low-end GPU comparison (AMD vs. Nvidia)

Turbonium · Post by **Turbonium** » Thu Oct 23, 2014 5:15 pm

AMD Radeon R7 240
vs.
Nvidia GeForce GT 620

I realize both options are fairly abysmal, but if you had to choose between the two (in terms of computing power only, and ignoring TDP or power draw), which would it be?

codysluder · Post by **codysluder** » Thu Oct 23, 2014 6:24 pm

The GeForce GT 620 card is rated at about 270 GFLOPs. The Radeon R7-240 is about 470 GFLOPs.

Turbonium · Post by **Turbonium** » Thu Oct 23, 2014 10:09 pm

codysluder wrote:The GeForce GT 620 card is rated at about 270 GFLOPs. The Radeon R7-240 is about 470 GFLOPs.

Well then, that settles it.

I wasn't aware you could just look up the GFLOPs for a given GPU and use it as an accurate measure of compute power in terms of Folding (actually, I didn't even know you could look up the GFLOPs rating to begin with).

Are the GFLOPs for a given GPU determined based on the Folding userbase (taking an average)? Or is the value provided by the chipset manufacturer?

Also, what is the estimated PPD for the R7 240? I'm guessing around 75,000+?

Post by **Joe_H** » Thu Oct 23, 2014 10:22 pm

Easy enough to look up on wikipedia, persons have posted lists of the GPU's from AMD and nVidia including specifications from the manufacturers. However, as a fairly low end card I suspect it would not be fast enough to process current GPU folding WU's within the deadlines. Someone who has a similar card may comment on that. Given my experience folding on a 5870 which is rated at 2720 GFLOPS, a higher end Radeon R7 or a R9 series card might be about the minimum needed to meet the deadlines.

Turbonium · Post by **Turbonium** » Thu Oct 23, 2014 10:41 pm

Joe_H wrote:Easy enough to look up on wikipedia, persons have posted listed of the GPU's from AMD and nVidia including specifications from the manufacturers. However, as a fairly low end card I suspect it would not be fast enough to process current GPU folding WU's within the deadlines. Someone who has a similar card may comment on that. Given my experience folding on a 5870 which is rated at 2720 GFLOPS, a higher end Radeon R7 or a R9 series card might be about the minimum needed to meet the deadlines.

I went straight to AMD's own website, and saw no information on GFLOPs. I also didn't find the info on the Wikipedia page, despite visiting it earlier.

Anyway, I'm assuming they're fairly accurate calculations, regardless of whether or not the manufacturer provides the value:

4 Single precision performance is calculated as two times the number of shaders multiplied by the base core clock speed.
5 Double precision performance of Hawaii is 1/8 of single precision performance, Tahiti is 1/4 of single precision performance, others 28 nm chip is 1/16 of single precision performance.

As suggested, if someone can confirm whether or not a low-end R7 series GPU will fail to meet deadlines, that would be helpful.

7im · Post by **7im** » Thu Oct 23, 2014 10:42 pm

codysluder · Post by **codysluder** » Thu Oct 23, 2014 11:16 pm

I got both those numbers from Wikipedia, so I know you can find a chart that shows them.

GFLOPs does not provide an exact indication of F@H performance but it does provide a reasonable indication.

Turbonium · Post by **Turbonium** » Thu Oct 23, 2014 11:35 pm

7im wrote:Not.

Are you saying an R7 240 will not fail to meet deadlines? Please clarify.

codysluder wrote:I got both those numbers from Wikipedia, so I know you can find a chart that shows them.

GFLOPs does not provide an exact indication of F@H performance but it does provide a reasonable indication.

Yea, I'm saying I found it as well after you referred to Wikipedia. The same page is where I got the quote from regarding calculating GFLOPs.

7im · Post by **7im** » Fri Oct 24, 2014 12:51 am

Turbonium wrote:
7im wrote:Not.
Are you saying an R7 240 will not fail to meet deadlines? Please clarify.

I answered the question you asked.

Turbonium wrote:snip...

As suggested, if someone can confirm whether or not a low-end R7 series GPU will fail to meet deadlines, that would be helpful.

"will not fail to meet deadlines" aka will meet deadlines, from personal experience...

Code: Select all

14:49:32:Enabled folding slot 01: gpu:0:Hawaii [Radeon R7 240/HD8570]

jrweiss · Post by **jrweiss** » Fri Oct 24, 2014 2:23 am

I Fold on AMD7750s which are "old, low-end" cards, in 2 separate machines. They make deadlines just fine.

At http://www.overclock.net/t/502403/graphics-card-ranking the 7750 rates #99; your 2 rate 125 and 155.

If you're thinking about buying a new card, I wouldn't bother with either one. Wait until you can afford a 750 Ti.

A quick comparison is at http://gpuboss.com/gpus/GeForce-GTX-750 ... rce-GT-620 or http://gpuboss.com/gpus/GeForce-GTX-750 ... rce-GT-620 or http://gpuboss.com/gpus/Radeon-R7-240-vs-GeForce-GT-620

Turbonium · Post by **Turbonium** » Fri Oct 24, 2014 2:49 am

7im wrote:
Turbonium wrote:
7im wrote:Not.
Are you saying an R7 240 will not fail to meet deadlines? Please clarify.
I answered the question you asked.

Turbonium wrote:snip...

As suggested, if someone can confirm whether or not a low-end R7 series GPU will fail to meet deadlines, that would be helpful.
"will not fail to meet deadlines" aka will meet deadlines, from personal experience...

Yes, I figured. I just wanted clarification to be certain. You could have just said "Correct." to keep up the trend.

jrweiss wrote:I Fold on AMD7750s which are "old, low-end" cards, in 2 separate machines. They make deadlines just fine.

At http://www.overclock.net/t/502403/graphics-card-ranking the 7750 rates #99; your 2 rate 125 and 155.

If you're thinking about buying a new card, I wouldn't bother with either one. Wait until you can afford a 750 Ti.

A quick comparison is at http://gpuboss.com/gpus/GeForce-GTX-750 ... rce-GT-620 or http://gpuboss.com/gpus/GeForce-GTX-750 ... rce-GT-620 or http://gpuboss.com/gpus/Radeon-R7-240-vs-GeForce-GT-620

Around 3x the GFLOPs, slightly higher TDP?

7im · Post by **7im** » Fri Oct 24, 2014 2:53 am

JobElihu · Post by **JobElihu** » Fri Oct 24, 2014 8:27 am

I fold on a GTX 460 and I've never missed a deadline so far.

Turbonium · Post by **Turbonium** » Fri Oct 24, 2014 8:39 pm

Turbonium wrote:Around 3x the GFLOPs, slightly higher TDP?

Regarding this:

So if i leave the video card BIOS as-is, it will not go over 40W TDP, even within a Linux environment (meaning roughly 300% the GFLOPs, at a cost of 133% TDP)?

I'm asking due to the discrepancy between Nvidia's advertised TDP for the card (60W), versus the actual TDP (~40W) as dictated by its BIOS.

TonyStewart14 · Post by **TonyStewart14** » Sat Oct 25, 2014 3:07 am

I fold on a 620 using the v6 client. The WUs allow about 45 days for work units that take about 3 days to complete if you have it running 24/7, so deadlines aren't a problem. I get about 5K PPD, which isn't great compared to the high end cards, but I don't have to worry about deadlines or QRB affecting the points.

Folding Forum

low-end GPU comparison (AMD vs. Nvidia)

low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)

Re: low-end GPU comparison (AMD vs. Nvidia)