mersenneforum.org > Data COMPLETE!!!! Thinking out loud about getting under 20M unfactored exponents
 Register FAQ Search Today's Posts Mark Forums Read

 2021-12-01, 05:36 #958 axn     Jun 2003 5·29·37 Posts Unlike previous versions which run into diminishing returns the more RAM you throw at it, 30.8 efficiency keeps on going up as you increase your RAM allocation. So any recommendation that doesn't take into account the available RAM is going to be sub-optimal (and not just by a little) We don't need to cut it so close to the required probabilities. Pick reasonably large B1, let P95 choose the B2. We'll still complete the range _much_ faster than we would with previous versions. If it turns out that we're getting much larger probabilities than is needed, we can dial back B1 a bit. Last fiddled with by axn on 2021-12-01 at 05:37 Reason: s/that/than
 2021-12-02, 00:35 #959 Prime95 P90 years forever!     Aug 2002 Yeehaw, FL 1ED016 Posts I had wanted to keep 30.8 out of the general public's eye, but someone created a 30.8 thread. I've moved all the 30.8 posts there: https://www.mersenneforum.org/showthread.php?t=27366 We now return this thread to its regular programming....
 2021-12-02, 05:07 #960 petrw1 1976 Toyota Corona years forever!     "Wayne" Nov 2006 Saskatchewan, Canada 3·11·157 Posts Some stats in the 2x.xM ranges. No crashes or weird results so far... i5-2500 with 12GB RAM allocated; B1/B2 of 600K/90M: 28.0M exponent Stage 1: 51 Min Stage 2: 55 Min i5-3570 with 14GB RAM; B1/B2 of 600K/90M: 24.7M exponent Stage 1: 36 Min Stage 2: 31 Min i5-3570 with 6.5GB RAM; B1/B2 of 600K/75M: 24.7M exponent Stage 1: 35 Min Stage 2: 51 Min i5-3570K with 3GB RAM; B1/B2 of 600K/75M: 24.7M exponent Stage 1: 31 Min Stage 2: 85 Min i5-3570 with 14GB RAM; B1/B2 of 600K/90M: 6.1M exponent Stage 1: 7 Min Stage 2: 3 Min i5-3570K with 3GB RAM; B1/B2 of 600K/75M: 6.1M exponent Stage 1: 6 Min Stage 2: 5 Min My take: in the 2x.xM ranges 12+ GB seems decent 6.5GB so-so 3GB not enough ... better suited for the lower exponents (i.e 6M) P.S. 2 cases now trying to stop Prime95 early in Stage 2; right after Stage 2 Init complete. And it takes a very long time ... more than 5 minutes and still waiting. ...it stopped only after Stage 2 completed. Last fiddled with by petrw1 on 2021-12-02 at 05:39 Reason: last line
2021-12-02, 06:05   #961
axn

Jun 2003

5×29×37 Posts

Quote:
 Originally Posted by petrw1 My take: in the 2x.xM ranges 12+ GB seems decent
Do you have machines with more RAM? Like 32 GB or more? You'll be surprised (or not) at how much more speedup could be gained by using more RAM. For 2x.x, I would think something like 100+ GB will be "optimal".

2021-12-02, 12:26   #962
alpertron

Aug 2002
Buenos Aires, Argentina

1,447 Posts

Quote:
 Originally Posted by petrw1 No crashes or weird results so far... ... i5-3570 with 14GB RAM; B1/B2 of 600K/90M: 6.1M exponent Stage 1: 7 Min Stage 2: 3 Min i5-3570K with 3GB RAM; B1/B2 of 600K/75M: 6.1M exponent Stage 1: 6 Min Stage 2: 5 Min
I'm running a more intensive test. The contents of the file worktodo.txt is:

Code:
[Worker #1]
Pminus1=1,2,9325159,-1,50000000,50000000000,"8056937377,16207126343,95788033249,799651034569,2894827758689,1416068886691577,36098431074477463,635315498297290933609,6214794856879965540791"
It took 26 hours for step 1 in an I5 3470. and now it is running step 2 for 35 hours with 8 GB used for Prime95. According to the values above, probably Prime95 will require 10 days to complete step 2.

2021-12-02, 13:02   #963
axn

Jun 2003

536510 Posts

Quote:
 Originally Posted by alpertron It took 26 hours for step 1 in an I5 3470. and now it is running step 2 for 35 hours with 8 GB used for Prime95. According to the values above, probably Prime95 will require 10 days to complete step 2.
You should have given it a _lot_ more RAM.

 2021-12-02, 13:19 #964 alpertron     Aug 2002 Buenos Aires, Argentina 1,447 Posts My system has 16 GB of RAM, and I have several programs running at the same time. That's why I give 8 GB to Prime95. Otherwise the system would run slower.
 2021-12-02, 13:42 #965 Xyzzy     Aug 2002 3×2,819 Posts We have 256GB of memory. If you want a test/benchmark done just post the worktodo entry that we need to use.
2021-12-02, 14:02   #966
petrw1
1976 Toyota Corona years forever!

"Wayne"
Nov 2006

3×11×157 Posts

Quote:
 Originally Posted by axn Do you have machines with more RAM? Like 32 GB or more? You'll be surprised (or not) at how much more speedup could be gained by using more RAM. For 2x.x, I would think something like 100+ GB will be "optimal".
Yes I have a 32GB machine that with 30.7 did P1 4x faster than these machines.
It's at the doctor. I'll have it later today.

2021-12-02, 16:51   #967
axn

Jun 2003

14F516 Posts

Quote:
 Originally Posted by petrw1 Yes I have a 32GB machine that with 30.7 did P1 4x faster than these machines. It's at the doctor. I'll have it later today.

Quote:
 Originally Posted by alpertron My system has 16 GB of RAM, and I have several programs running at the same time. That's why I give 8 GB to Prime95. Otherwise the system would run slower.
Yes, with only 16GB available RAM, I understand that you can't give more. Unfortunately, you could've saved several days of computation if you had more RAM.

2021-12-03, 19:51   #968
petrw1
1976 Toyota Corona years forever!

"Wayne"
Nov 2006

518110 Posts

Quote:
 Originally Posted by petrw1 i5-3570K with 3GB RAM; B1/B2 of 600K/75M: 24.7M exponent Stage 1: 31 Min Stage 2: 85 Min 3GB not enough ... better suited for the lower exponents (i.e 6M)
I squeezed this PC from 3GB to 5GB and the Stage 2 time dropped from 85 minutes to 50 minutes.

 Similar Threads Thread Thread Starter Forum Replies Last Post jschwar313 GPU to 72 3 2016-01-31 00:50 Batalov Factoring 6 2011-12-27 22:40 jasong jasong 1 2008-11-11 09:43 devarajkandadai Math 4 2007-07-25 03:01 WraithX GMP-ECM 1 2006-03-19 22:16

All times are UTC. The time now is 07:47.

Wed May 25 07:47:24 UTC 2022 up 41 days, 5:48, 0 users, load averages: 1.11, 1.36, 1.36