mersenneforum.org  

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

Reply
 
Thread Tools
Old 2021-05-26, 16:26   #23
drkirkby
 
"David Kirkby"
Jan 2021
Althorne, Essex, UK

1110010102 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
Aren't you able to get around 380 GHz-days/day on that?
If I had that at my disposal, I would be overjoyed.
Unfortunately I believe you are thinking of another graphics card. I posted a thread here

https://mersenneforum.org/showthread.php?t=26790

on a PRP test benchmark of the exponent 103750501. Rounding to the nearest hour, a single Intel Xeon 8167M CPU would have completed the exponent in 45 hours (1.88 days), and the Nvidia Quadro P2200 in 295 hours (12.3 days). Given the exponent

https://www.mersenne.ca/exponent/103750501

is 413 GHz days, I assume that works out at 413/12.3=33.6 GHz days per day for the Nvidia Quadro P2200 GPU, and 413/45= 220 GHz days /day on the Intel Xeon 8167M CPU, making the CPU 6.5x faster than the GPU.

It so happens that the Nvidia Quadro P2200 GPU cost me £300 ($424 USD), and the CPUs cost £300 each. In terms of cost, the GPU is beaten by a factor of 6.5.

The information available on the Intel Xeon 8167M is sketchy, as Intel will not release any information. But this website
https://www.cpu-world.com/CPUs/Xeon/...n%208167M.html
says the TDP is 165 W. The TDP on the Nvidia Quadro P2200 is 75 W, so in terms of energy consumed per exponent, the CPU wins by a factor of 6.5*75/165=3.0
Using mfaktc for trial-factoring, the GPU is more comparable in performance to the CPU, but I am not interested in trial-factoring.

So I think I have a pretty poor GPU, but I bought it for a bit of very expensive engineering software that supported the Quadro, but not the mainstream gaming graphics cards.

Dave

Last fiddled with by drkirkby on 2021-05-26 at 16:28
drkirkby is offline   Reply With Quote
Old 2021-05-26, 17:02   #24
tuckerkao
 
"Tucker Kao"
Jan 2020
Head Base M168202123

37016 Posts
Default

Quote:
Originally Posted by axn View Post
380 Gd/d of TF is not that impressive. Even a lowly 1650 would get you ~1000. On gpuowl, I think it gets something like 40 Gd/d?
My old video card runs @ 340 Gd/d and it was from year 2013.

I'd like to have the stats of Geforce RTX 3070 Ti, 3080 Ti and 3090.
tuckerkao is offline   Reply With Quote
Old 2021-05-26, 17:20   #25
drkirkby
 
"David Kirkby"
Jan 2021
Althorne, Essex, UK

2×229 Posts
Default

Quote:
Originally Posted by tuckerkao View Post
My old video card runs @ 340 Gd/d and it was from year 2013.

I'd like to have the stats of Geforce RTX 3070 Ti, 3080 Ti and 3090.
What model is your old graphics card? With what software does it achieve 340 GHz day/day? Doing what type of work? Unless you give more information, the data is meaningless.

Somewhere on the forum there are benchmark data of GPU performance using both gnuowl & mfaktc, but I an unaware of any such data for CPUs with mprime.

Last fiddled with by drkirkby on 2021-05-26 at 17:51 Reason: Spelling error
drkirkby is offline   Reply With Quote
Old 2021-05-26, 17:45   #26
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

24·3·163 Posts
Default

Quote:
Originally Posted by drkirkby View Post
Somewhere on the forum there are benchmark data of GPU performance for gpuowl & mfaktc, but I an unaware of any such data for CPUs with mprime.
gpu ratings for cudalucas/cllucas & mfaktc can easily be found at mersenne.ca also. Which used to have cpu prime95 as well, but that "temporarily unavailable" has been up for months now: https://www.mersenne.ca/throughput.php
Bookmark https://www.mersenne.ca/. It has a dropdown menu in the upper right, a link to the GIMPS software mirror, etc, a very useful site.
kriesel is online now   Reply With Quote
Old 2021-05-26, 18:38   #27
tuckerkao
 
"Tucker Kao"
Jan 2020
Head Base M168202123

24×5×11 Posts
Default

Quote:
Originally Posted by drkirkby View Post
What model is your old graphics card? With what software does it achieve 340 GHz day/day? Doing what type of work? Unless you give more information, the data is meaningless.

Somewhere on the forum there are benchmark data of GPU performance using both gnuowl & mfaktc, but I an unaware of any such data for CPUs with mprime.
Nvidia Geforce GTX 780

It's @ 355 Gd/d for mfaktc when trial factoring between 2^73 to 2^76 and 325 Gd/d when trial factoring between 2^76 to 2^78.
tuckerkao is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
2020 the year in space prediction thread. Uncwilly Science & Technology 14 2021-05-13 07:17
General Hardware Discussion veganjoy jvang 21 2020-10-31 14:20
Do-it-yourself, crank, mersenne prediction thread. Uncwilly Miscellaneous Math 85 2017-12-10 16:03
2045 eclipse/future prediction thread MooMoo2 Lounge 2 2017-08-26 23:39
Old Hardware Thread E_tron Hardware 0 2004-06-18 03:32

All times are UTC. The time now is 16:39.


Fri Jul 7 16:39:36 UTC 2023 up 323 days, 14:08, 1 user, load averages: 3.59, 3.01, 2.39

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2023, 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.

≠ ± ∓ ÷ × · − √ ‰ ⊗ ⊕ ⊖ ⊘ ⊙ ≤ ≥ ≦ ≧ ≨ ≩ ≺ ≻ ≼ ≽ ⊏ ⊐ ⊑ ⊒ ² ³ °
∠ ∟ ° ≅ ~ ‖ ⟂ ⫛
≡ ≜ ≈ ∝ ∞ ≪ ≫ ⌊⌋ ⌈⌉ ∘ ∏ ∐ ∑ ∧ ∨ ∩ ∪ ⨀ ⊕ ⊗ 𝖕 𝖖 𝖗 ⊲ ⊳
∅ ∖ ∁ ↦ ↣ ∩ ∪ ⊆ ⊂ ⊄ ⊊ ⊇ ⊃ ⊅ ⊋ ⊖ ∈ ∉ ∋ ∌ ℕ ℤ ℚ ℝ ℂ ℵ ℶ ℷ ℸ 𝓟
¬ ∨ ∧ ⊕ → ← ⇒ ⇐ ⇔ ∀ ∃ ∄ ∴ ∵ ⊤ ⊥ ⊢ ⊨ ⫤ ⊣ … ⋯ ⋮ ⋰ ⋱
∫ ∬ ∭ ∮ ∯ ∰ ∇ ∆ δ ∂ ℱ ℒ ℓ
𝛢𝛼 𝛣𝛽 𝛤𝛾 𝛥𝛿 𝛦𝜀𝜖 𝛧𝜁 𝛨𝜂 𝛩𝜃𝜗 𝛪𝜄 𝛫𝜅 𝛬𝜆 𝛭𝜇 𝛮𝜈 𝛯𝜉 𝛰𝜊 𝛱𝜋 𝛲𝜌 𝛴𝜎𝜍 𝛵𝜏 𝛶𝜐 𝛷𝜙𝜑 𝛸𝜒 𝛹𝜓 𝛺𝜔