.08 95bit_mul32]
no factor for M3321928219 from 2^78 to 2^79 [mfaktc 0.08 95bit_mul32]
M3321928219 has a factor: 670475084138875461500081
found 1 factor(s) for M3321928219 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]
[/code]

Sorry Merfighters, you have to edit the mersennewiki again. This beats the old record by 3 bits! It is 2[sup]2.775[/sup] times bigger than the old record. :smile:

Oliver[/quote]

That's a nice record!! :bow wave:
[URL]http://www.mersennewiki.org/index.php/Operation_Billion_Digits[/URL]

But why do you talking about me?? :innocent:

Nice factor judger! :curtisc:

79.15 bits.

[QUOTE=TheJudger;219081][CODE]
no factor for M3321928219 from 2^76 to 2^77 [mfaktc 0.08 95bit_mul32]
no factor for M3321928219 from 2^77 to 2^78 [mfaktc 0.08 95bit_mul32]
no factor for M3321928219 from 2^78 to 2^79 [mfaktc 0.08 95bit_mul32]
M3321928219 has a factor: 670475084138875461500081
found 1 factor(s) for M3321928219 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]
[/CODE]

Sorry Merfighters, you have to edit the mersennewiki again. This beats the old record by 3 bits! It is 2[SUP]2.775[/SUP] times bigger than the old record. :smile:

Oliver[/QUOTE]

:surprised:surprised:surprised

Hi,

[QUOTE=Merfighters;219092]But why do you talking about me?? :innocent:[/QUOTE]

Because you did the last few edits. :smile:

---
[CODE]
no factor for M3321928241 from 2^76 to 2^77 [mfaktc 0.08 95bit_mul32]
no factor for M3321928241 from 2^77 to 2^78 [mfaktc 0.08 95bit_mul32]
no factor for M3321928241 from 2^78 to 2^79 [mfaktc 0.08 95bit_mul32]
no factor for M3321928241 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]

no factor for M3321928307 from 2^76 to 2^77 [mfaktc 0.08 95bit_mul32]
no factor for M3321928307 from 2^77 to 2^78 [mfaktc 0.08 95bit_mul32]
no factor for M3321928307 from 2^78 to 2^79 [mfaktc 0.08 95bit_mul32]
no factor for M3321928307 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]
[/CODE]


Oliver

Calculating an average bit level respecting the fact that higher bit levels take more work to achieve, I calculate it to be 78.086654 bits.

In other words if the current crop of unfactored candidates were trial factored equally deep, they would now all be just beyond 78 bits.

[code]
no factor for M3321928171 from 2^79 to 2^80 [mfaktc 0.08-pre5 95bit_mul32]
no factor for M3321928171 from 2^80 to 2^81 [mfaktc 0.08-pre5 95bit_mul32]
[/code]

Luigi :smile:

[QUOTE]no factor for M3321928319 from 2^76 to 2^77 [mfaktc 0.08Winx64 95bit_mul32]
no factor for M3321928319 from 2^77 to 2^78 [mfaktc 0.08Winx64 95bit_mul32]
no factor for M3321928319 from 2^78 to 2^79 [mfaktc 0.08Winx64 95bit_mul32]
no factor for M3321928319 from 2^79 to 2^80 [mfaktc 0.08Winx64 95bit_mul32]
no factor for M3321928319 from 2^80 to 2^81 [mfaktc 0.08Winx64 95bit_mul32]
[/QUOTE]

Dave