mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Hardware (https://www.mersenneforum.org/forumdisplay.php?f=9)
-   -   Prime95 On Graphics Card (https://www.mersenneforum.org/showthread.php?t=3309)

louis_net 2004-11-14 20:47

Prime95 On Graphics Card
 
I have heard that there was a way to convert data into something the gpu can process and run stuff like pixel shaders on it, then convert that back to data that prime95 can use.

Any ideas?

PrimeCruncher 2004-11-15 00:43

[QUOTE=louis_net]I have heard that there was a way to convert data into something the gpu can process and run stuff like pixel shaders on it, then convert that back to data that prime95 can use.

Any ideas?[/QUOTE]

*Sigh* Here we go with this thread again.

Prime95 can NOT run on GPUs. There's some issue, forget exactly what it is. I think it's because GPUs only have single-precision floating point and Prime95 requires double-precision.

sdbardwick 2004-11-15 01:20

Previous discussion on this topic: [url="http://www.mersenneforum.org/showthread.php?t=432"]http://www.mersenneforum.org/showthread.php?t=432[/url]

cheesehead 2004-11-18 06:09

As I wrote in an earlier thread,

[quote]The potential use of graphics cards to do L-L testing or factoring has been discussed several times in the Hardware forum. (Search on "graphics" or "video" there.)

The main reason video cards are not suitable for GIMPS work is that GIMPS calculations require high precision and so use double-precision floating-point arithmetic, but video cards operate with only single-precision floating-point numbers. Single-precision FP is all that video work requires, so it is unlikely that any future video cards will be able to perform double-precision FP, not matter how advanced their other capabilities or speed.

Q: Why not use single-precision?
A: It has to do with the technical details of FFT arithmetic, especially the need to guard against losing low-order result bits because of FP rounding/truncation. For more information, search the Math forum threads discussing this subject.[/quote]

E_tron 2004-11-23 18:36

Another big bottle neck is the AGP interface. There is alot of bandwidth to the GPU, but little bandwidth from it.

PCIexpress opens up a interesting possibility. We might see secondary CPU's in the form of an add-in card, because the interface is so flexable. It would be funny to see a machine with both and intel and AMD chip in it.

Xyzzy 2004-11-23 20:40

[QUOTE=E_tron]AWe might see secondary CPU's in the form of an add-in card, because the interface is so flexable. It would be funny to see a machine with both and intel and AMD chip in it.[/QUOTE]
This was posted today, but I haven't been able to find any info on it:

[url]http://lists.duskglow.com/open-graphics/2004-November/000292.html[/url]

PrimeCruncher 2004-11-24 00:08

[QUOTE=Xyzzy]This was posted today, but I haven't been able to find any info on it:

[url]http://lists.duskglow.com/open-graphics/2004-November/000292.html[/url][/QUOTE]

If that exists, it would be most impressive. It would be more impressive if we could find a way to run Prime95 on it, considering I have two MP 1900+s collecting dust at the moment...

E_tron 2004-11-25 23:11

ASRock i think will get there first. They were actually brave enough to create a CPU AGP addin card.

PCIe has alot of unspecified specs, so all sorts of traffic can cross it.

Ivan Semenov 2004-11-26 06:06

No holiday today :mellow:

[url]http://lists.duskglow.com/open-graphics/2004-November/000346.html[/url]

PrimeCruncher 2004-11-26 16:10

[QUOTE=Ivan Semenov]No holiday today :mellow:

[url]http://lists.duskglow.com/open-graphics/2004-November/000346.html[/url][/QUOTE]

*Sigh* An April fools joke. Damn. Too bad.

ColdFury 2004-11-26 22:05

Reminds me of the Seti@Home accelerator card hoax.


All times are UTC. The time now is 23:20.

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