mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Software

Reply
 
Thread Tools
Old 2020-08-03, 11:27   #562
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

32·337 Posts
Default

Quote:
Originally Posted by preda View Post
P-1 bounds getting lower too. A pity, I liked P-1 :)
(I always thought P-1 is so much more beautiful than TF :), but TF is done early to high-bits and too little opportunity is left for P-1)
GPU-TF threw off the balance of TF-PM1. In the old days (pre-GPU) there was alternation with low-TF, then P-1, then high-TF, then LL. GPU-TF changed the equation by making TF much cheaper. It might be useful to reexamine the methodology with the current set of GPU-TF, CPU/GPU-PM1, CPU/GPU-PRP+Cert software and see if the current order and limits of factoring methods is still optimal.
James Heinrich is offline   Reply With Quote
Old 2020-08-03, 19:02   #563
Aramis Wyler
 
Aramis Wyler's Avatar
 
"Bill Staffen"
Jan 2013
Pittsburgh, PA, USA

6048 Posts
Default

I'm working from very old and sluggish memories here, but the problem with running p-1 on GPUs was having enough memory to set useful bounds, right? That's why we do the gpu-tf; more discrete sizes?
Aramis Wyler is offline   Reply With Quote
Old 2020-08-03, 23:27   #564
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

22×1,109 Posts
Default

Quote:
Originally Posted by Aramis Wyler View Post
I'm working from very old and sluggish memories here, but the problem with running p-1 on GPUs was having enough memory to set useful bounds, right? That's why we do the gpu-tf; more discrete sizes?
CUDAPm1 was rather limiting. I've run full GPUto72 labeled mersenne.ca bounds P-1 up to 999M exponents on 16-GB Tesla P100 GPUs in gpuowl on Colab. 999999937, and PrimeNet labeled bounds to 500M on an 8GB RX480.
GPUs go higher on TF because they are much faster at the single precision computations used in TF relative to DP-type computations used in primality testing or P-1; ratios ~11 to 40 or so are common. CPUs' speed ratios are much smaller, typically 0.7 to 1.5.
kriesel is offline   Reply With Quote
Old 2020-08-06, 13:46   #565
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

29·53 Posts
Default

Quote:
Originally Posted by preda View Post
P-1 bounds getting lower too. A pity, I liked P-1 :)

(I always thought P-1 is so much more beautiful than TF :), but TF is done early to high-bits and too little opportunity is left for P-1)
P-1 bounds. Really? I posted a topic about this in the gpuOwl forum group two days ago. It has pretty much been ignored. I pointed out the diversity of what the different programs suggest for bounds on the same test. I suppose there is no such thing as going too high. Too low would be different. Doing so may require a repeated test with the bounds higher. I do not want to be responsible for that.

I ran a lot of P-1's in the pre-GPU days despite not being able to make a direct connection between P-1 bounds and TF start and stop bits.
storm5510 is offline   Reply With Quote
Old 2020-08-06, 13:54   #566
kruoli
 
kruoli's Avatar
 
"Oliver"
Sep 2017
Porta Westfalica, DE

307 Posts
Default

I guess in this case, your topic was "ignored" because no one had a good answer. Especially, when P-1 are going to be reevaluated now.
kruoli is offline   Reply With Quote
Old 2020-08-06, 15:34   #567
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

30018 Posts
Default

Quote:
Originally Posted by kruoli View Post
I guess in this case, your topic was "ignored" because no one had a good answer. Especially, when P-1 are going to be reevaluated now.
I understand what you mean by reevaluated, in this case. Their necessity, now that GPU's are becoming more powerful all the time. Recently, I have seen P-1 tests which indicate the exponent was trial-factored to 78 bits. 80, and beyond, TF's will not be all that far behind for those who can run them in what they feel is a practical period of time.
storm5510 is offline   Reply With Quote
Old 2020-08-06, 15:42   #568
kruoli
 
kruoli's Avatar
 
"Oliver"
Sep 2017
Porta Westfalica, DE

307 Posts
Default

Quote:
Originally Posted by kruoli View Post
...when P-1 bounds are going to be reevaluated now.
Oops, I was missing a word there. Didn't catch it fast enough.
kruoli is offline   Reply With Quote
Old 2020-08-06, 16:17   #569
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

32×337 Posts
Default

Quote:
Originally Posted by storm5510 View Post
I understand what you mean by reevaluated, in this case.
Their necessity, now that GPU's are becoming more powerful all the time.
I don't think that's what kruoli meant, but rather the the calculus of optimal P-1 bounds (and TF limits) will change with the new PRP+Cert worktype that should, eventually replace LL+DC (or equivalent) with a single test and small verification, thereby eliminating (approximately) half the effort, therefore the effort applied to TF and P-1 should be approximately half what it currently is (in broad terms).
James Heinrich is offline   Reply With Quote
Old 2020-08-06, 23:47   #570
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

29·53 Posts
Default

Quote:
Originally Posted by kruoli
Oops, I was missing a word there. Didn't catch it fast enough.
Do not feel bad, this is a daily occurrence for me. James made one below. He missed it. I do that one too.

Quote:
Originally Posted by James Heinrich
I don't think that's what kruoli meant, but rather the the calculus of optimal P-1 bounds (and TF limits) will change with the new PRP+Cert worktype that should, eventually replace LL+DC (or equivalent) with a single test and small verification, thereby eliminating (approximately) half the effort, therefore the effort applied to TF and P-1 should be approximately half what it currently is (in broad terms).
Half the effort from TF and P-1 suggests finishing at lower end-bits and lower bounds. That would speed things up. Mid 70's TF levels have jumped out about 10-million beyond the wavefront. chatsall wrote something not long ago about increasing to 77 bits for TF regarding GPUto72. In any case, things are changing.
storm5510 is offline   Reply With Quote
Old 2020-08-07, 00:37   #571
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

22×5×431 Posts
Default

Remember that each bit level of TF is twice the effort of the one before it. So, 1 bit level less. And that leaves more factors in the low end of the P-1 search to be found.
We live in interesting times.
Uncwilly is offline   Reply With Quote
Old 2020-08-07, 14:55   #572
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

30018 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
We live in interesting times.

That we do. Any idea when George might let Prime95 v30 out of its cage?
storm5510 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Prime95 version 29.2 Prime95 Software 71 2017-09-16 16:55
Prime95 version 29.1 Prime95 Software 95 2017-08-22 22:46
Prime95 version 26.5 Prime95 Software 175 2011-04-04 22:35
Prime95 version 25.9 Prime95 Software 143 2010-01-05 22:53
Prime95 version 25.8 Prime95 Software 159 2009-09-21 16:30

All times are UTC. The time now is 08:12.

Sun Sep 27 08:12:53 UTC 2020 up 17 days, 5:23, 0 users, load averages: 1.18, 1.35, 1.35

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.