mersenneforum.org  

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

Reply
 
Thread Tools
Old 2016-03-16, 19:06   #1
Rodrigo
 
Rodrigo's Avatar
 
Jun 2010
Pennsylvania

11100110112 Posts
Default Running CUDA on non-Nvidia GPUs

I just came across the following and thought that it might be of interest: http://www.extremetech.com/computing...on-nvidia-gpus

Of course, I wouldn't be surprised to learn that it's old news to people here.

If one could run MFAKTC on an AMD card, it might be interesting to learn whether it would run faster, slower, or about as fast as MFAKTO on the same card.

Rodrigo
Rodrigo is offline   Reply With Quote
Old 2016-03-17, 22:33   #2
TheJudger
 
TheJudger's Avatar
 
"Oliver"
Mar 2005
Germany

110410 Posts
Default

my bet: best case it runs much slower than mfakto.
TheJudger is offline   Reply With Quote
Old 2016-05-10, 16:15   #3
diep
 
diep's Avatar
 
Sep 2006
The Netherlands

67510 Posts
Default

Rodrigo - architecture of AMD cards versus Nvidia cards is different.

At Nvidia the 32x32 == 64 bits (requiring 2 instructions) is very fast starting at Fermi compared to AMD which historically is very fast for 24x24 bits == 24 bits in unsigned integers.

At AMD you run their OpenCL stuff (avoiding the word crap).

There is a special mfockt version that runs on AMD. Forgot exact name will be 1 letter difference from the nvidia thing.

So it's a total different implementation doing exactly the same thing as other instructions in other combinations is a lot faster at AMD than what is faster at Nvidia.

edit: if you wonder about timing details, at nvidia a 32x32 bits multiplication that gives the high bits requires at AMD 4 units to produce the result (6000+ and 7000+ generation). You could see that arguably as 4 times slower (of course the AMD has more units let's not forget about that).

Newer Nvidia's are doing something in between the old Fermi and what AMD is doing for 64 bits double precision. Not clear to me how they solve 32 bits multiplications nowadays.

Last fiddled with by diep on 2016-05-10 at 16:19
diep is offline   Reply With Quote
Old 2016-05-17, 05:43   #4
bayanne
 
bayanne's Avatar
 
"Tony Gott"
Aug 2002
Yell, Shetland, UK

307 Posts
Default

Has anyone compiled mfakto for use on a Mac?
bayanne is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
ECM for CUDA GPUs in latest GMP-ECM ? Karl M Johnson GMP-ECM 493 2020-04-19 13:59
ubuntu-15.10 nvidia-355 driver fails to detect two dissimilar GPUs fivemack GPU Computing 9 2015-11-18 12:55
Cuda-Z for nvidia and radeon ET_ GPU Computing 0 2013-11-20 11:38
ECM projects @ Nvidia GPUs Karl M Johnson Factoring 2 2010-07-25 15:50
NVIDIA CUDA C toolkit for G80 GPU available dsouza123 Programming 2 2007-02-18 12:50

All times are UTC. The time now is 17:36.

Wed Jul 15 17:36:20 UTC 2020 up 112 days, 15:09, 1 user, load averages: 1.95, 1.83, 1.76

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.