GTX 780 F@H benchmarks

[warmon6]

http://www.tomshardware.com/reviews/geforce-gtx-780-performance-review,3516-29.html

 

 

 

 

Folding@Home (FP32)

The Folding@Home benchmark puts Nvidia's GeForce GTX 780 exactly halfway between the GeForce GTX Titan and GeForce GTX 680.

 

 

   

Looking at these results, two things become clear. First, Nvidia regrettably still appears to treat OpenCL as a second-class API. Second, the F@H benchmark proves that porting a CUDA application to OpenCL is not as trivial as it may sound. While the GeForce cards dominate when it comes to protein folding with the explicit solvent, AMD’s Radeon HD 7970 GHz Edition comes out on top in financial analysis performance.

Of course, it’s easy to find other benchmarks that show either one or the other company in the lead, but that’s not what we're after. Rather, we want to show how our field performs in these same tests switching to double-precision.

 

 

 

 

http://www.anandtech.com/show/6973/nvidia-geforce-gtx-780-review/18

 

 

Moving on, our 4th compute benchmark is FAHBench, the official Folding @ Home benchmark. Folding @ Home is the popular Stanford-backed research and distributed computing initiative that has work distributed to millions of volunteer computers over the internet, each of which is responsible for a tiny slice of a protein folding simulation. FAHBench can test both single precision and double precision floating point performance, with single precision being the most useful metric for most consumer cards due to their low double precision performance. Each precision has two modes, explicit and implicit, the difference being whether water atoms are included in the simulation, which adds quite a bit of work and overhead. This is another OpenCL test, as Folding @ Home has moved exclusively to OpenCL this year with FAHCore 17.

 

The Folding@Home group recently pushed out a major core update(FAHBench 1.2.0), which we’ve rerun on a number of cards and is reflected in our results. Unfortunately this version also broke single precision implicit on AMD GPUs and AMD’s latest drivers, so we only have NVIDIA GPUs for that section.

 

 

In any case, despite the fact that this is an OpenCL benchmark this is one of the cases where NVIDIA GPUs do well enough for themselves in single precision mode, with GTX 780 surpassing 7970GE, and falling behind only GTX Titan and the 7990. GTX 780 doesn’t necessarily benefit from GK110’s extra compute functionality, but it does see a performance improvement over GTX 680 that’s close to the theoretical difference in shader performance. Meanwhile in double precision mode, the lack of an uncapped double precision mode for GTX 780 means that it brings up the bottom of the charts compared to Titan and its 1/3 FP64 rate. Compute customers looking for a bargain NVIDIA card (relatively speaking) will need to stick with Titan.

 

 

To bad that I'm not finding PPD numbers. Although since we know what the Titan and 680 are producing, I guess it isn't all that hard to figure out. 

From the F@H blog here with the new Core_17: http://folding.typepad.com/news/2013/05/core-17-update-video-and-reddit-live-qa.html

 

GTX Titan: about 120K ppd 

GTX 680: 80K PPD

Then a GTX 780 being shown to perform in the middle of these 2 cards based on these benchmarks, it should be about 100K ppd.

 

 

 

 

If I find any more reviews with F@H in them, I'll add-on to this post.  :)

[Kitson]

Seems to fit in where it should i guess. Might seem like a stupid question but what does nanoseconds per day mean in relation to F@H?

[warmon6]

Seems to fit in where it should i guess. Might seem like a stupid question but what does nanoseconds per day mean in relation to F@H?

 

I dont know the full answer myself but what I do know is that it has something to do with how fast a protein folds. Beyond that, i dont have anymore as this has more meaning the Scientist in the F@H group than us non-scientist.

 

I'll try looking more into it and see what else i can learn.

 

 

EDIT: Well that didn't take long to find.

 

http://foldingforum.org/viewtopic.php?p=60355#p60355

 

 

ns/day : a WU is a part of simulation which is representing a certain amount of time. For example, a 1 000 ns simulation divided in 100 WU will make a single WU represents 10 ns. If you're folding one WU per day, you'll produce 10 ns/day ... two WU per day, 20 ns/day ... this value (ns/day) will vary, depending on the complexity of the protein (more atom typically means more iteration between two states of the protein, which means less ns/day).

 

Hope this clear things up!   :)

[Kitson]

I dont know the full answer myself but what I do know is that it has something to do with how fast a protein folds. Beyond that, i dont have anymore as this has more meaning the Scientist in the F@H group than us non-scientist.

 

I'll try looking more into it and see what else i can learn.

 

 

EDIT: Well that didn't take long to find.

 

http://foldingforum.org/viewtopic.php?p=60355#p60355

 

 

Hope this clear things up!   :)

 

Well that kinda clears it up, basically part of a work unit? .. Non scientist here haha (Im an Engineer).

Thanks for that :)

[Whaler_99]

The caveat here, if AMD with more new cores optimized for OpenCL stay where they are, NVidia is pretty much done for Folding when you look at the cost comparison... 

 

Be interesting to see how things come out once the Beta is over...

[Guest]

110,000 PPD for AMD Radeon 7970 and 120,000 PPD for Nvidia GTX Titan. Wow That's Crazy!

[Whaler_99]

110,000 PPD for AMD Radeon 7970 and 120,000 PPD for Nvidia GTX Titan. Wow That's Crazy!

 

yes, as has been mentioned - the FAH client itself hasn't been optimized yet to take advantage of the new GK processor in the 780/Titan series. If they did release something, I would expect those numbers to be a bit higher on the Titan.

[blindsighted26]

well i wish they would optimize it for GK110 but what i get on my 780 is ok i gess i get just above 103K ppd

[Guest]

yes, as has been mentioned - the FAH client itself hasn't been optimized yet to take advantage of the new GK processor in the 780/Titan series. If they did release something, I would expect those numbers to be a bit higher on the Titan.

True. Though it seems to me that these new F@Home benchmarks are starting to resemble OpenCL benchmarks like these

http://clbenchmark.com/result.jsp

and others I've seen. I don't think we will see much improvement through optimizing,maybe through new Nvidia OpenCL drivers but even then I think it will be small and prolly will not justify the price for a folding rig. 780/Titan still  the best for a gaming rig.

[Theo]

The 680 gets around 90-100k ( mine does, slightly OC'ed) on core17. I sure hope the titan/780 get more than what the OP said.

[jmrios]

well i wish they would optimize it for GK110 but what i get on my 780 is ok i gess i get just above 103K ppd

If they manage to optimize even more..I will be happy. 

My GTX 780 is producing 128k PPD on the 8900 beta WU. Compared to my GTX 570, 30k PPD, with the same 8900 WU. I think that they introduced some changes already. 

[blindsighted26]

If they manage to optimize even more..I will be happy. 

My GTX 780 is producing 128k PPD on the 8900 beta WU. Compared to my GTX 570, 30k PPD, with the same 8900 WU. I think that they introduced some changes already. 

is your 780 OC at all my PPD goes up and down so much i just averaged it out. my 780 is at stock clocks 

[warmon6]

The 680 gets around 90-100k ( mine does, slightly OC'ed) on core17. I sure hope the titan/780 get more than what the OP said.

 

1. Well when they report numbers, they mean in stock config, so if you overclock, you'll have more PPD.

 

2. Because it's been 3 months since I posted that, somethings have already changed. Different WU's, newer drivers that can affect performance (in a good direction), ect.

 

[jmrios]

is your 780 OC at all my PPD goes up and down so much i just averaged it out. my 780 is at stock clocks 

My GTX 780 is an EVGA, the reference one, no OC at all.. 

Do you run the beta WU's (8900, 7811 or 7810)? , If your PPD goes up and down I assume that you are getting 7811/7810 WU's.

You can try stopping your cpu folding, leaving only the GPU and wait and see if the PPD of the GTX 780 increases. I'm folding smp 6 on my 3770k because of that.. Since I have 2 cards,I leave 2 threads for them while folding the beta WU's

[blindsighted26]

My GTX 780 is an EVGA, the reference one, no OC at all.. 

Do you run the beta WU's (8900, 7811 or 7810)? , If your PPD goes up and down I assume that you are getting 7811/7810 WU's.

You can try stopping your cpu folding, leaving only the GPU and wait and see if the PPD of the GTX 780 increases. I'm folding smp 6 on my 3770k because of that.. Since I have 2 cards,I leave 2 threads for them while folding the beta WU's

hi im running MSI 780 reference card no OC , im running 7811 beta WU on a 3570k soon to be a 3770k as im upgrading my rig to watercool it .

 

I am also running an oc MSI 660 TI power edition card as well im not running my CPU as it a beta WU and found it use a lot of it 

 

 

thanks for relpying

[Whaler_99]

I have a Zotac OC GTX 780 due to arrive next Monday with an XSPC Razor full water block. Then I can crank that GPU Boost thermal and hopefully get some wicked overclocks...

[blindsighted26]

i have just order an ek waterblock for the gtx 780 so i will hopefully be the same

 

I have a Zotac OC GTX 780 due to arrive next Monday with an XSPC Razor full water block. Then I can crank that GPU Boost thermal and hopefully get some wicked overclocks...

[blindsighted26]

just got a meger WU on my 780 
 

 

 

[jmrios]

just got a meger WU on my 780 

 

I personally prefer those 8900 WU over the 7810/7811 for the GTX 780, they have a stable PPD, with the 7810/7811 the PPD is all over the place.. 

Your PPD seems ok, I got the same PPD with my GTX 780 on the 8900 WU. 

How many PPD is producing your GTX 660Ti? I'm thinking to retire my GTX 570. 

[blindsighted26]

my 660ti is doing 8900 WU's aswell just takes about 2 1/2 hours longer lol but its getting 26K so 32K on the 780 and 26K on the 660ti

[Guest]

I personally prefer those 8900 WU over the 7810/7811 for the GTX 780, they have a stable PPD, with the 7810/7811 the PPD is all over the place.. 

Your PPD seems ok, I got the same PPD with my GTX 780 on the 8900 WU. 

How many PPD is producing your GTX 660Ti? I'm thinking to retire my GTX 570. 

How do you make it/force it to do the 8900 WU instead of the 7811 WU?

[orangecat]

I wish I had a 780 not just for gaming but i would love to be in the 100K PPD+ club.

[Whaler_99]

How do you make it/force it to do the 8900 WU instead of the 7811 WU?

 

You cannot - at most you can make your card "available" for Core 17 - but as to what specific project you get on the Core, you have no control over.