New GPU3/Fermi Project 6806 + New Benchmarking GPU

[LCB001]

Pande Group Member Yslin has added a new project for GPU3/Fermi P6806 to advmethods, this one seems to give slightly better PPD than the 6800/6801 while impacting the CPU a bit less than 6800 but a bit more than the 6801's, YMMV.

The Pande Group has also changed the GPU used on the benchmarking machine from a GTX480 to a GTX460...

Hi,

A new GPU3/Fermi project 6806 will soon be added to adv. This is our second-stage sampling of the amyloid beta peptide (Abeta) mutant configurations. These beta tests are to evaluate the performance of our new work server code, new version of core15 (2.15) and the results will be used to build Markov states for Abeta monomer, which will serve as great starting points for future oligomerization studies.

Points and deadlines: 1348 points, preferred deadline 5 days, final deadline 10 days

Note: We have changed our benchmarking machine for GPU3/Fermi WUs from GTX480 to GTX460. Before, we would have a consistent PPD=12889 for GTX480. Now we are using PPD=9787 for GTX460. This PPD=9787 for GTX460 was chosen using Villin (p10632; 611 points) as the standard since Villin (number of atoms=582) is of medium size.

Thanks!

yslin

Enjoy

---------------


Edit/ For those wanting more info on the GPU3 Benchmarking machine used by Stanford;

Benchmark machine: GeForce GTX460 . Details:

• GPU: VGA GeForce GTX460 768MB DDR5 PCI-Express 2.0 Graphics Card 768-P3-1360-TR
• Processor: Dual Core AMD Opteron(tm) Processor 275 (2210 MHz)
• Operating System: Windows 7 Professional, 32-bit
• DirectX version: 11.0
• GPU processor: GeForce GTX 460
• Driver version: 258.96
• CUDA Cores: 336
• Core clock: 720 MHz
• Shader clock: 1440 MHz
• Memory clock: 1800 MHz (3600 MHz data rate)
• Memory interface: 192-bit
• Total available graphics memory: 2047 MB
• Dedicated video memory: 768 MB GDDR5
• System video memory: 0 MB
• Shared system memory: 1279 MB
• Video BIOS version: 70.04.1B.00.60
• Bus: PCI Express x16

PPD: 9787

For more info including the specs on the systems used for benchmarking other clients; Folding@home - FAQ-Points

[sswilson]

Thanks for the heads-up. I'll have to swap a couple of 450s over to -advmethods to check them out.

[Sagath]

Wow! Thats really interesting to note that the 'standard' was based off a GTX480, and that theyve changed it to a 460...

I wonder if this will retroactively change the other WU values, since that would mean that the larger units (like the 6800 and 6811s) should give a larger point total on completion...Somehow I doubt it

[sswilson]

I doubt it will change the values of current WUs, but the scaling from performance on a 460 to a 480 should prove beneficial to 480 owners. :)

I predict my OC'd 450s will be putting out 10.5, but won't know for 2+ hours.... ;)

[Sagath]

Originally Posted by sswilson

I doubt it will change the values of current WUs, but the scaling from performance on a 460 to a 480 should prove beneficial to 480 owners. :)

I predict my OC'd 450s will be putting out 10.5, but won't know for 2+ hours.... ;)

I agree. I just snagged one on my 580...TPF 1:04, same as the 6811's. Better PPD for sure, the WU is worth about 150 more points on completion.

Just pulled PPD. It works out to be the same as the 6811's, 18k ppd...It doesnt seem to be stressing the card as much as the 6811's, the card is running cooler. That being said this will probably be a good WU for the 450/460's then when directly compared to the 6800/6811 WU's since they upped the point value vs them.

Edit: CPU usage is down, but not much. Pulls about 1.5% on my 970.

[stoanee]

Looks like about 10900ppd +- compared to 10500ppd on my medium overclocked GTX460's...............

[sswilson]

Heh.... "slight" improvement indeed.... ;)

9.7K vice 9.5K (6800) for my highest clocked 450.

Temps are the same as the 6800s and virtually no change to SMP TPF on a dual card rig with both cards folding 6806 projects.

edit: BTW... there's a core update (outdated core) attached to these, so if funky stuff starts happening when you return to the older WUs, that'd be the first place I'd look.

[chrisk]

Originally Posted by sswilson

edit: BTW... there's a core update (outdated core) attached to these, so if funky stuff starts happening when you return to the older WUs, that'd be the first place I'd look.

Spoken like a wily folding veteran.

I'll keep an eye on the folding forums for any info on this, as I am currently only folding on the CPU while I get my main rig stable.

[sswilson]

It should also be noted (slightly off topic, but folks might start looking here first because of the new WUs) that stanford is going to be doing server maintenance over the next week or two, so expect to see some idle clients from time to time. There have also been reports of data being turned in without credit, and it looks like they don't have the information available to re-credit.

I guess we'll just have to grin and bare it..... ;)

Folding@home

[G777]

Hey, I wonder if any of you motherfolders who have gpus folding the new P6806s can check and see if your points are updating. My 2 gtx480s have been chewing through WUs since yesterday, but as you can see no points have been given. Started using FAH GPU Tracker V2 a week ago, have no idea if that has caused some sort of problem... can't imagine how, 6800s were fine.

Also saw a post on Ncix forum saying Stanford was warning servers would be down for maintenance:

"Tomorrow and over the holidays there will be a few wu shortages and the servers will go down for maintenance. They should be brief and not too much to worry about though.

From Stanford:
We've been working to try to minimize the effect of the server room maintenance coming up tomorrow. Since this is our main room, FAH will be stretched pretty thin during the outage and we expect there will be WU shortages. Also, clients will not be able to send WUs back to servers that will be unavailable during the outage.

However, some good news is that we have been able to get power to a few key machines so the stats and web page should be up as well as most of the key servers. If you have trouble getting or returning WUs tomorrow, please just wait it out until we get a chance to get everything back on line."

Daily Production
Day Points WUs
12.15.10 0 0
12.14.10 28,906 22
12.13.10 92,029 24
12.12.10 20,768 16
12.11.10 7,788 6
12.10.10 12,980 10
12.09.10 84,083 11
12.08.10 11,682 9
12.07.10 12,980 10
12.06.10 83,026 12
12.05.10 12,980 10
12.04.10 11,682 9
12.03.10 86,034 12
Hourly Production
Time Points WUs
12.15, 3pm 0 0
12.15, 12pm 0 0
12.15, 9am 0 0
12.15, 6am 0 0
12.15, 3am 0 0
12.15, 12am 4,044 3
12.14, 9pm 2,696 2
12.14, 6pm 5,292 4
12.14, 3pm 2,596 2
12.14, 12pm 3,894 3
12.14, 9am 3,894 3
12.14, 6am 2,596 2
12.14, 3am 3,894 3