• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

GPU folding results

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
That is odd, I thought the 8800GT was stripped down from 8800gts..why so much faster than it? Its about 1000ppd faster than mine..
 
Originally posted by: imaheadcase
That is odd, I thought the 8800GT was stripped down from 8800gts..why so much faster than it? Its about 1000ppd faster than mine..
Click to expand...

You've got a 320MB 8800GTS (G80) while the 8800GT uses the G92 processor. It's considerably faster than the older G80 cards.

On a side note, there doesn't appear to be any real difference between cards with 320MB and 640MB memory and the G92-based 8800GT and 8800GTS run the same ppd. So basically for those folding farms, just buy the cheapest card you can with the right core and crunch away. (There are 8800GT 256MB cards on ZZF for ~$110 AR with upgraded cooling that should be superb for folding.)
 
I bought two 4850s at BB. For folding purposes, would it be better to put each card in a seperate machine, or to put them in teh same machine? I want to crossfire them for gaming, but as I understand it, the GPU2 client is incompatible with crossfire. Can you leave them in a physical crossfire configuration (with the bridges installed), and then just disable crossfire in software, in order to run dual GPU2 clients?

The systems would have quad-cores. The other question is if running two GPU2 clients under XP would require two dedicated CPU cores, thus fairly prohibiting running the SMP CPU client.
 
Since you have multiple boxes available I would say run them separately, at least to start. Someone more knowledgeable on CF & F@H will have to comment on running on a single machine.
 
Originally posted by: imaheadcase
That is odd, I thought the 8800GT was stripped down from 8800gts..why so much faster than it? Its about 1000ppd faster than mine..
Click to expand...

Here's the link to the one I bought. It's OCd from 600 to 650 MHZ and has 512 MB VRAM. Not sure what processor it has but it looks like I got lucky since I'm no expert.😉

4750 ppd was initial estimate from fahmon so time will tell on how well it does.

:beer:
 
hehe im not gonna downgrade to a 8800gt just for folding though 😛 Im gonna get me a ATI 4870 next month sometimes anyways.
 
The 8800GS I picked up for $80AR has given ~4000 points over the last 24 hours.

Talk about great value for such a low cost! 😀

And it's dead silent also, can't even hear it in the case next to me with the side off.
 
8800GT w/ 512MB, OC'd to 650 MHz.

After 4 days, the fahmon "effective rate" is 4702 ppd.

fahmon benchmarks for different WUs:

Project : 5002
Core : GPUv2 Gromacs
Frames : 100
Credit : 98


-- GPU --

Min. Time / Frame : 17s - 4980.71 ppd
Avg. Time / Frame : 17s - 4980.71 ppd
Cur. Time / Frame : 18s - 4704.00 ppd
R3F. Time / Frame : 17s - 4980.71 ppd
Eff. Time / Frame : 17s - 4980.71 ppd


Project : 5003
Core : GPUv2 Gromacs
Frames : 100
Credit : 479


-- GPU --

Min. Time / Frame : 1mn 28s - 4702.91 ppd
Avg. Time / Frame : 1mn 28s - 4702.91 ppd
Cur. Time / Frame : 1mn 29s - 4650.07 ppd
R3F. Time / Frame : 1mn 28s - 4702.91 ppd
Eff. Time / Frame : 1mn 28s - 4702.91 ppd

Project : 5004
Core : GPUv2 Gromacs
Frames : 100
Credit : 479


-- GPU --

Min. Time / Frame : 1mn 28s - 4702.91 ppd
Avg. Time / Frame : 1mn 28s - 4702.91 ppd
Cur. Time / Frame : 1mn 28s - 4702.91 ppd
R3F. Time / Frame : 1mn 28s - 4702.91 ppd
Eff. Time / Frame : 1mn 28s - 4702.91 ppd

:beer:
 
Ok, I managed to piece together another system from the ol' parts closet.

XP Pro SP3
X2 3800+
abit AN-M2
512MB DDR2
Visiontek 4850

Used the Overdrive profile edit trick to set the fanspeed to 55% and it now idles at 58C while crunching (standing on a box on my desk, not in a case atm).

I started it running the GPU2 client this evening, I hope to have ppd to report in a day or two.
 
Consistently getting 4868 PPD per Fahmon...

Specs:

EVGA 8800 GT
Q6600 - 3.3 ghz
P35-DS3R
4 GB RAM
Vista 64
177.35 drivers


It has been pretty consistent in that it will download a project worth 479 points every 2.5 hours.

Pretty sweet compared to the 1000 PPD I was getting before with the SMP client alone.
 
Getting around 1700 PPD on the video card and another 235 PPD on the CPU

Specs:
XFX 8600GT XXX @ 620 core / 1600 mem
E4500 @ 2.2GHz
P35-DS3L
2 GB RAM
XP Pro
177.35 drivers
 
Yeah, I found the same info. Looks like the Folding team hasn't optimized for the much increased shader power yet. Probably too busy working out how to get maximum performance out of nVidia hardware, if I had to guess.

I'll bet that changes fairly quickly, there are going to be a lot of people making noise on the boards to "equalize" points between the camps.
 
The HD3xxx and HD2xxx should get faster too since they are only at the starting phase of the development. Nvidia currently pushes performance, while the ATI team pushes stability and compatibility for the F@H client development
 
Yeah, the ATI card owners are a little upset. thread

F@H is never boring when it comes to hardware/ppd controversies!😉

:beer:
 
Originally posted by: biodoc
Yeah, the ATI card owners are a little upset. thread

F@H is never boring when it comes to hardware/ppd controversies!😉

:beer:
Click to expand...

Ain't it the truth :roll:

I've been on both sides of it over the years. If I learned anything, it's that I need to house the hardware that is good for my PC. (As opposed to what's good for F@H points.)

Hardware and Stanford priorities can both change pretty fast.

-Sid

 
Originally posted by: Insidious
I've been on both sides of it over the years. If I learned anything, it's that I need to house the hardware that is good for my PC. (As opposed to what's good for F@H points.)

Hardware and Stanford priorities can both change pretty fast.

-Sid
Click to expand...

I agree with Sid, get what you want or need in real life and if it can be used to crunch then great.

Especially for GPU as the fastest card will change in a few days either due to better hardware or different types of WUs.

Or like GPU1, they will drop it completely :frown: :| 🙁
 
Originally posted by: Insidious
I've been on both sides of it over the years. If I learned anything, it's that I need to house the hardware that is good for my PC. (As opposed to what's good for F@H points.)

Hardware and Stanford priorities can both change pretty fast.

-Sid


Reply from: GLeeM
I agree with Sid, get what you want or need in real life and if it can be used to crunch then great.

Especially for GPU as the fastest card will change in a few days either due to better hardware or different types of WUs.

Or like GPU1, they will drop it completely
Click to expand...
Click to expand...

Dang, GLeeM - Might as well retire my ati 1950 card on your last statement. :| I was running the smp on my AMD am2 duo core box and shut that down to see if the new workfiles for the 1950 card would show some effort on the GPU1 side - let it run for about a day doing 2 projects. it averaged out to about less than 200ppd!

I will try to get one of those o/c'ed 9800gt's and slap it in it's place and see how it goes.
I will probably donate the 1950 card to a retired vietnam vet who runs a flower shop in my neighborhood and cobbles together pc's from spare parts for other vets & their families who don't have pc's or who need an extra one.

TurtleBlue

Major Kudo's for 😎Denithor:thumbsup: for starting up this topic - really appreciated it!🙂
 
Err, IIRC they killed the GPU1 client several weeks ago.

Must admit I can't remember what ppd I was getting on my X1950Pro, I thought it was more than 200 though? Anyone with better memory remember their output? (or find my old thread about GPU temps).
 
Err, IIRC they killed the GPU1 client several weeks ago.

Must admit I can't remember what ppd I was getting on my X1950Pro, I thought it was more than 200 though? Anyone with better memory remember their output? (or find my old thread about GPU temps).
Click to expand...

😕😕Ooops, Assimilator1 - now I know that I am suffering from a full-blown case of dementia!😕😕

Looked at my log files under my GPU subfolder and saw that it was NOT the GPU client, I was running FAHcore82.exe and not the Core 10 gpu client

Sorry about the erroneous post!🙁🙁🙁

No wonder the temps on the 1950 card never changed!

I need to go to my corner and put on my "D" cap now...

TurtleBlue
 
my Q9450/3870 approx (was) 1917 ppd now 2559ppd
E7200/2900Pro Approx (was) 1300 ppd now 1657ppd

BTW, this all depends on the project. +/- a few


So, not to bad with a combined total ppd. 🙂



Edit: updated bios in both cards:

 
You must log in or register to reply here.

TRENDING THREADS

Back
Top