mersenneforum.org  

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

Reply
 
Thread Tools
Old 2021-05-05, 09:27   #12
drkirkby
 
"David Kirkby"
Jan 2021
Althorne, Essex, UK

3×149 Posts
Default

I increased the P-1 memory resources up to 170 GB, and requested a P-1 test as a manual assignment

PFactor=xxxxxxxxxxxx,1,2,103747177,-1,76,2
Code:
[Worker #1 May 5 10:20] M103747177 stage 1 is 9.05% complete. Time: 203.402 sec.
[Worker #1 May 5 10:20] M103747177 stage 1 is 9.05% complete. Time: 203.402 sec.
[Worker #1 May 5 10:20] M103747177 stage 1 is 9.05% complete. Time: 203.402 sec.
The manual assignment is now running, but using 0.1% of my RAM! So despite allocating mprime a load of RAM for P-1, it is not actually using it. At least not at this point in time - I only started a few minutes ago.

What does the 203.402 ms means?

Dave

Last fiddled with by drkirkby on 2021-05-05 at 09:28
drkirkby is offline   Reply With Quote
Old 2021-05-05, 09:36   #13
kruoli
 
kruoli's Avatar
 
"Oliver"
Sep 2017
Porta Westfalica, DE

571 Posts
Default

Stage 1 of P-1 actually needs very little memory. Stage 2 is where the RAM usage will be much more.

203.402 seconds (!) is the time it took since the last status update. That way you can estimate the ETA of stage 1. I'm only wondering why the message is there three times; that should not be the case.

P-1 can be quite satisfying: Having "success" is much more common than when doing PRP/LL, i.e. finding a factor and maybe even finding a top 100 factor.

Last fiddled with by kruoli on 2021-05-05 at 09:37 Reason: Grammar.
kruoli is online now   Reply With Quote
Old 2021-05-29, 18:01   #14
tuckerkao
 
"Tucker Kao"
Jan 2020
Head Base M168202123

22×107 Posts
Default

Quote:
Originally Posted by pepi37 View Post
I have 12 GTX/RTX cards. I do mining. ASIC is better option, but it is limited to algo, and cannot be used on anything else. GPU can do many thing, and can also be sold years after with decent price. Current ASIC in two years from will be useless ( or near that point)
Current GPU , that is different story
I'm really glad Nvidia has imposed the "Hash Rate Lite" on all the video cards except Geforce 3090 now. Since AMD has no issues with the crypto miners, all the mathematicians ought to be excited for the upcoming RDNA 3 by the beginning of year 2023.
tuckerkao is offline   Reply With Quote
Old 2021-05-29, 19:13   #15
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

493810 Posts
Default

Why do you think AMD has no problems with miners? If an AMD card comes along that is very useful for we mathematicians, why wouldn't it also be very useful for mining?
VBCurtis is offline   Reply With Quote
Old 2021-05-29, 19:52   #16
drkirkby
 
"David Kirkby"
Jan 2021
Althorne, Essex, UK

1101111112 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
Why do you think AMD has no problems with miners? If an AMD card comes along that is very useful for we mathematicians, why wouldn't it also be very useful for mining?
See
https://blogs.nvidia.com/blog/2021/02/18/geforce-cmp/
where Nvidia state the GeForce RTX 3060 software drivers are designed to detect specific attributes of the Ethereum cryptocurrency mining algorithm, and limit the hash rate, or cryptocurrency mining efficiency, by around 50 percent. They want gamers to have the card - not crypto miners. Where would prime hunters fall? It's not gaming, but neither is it mining.

It might be a game of cat-and-mouse, with a challenge to miner developers to use the card, without it being detected as mining.

Last fiddled with by drkirkby on 2021-05-29 at 19:55
drkirkby is offline   Reply With Quote
Old 2021-05-29, 20:25   #17
drkirkby
 
"David Kirkby"
Jan 2021
Althorne, Essex, UK

3×149 Posts
Default

Quote:
Originally Posted by tuckerkao View Post
I hope that the DDR5 Memory modules will be available for the compatible motherboards and CPUs by mid year 2022. If the total memory capacities reaches 512 gigabytes per machine, the P-1 factoring should be under 2 hours for most exponents under 200M.
To give just one data point, for an exponent of 108,792,767 the P-1 factoring is only using 321 GB RAM despite I had given it access to 370 GB. I don't know how that would scale to exponent of 200 M, but it would not totally surprise me if more than 512 GB could be used.
Code:
[Worker #2 May 29 20:18] M108792767 stage 1 complete. 2678570 transforms. Time: 2019.648 sec.
[Worker #2 May 29 20:18] Starting stage 1 GCD - please be patient.
[Worker #2 May 29 20:19] Stage 1 GCD complete. Time: 47.641 sec.
[Worker #2 May 29 20:19] Available memory is 376728MB.
[Worker #2 May 29 20:19] D: 2730, relative primes: 7192, stage 2 primes: 3211863, pair%=95.71
[Worker #2 May 29 20:19] Using 328771MB of memory.
[Worker #1 May 29 20:19] Iteration: 76800000 / 103866121 [73.94%], ms/iter:  1.492, ETA: 11:13:15
[Worker #1 May 29 20:22] Iteration: 76900000 / 103866121 [74.03%], ms/iter:  1.488, ETA: 11:08:52
[Worker #1 May 29 20:24] Iteration: 77000000 / 103866121 [74.13%], ms/iter:  1.560, ETA: 11:38:31
[Worker #1 May 29 20:24] Gerbicz error check passed at iteration 77000000.
[Worker #2 May 29 20:26] Stage 2 init complete. 75668 transforms. Time: 425.106 sec.
Be aware that server grade CPUs will most likely require ECC RAM, and that is significantly more expensive than RAM without error correction. Dell want just over £500 ($709 USD) for a single 32 GB stick of DDR4 ECC RAM for this Dell 7920 tower workstation. Even third party (Kingston) RAM is about £180 for 32 GB from a reseller. A distributor is a bit cheaper, but they can't currently get any 32 GB RDIMMs from Kingston.

Dave

Last fiddled with by drkirkby on 2021-05-29 at 20:42
drkirkby is offline   Reply With Quote
Old 2021-05-29, 21:46   #18
tuckerkao
 
"Tucker Kao"
Jan 2020
Head Base M168202123

22·107 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
Why do you think AMD has no problems with miners? If an AMD card comes along that is very useful for we mathematicians, why wouldn't it also be very useful for mining?
AMD seems to be willing to grant more freedom to its buyers because it wants more buyers and markets anyway. The current Radeon 6900 XT seems to be a bit slower than the very top end, I expect the next generation 7900 XT will fulfill all the expectations including the mining.

Quote:
Originally Posted by drkirkby View Post
[Worker #2 May 29 20:26] Stage 2 init complete. 75668 transforms. Time: 425.106 sec.
[/CODE]Be aware that server grade CPUs will most likely require ECC RAM, and that is significantly more expensive than RAM without error correction. Dell want just over £500 ($709 USD) for a single 32 GB stick of DDR4 ECC RAM for this Dell 7920 tower workstation. Even third party (Kingston) RAM is about £180 for 32 GB from a reseller. A distributor is a bit cheaper, but they can't currently get any 32 GB RDIMMs from Kingston.

Dave
I hope I can get the DDR5 memories by the time I purchase my next PC if possible. Nvidia Geforce RTX 4090 will be available at that time.

Quote:
Originally Posted by Uncwilly View Post
This is correct.
Once we have tested those below that range, we will test them too. We want to make sure that we know the M54 is M54 and not M55. But it makes no sense to test them until we get to that range. The machines that we will have then will be better for testing those numbers.
AMD Threadripper 5970X should be able to finish 1 PRP test of M168,xxx,xxx every 4 and a half day, so good enough to me.

Last fiddled with by tuckerkao on 2021-05-29 at 22:11
tuckerkao is offline   Reply With Quote
Old 2021-05-29, 22:11   #19
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

11000011100012 Posts
Default

Quote:
Originally Posted by tuckerkao View Post
... willing to grant more freedom to its buyers ...
How did we ever let this situation come to pass. The seller dictating what the buyer is permitted to do.

Boycott. Stop buying encumbered things.
retina is online now   Reply With Quote
Old 2021-07-21, 09:12   #20
tuckerkao
 
"Tucker Kao"
Jan 2020
Head Base M168202123

6548 Posts
Default

I hope Intel Core-i9 12900k can be released as soon as possible - Nov 2021 the earlier possible, so at least a competitive rival stands next to AMD Threadripper 5970X, thus the price won't be too expensive for both.
tuckerkao is offline   Reply With Quote
Old 2021-07-21, 11:58   #21
M344587487
 
M344587487's Avatar
 
"Composite as Heck"
Oct 2017

7×112 Posts
Default

Quote:
Originally Posted by tuckerkao View Post
I hope Intel Core-i9 12900k can be released as soon as possible - Nov 2021 the earlier possible, so at least a competitive rival stands next to AMD Threadripper 5970X, thus the price won't be too expensive for both.
The leaked Cinebench score seems too good to be true for the 12900k, at least my expectation of the LITTLE in big.LITTLE is that it would not be so performant. I guess removing AVX512 and SMT does make quite a large difference to the die size of the LITTLE. In any case, unless DDR5 speeds ramp very quickly the 12900k (dual channel ddr5?) should sit somewhere between the 5950X (dual channel DDR4) and 5970X (quad channel DDR4) for Prime95. Another question mark is if AVX512 and all 16 cores can be used simultaneously, it's my understanding that windows at least in its current state cannot and I don't know about Linux. Maybe pinning to the rescue?
M344587487 is offline   Reply With Quote
Old 2021-07-24, 23:04   #22
ewmayer
2ω=0
 
ewmayer's Avatar
 
Sep 2002
República de California

11,657 Posts
Default

@drkirby:

I believe once you get beyond 500-1000 memory buffers in stage 2, you get rapidly diminishing returns, as the stage 2 prime-pairing percentage approaches 100%. Do your mprime logs list #bufs and pairing %? Suggest you compare the for several different "how much mem" settings.
ewmayer is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Threadripper 3990X vs. Threadripper 3970X Viliam Furik Hardware 21 2020-04-19 08:24
Updated Bios on B350M Board for Ryzen Series 3000! megav13 Hardware 1 2019-09-03 15:10
Has anyone tried linear algebra on a Threadripper yet? fivemack Hardware 3 2017-10-03 03:11
5000 < k < 6000 justinsane Riesel Prime Data Collecting (k*2^n-1) 26 2010-12-31 12:27
6000 < k < 7000 amphoria Riesel Prime Data Collecting (k*2^n-1) 0 2009-04-12 16:58

All times are UTC. The time now is 10:23.


Thu Sep 16 10:23:51 UTC 2021 up 55 days, 4:52, 0 users, load averages: 1.59, 1.69, 1.58

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, 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.