Message boards : Number crunching : GPU advertised for LHC, but they don't do it?
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Jim1348

Send message
Joined: 15 Nov 14
Posts: 395
Credit: 11,782,400
RAC: 2,610
Message 40265 - Posted: 24 Oct 2019, 16:32:49 UTC - in response to Message 40263.  

For some curious reason, my Radeon RX 570 does well on MilkyWay (Win7 64-bit), even though it is merely OK, not great on DP.
https://milkyway.cs.rpi.edu/milkyway/results.php?hostid=803662

So there are various factors involved. We will really need to try out a card and see how it does.
ID: 40265 · Report as offensive     Reply Quote
Profile MAGIC Quantum Mechanic
Avatar

Send message
Joined: 24 Oct 04
Posts: 902
Credit: 39,296,766
RAC: 12,701
Message 40266 - Posted: 24 Oct 2019, 17:31:08 UTC

The Einstein Project is the place to go if you want to run GPU tasks.

Or find out anything about running them every day from the experts.
ID: 40266 · Report as offensive     Reply Quote
[VENETO] boboviz

Send message
Joined: 7 May 08
Posts: 39
Credit: 218,356
RAC: 1
Message 40271 - Posted: 24 Oct 2019, 18:57:10 UTC - in response to Message 40258.  

We are still in the process of developing SixTrackLib, validating the physics and tackle the integration into the main SixTrack. Our time scale is still probably a year or two to expose the functionality into LHC@Home, but we already performing simulations using SixTrackLib in GPU for specific problems.


That's great!
If you need help to debug gpu app.... (i think you will use lhc-dev project) :-P
ID: 40271 · Report as offensive     Reply Quote
Peter Hucker

Send message
Joined: 12 Aug 06
Posts: 80
Credit: 318,225
RAC: 1,674
Message 40276 - Posted: 25 Oct 2019, 11:05:16 UTC - in response to Message 40245.  

Then how come there are projects which only use ATI but not Nvidea?

Which project? All opencl app support both Amd and Nvidia gpu.
The only "problem" may be the different optimization of the code.


There aren't, but there ones that don't list Intel, but do list AMD, which is strange if OpenCL works on both.
ID: 40276 · Report as offensive     Reply Quote
Peter Hucker

Send message
Joined: 12 Aug 06
Posts: 80
Credit: 318,225
RAC: 1,674
Message 40278 - Posted: 25 Oct 2019, 11:10:45 UTC - in response to Message 40258.  
Last modified: 25 Oct 2019, 11:13:35 UTC


Despite this, GPUs also support 64bit floating point calculations but at a lower speed, typically 16/32 times more slowly, however still the computing capabilities they provide are comparable and often exceed the ones of CPUs. Even the latest embedded GPUs in notebooks are fairly capable. There are also GPUs that are strong in 64bit arithmetic such as Nvidia Quadro GV100, Tesla V100/P100, Titan V, AMD Radeon VII, Firepro W9100, W8100 although they are very expensive and less common in workstations (instead more common in supercomputers and the cloud providers).


2nd hand cards are great for 64 bit - The R9 280X is the best I've found. £60 2nd hand, and it does 1024 GFLOPS at 64bit (that's 1:4 ratio - older cards tended to have more double precision). More expensive ones are faster, but you're better buying several of these instead for the same price and getting even more computing power. I'm going to experiment with PCI express splitters at some point when I get more cards - it could be possible to get about 20 GPUs connected to one motherboard - I know bitcoin miners have done it.
ID: 40278 · Report as offensive     Reply Quote
Peter Hucker

Send message
Joined: 12 Aug 06
Posts: 80
Credit: 318,225
RAC: 1,674
Message 40280 - Posted: 25 Oct 2019, 11:32:19 UTC - in response to Message 40258.  


Generally GPUs have very strong 32bit floating point capabilities, but we will not be able to profit from them. The calculations needed by SixTrack requires variables with large dynamic range, which is very different from tasks like machine learning.


Is it true that you can do a 64bit calculation using two or more 32bit floating point units on the card? What's the performance hit if you did?
ID: 40280 · Report as offensive     Reply Quote
Profile F6FGZ looking for DX !

Send message
Joined: 7 Jan 07
Posts: 18
Credit: 9,396,967
RAC: 13,321
Message 41137 - Posted: 1 Jan 2020, 14:46:35 UTC - in response to Message 40280.  

Is it true that you can do a 64bit calculation using two or more 32bit floating point units on the card? What's the performance hit if you did?


Typically reported performance is ~40% of FP32 performance, which is an order of madnitude better than the 1/24 (~4%) FP64:FP32 performance available in almost all consumer-grade GPUs.

Source

HNY 2020 !
ID: 41137 · Report as offensive     Reply Quote
Peter Hucker

Send message
Joined: 12 Aug 06
Posts: 80
Credit: 318,225
RAC: 1,674
Message 41139 - Posted: 1 Jan 2020, 18:37:06 UTC - in response to Message 41137.  

Is it true that you can do a 64bit calculation using two or more 32bit floating point units on the card? What's the performance hit if you did?


Typically reported performance is ~40% of FP32 performance, which is an order of madnitude better than the 1/24 (~4%) FP64:FP32 performance available in almost all consumer-grade GPUs.

Source

HNY 2020 !


Interesting. I tend to buy up 2nd hand cards from back when they had a better ratio. Still, mine is 1:4, so 25%. So I wonder why Milkyway uses double precision? According to what you said, they'd get more calculations out of even my card.
ID: 41139 · Report as offensive     Reply Quote
Profile F6FGZ looking for DX !

Send message
Joined: 7 Jan 07
Posts: 18
Credit: 9,396,967
RAC: 13,321
Message 41141 - Posted: 2 Jan 2020, 15:09:33 UTC - in response to Message 41139.  
Last modified: 2 Jan 2020, 15:11:03 UTC

Interesting. I tend to buy up 2nd hand cards from back when they had a better ratio. Still, mine is 1:4, so 25%. So I wonder why Milkyway uses double precision? According to what you said, they'd get more calculations out of even my card.

It's all about the needed precision to do the maths, no rounding errors, no overflow, etc ...
A reading to have some clues : Floating Point's not Real
ID: 41141 · Report as offensive     Reply Quote
David

Send message
Joined: 11 Apr 17
Posts: 38
Credit: 6,765,239
RAC: 27,947
Message 41946 - Posted: 18 Mar 2020, 9:16:13 UTC - in response to Message 41141.  

Well, 64 bit floats still give you rounding errors but they tend to be way out there past the decimal points. :)
ID: 41946 · Report as offensive     Reply Quote
Peter Hucker

Send message
Joined: 12 Aug 06
Posts: 80
Credit: 318,225
RAC: 1,674
Message 41991 - Posted: 24 Mar 2020, 19:45:28 UTC - in response to Message 41946.  

Well, 64 bit floats still give you rounding errors but they tend to be way out there past the decimal points. :)


Can more calculations at FP32 not be done to make the accuracy better? On cards with stupidly low ratios like 1:24, Even doing 10 calculations at FP32 would be faster than 1 at FP64.
ID: 41991 · Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jun 08
Posts: 1362
Credit: 68,429,088
RAC: 101,829
Message 41994 - Posted: 24 Mar 2020, 20:34:24 UTC - in response to Message 41991.  

https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html
This might be a good starting point to understand why it is not a good idea to work with reduced precision.
ID: 41994 · Report as offensive     Reply Quote
Peter Hucker

Send message
Joined: 12 Aug 06
Posts: 80
Credit: 318,225
RAC: 1,674
Message 41995 - Posted: 24 Mar 2020, 20:56:29 UTC - in response to Message 41994.  

https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html
This might be a good starting point to understand why it is not a good idea to work with reduced precision.


I know reduced precision is bad. I don't have enough maths skills to understand any papers on it. But can't more calculations be done at FP32 to make it more accurate? Even if it requires several, it would still be faster than FP64 on most cards.
ID: 41995 · Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : Number crunching : GPU advertised for LHC, but they don't do it?


©2020 CERN