GPUs are better then CPUs

[evilfrog]

"GPUs remain Folding@home's most powerful platform in terms of FLOPS; as of November 2012 GPU clients account for 87% of the entire project's x86 FLOPS throughput"
from Folding@home - Wikipedia, the free encyclopedia
and from Folding@home - FAQ-highperformance :
However, after much work, we have been able to write a highly optimized molecular dynamics code for GPU's, achieving a 20x to 40x speed increase over comparable CPU code for certain types of calculations in FAH. This means that we will be able to make an enormous advance over what we could do only just a few years ago.
Which GPUs will be supported?
We support CUDA-class GPUs for NVIDIA (primarily G80 and later) and R6xx or later GPUs for ATI. This essentially encapsulates most of the more modern GPUs.

So if you have a gaming rig, make sure your GPU slot on the client is active!

 

[Thrasher]

agreed, with the transistor count inside the GPU's almost tripling the CPU nowadays they are def more adept at crunching numbers and calculations.

 

[CountSmackula]

But it bogs down my display. I have to pause GPU folding to be able to use my machine.

 

[Orobas]

But it bogs down my display. I have to pause GPU folding to be able to use my machine.

This. I can't have it going constantly.

 

[MattBott]

I just picked up a Radeon HD 7950 Boosted with part of my tax return. Right now I'm running two monitors, one with netflix playing a movie, another with this chrome window open on half of the screen, WoW running in the background doing my auctioning (can see half of the frame) AND folding. No jitters, skips, anything. Love this card!

Before I was just folding on my i7 920 (stock speeds). Today is day 2 of folding with this card, I'm interested to see if I shoot up the ranks with it.

 

[jesusboots]

Amd card/cpu's are underwhelming in folding performance.

They change their core architecture every generation, by a great margin. While nvidia slightly tweaks each time. For the same price, and near the same wattage, a gtx 580 brand new triples, and for some work units quadruples the ppd of the 7950.

Great card for other purposes, games. Any other distributed computing AMD, seems to out pace Nvidia in performance, except folding.

 

[loft]

The cost per video card vs cost for a 2 cpu rig is terrible. You would need at least 3x NVIDIA video cards (gtx580 or better) to match an SMP only rig with 2 CPU/16 physical cores. The system with the video cards would be pushing roughly 700-800 watts at the wall, and the system with the 2cpu's would be pushing roughly 300 watts at the wall. Not only would the video card rig be more expensive to make, it will be more expensive to run on a PPD/Watt break down.

Overall it is cheaper and more efficient to have a dedicated folding machine for SMP (CPU) folding. Cheaper to build, cheaper to run, and generates more work.

 

[b_1620]

Amd card/cpu's are underwhelming in folding performance.

They change their core architecture every generation, by a great margin. While nvidia slightly tweaks each time. For the same price, and near the same wattage, a gtx 580 brand new triples, and for some work units quadruples the ppd of the 7950.

Great card for other purposes, games. Any other distributed computing AMD, seems to out pace Nvidia in performance, except folding.

Sorry for the hijack but is this the same jesusboots from OCN?

 

[Crash158]

I just picked up a Radeon HD 7950 Boosted with part of my tax return. Right now I'm running two monitors, one with netflix playing a movie, another with this chrome window open on half of the screen, WoW running in the background doing my auctioning (can see half of the frame) AND folding. No jitters, skips, anything. Love this card!

Before I was just folding on my i7 920 (stock speeds). Today is day 2 of folding with this card, I'm interested to see if I shoot up the ranks with it.

Best of luck to ya but you are the next target on my overtake report Although I will be tearing down the system tomorrow to put in a new mobo so I will lose most of the day. But remember it is for a good cause.
As jesusboots has pointed out ATI/AMD cards are underutilized for folding. I love their cards for price and performance but they are not optimized for folding. Whether that is stanford or ATI is really up for debate but I recently made the switch and bought a nice nvidia card. The difference is quite noticeable.

 

[jfalbanese]

i have a nvidia gtx630 w 4gigs ddr3. the program does not like my video card. results keep getting rejected. and i get a "failed" status on my browser page. anyone have any ideas?

 

[loft]

Is your card overclocked? What does the log say regarding the server rejecting the WU's?

 

[jfalbanese]

Is your card overclocked? What does the log say regarding the server rejecting the WU's?

as stated below really didn't pay to much attention to the messages from the server, other than "rejected", and a failed button in the browser page below my cpu status bar. card is not overclocked installed out of the box. not a gamer. more into movies and videos. i can also enable my onboard video card if that helps at all, but nothing connected to it.

 

[jfalbanese]

16:23:20:WU00:FS00:0x15:- Expanded 58293 -> 257358 (decompressed 441.4 percent)
16:23:20:WU00:FS00:0x15:Called DecompressByteArray: compressed_data_size=58293 data_size=257358, decompressed_data_size=257358 diff=0
16:23:20:WU00:FS00:0x15:- Digital signature verified
16:23:20:WU00:FS00:0x15:
16:23:20:WU00:FS00:0x15roject: 8070 (Run 158, Clone 7, Gen 12)
16:23:20:WU00:FS00:0x15:
16:23:20:WU00:FS00:0x15:Assembly optimizations on if available.
16:23:20:WU00:FS00:0x15:Entering M.D.
16:23:22:WU00:FS00:0x15:Tpr hash 00/wudata_01.tpr: 2493796616 2046815114 2030391350 2713380791 2571875720
16:23:22:WU00:FS00:0x15:GPU device id=0
16:23:22:WU00:FS00:0x15:Working on Gallium Rubidium Oxygen Manganese Argon Carbon Silicon t= 158.00000
16:23:22:WU00:FS00:0x15:Client config unavailable.
16:23:22:WU00:FS00:0x15:Finished fah_main status=59
16:23:22:WU00:FS00:0x15:mdrun_gpu returned 59
16:23:22:WU00:FS00:0x15:GPU memtest failure
16:23:22:WU00:FS00:0x15:
16:23:22:WU00:FS00:0x15:Folding@home Core Shutdown: GPU_MEMTEST_ERROR
16:23:22:WARNING:WU00:FS00:FahCore returned: GPU_MEMTEST_ERROR (124 = 0x7c)
16:24:19:WU00:FS00:Starting
16:24:19:WU00:FS00:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/John/AppData/Roaming/FAHClient/cores/www.stanford.edu/~pande/Win32/AMD64/NVIDIA/Fermi/Core_15.fah/FahCore_15.exe -dir 00 -suffix 01 -version 703 -lifeline 4924 -checkpoint 15 -gpu 0 -gpu-vendor nvidia
16:24:19:WU00:FS00:Started FahCore on PID 7496
16:24:19:WU00:FS00:Core PID:9016
16:24:19:WU00:FS00:FahCore 0x15 started
16:24:20:WU00:FS00:0x15:
16:24:20:WU00:FS00:0x15:*------------------------------*
16:24:20:WU00:FS00:0x15:Folding@Home GPU Core
16:24:20:WU00:FS00:0x15:Version 2.25 (Wed May 9 17:03:01 EDT 2012)
16:24:20:WU00:FS00:0x15:Build host AmoebaRemote
16:24:20:WU00:FS00:0x15:Board Type NVIDIA/CUDA
16:24:20:WU00:FS00:0x15:Core 15
16:24:20:WU00:FS00:0x15:
16:24:20:WU00:FS00:0x15:Window's signal control handler registered.
16:24:20:WU00:FS00:0x15reparing to commence simulation
16:24:20:WU00:FS00:0x15:- Looking at optimizations...
16:24:20:WU00:FS00:0x15eleteFrameFiles: successfully deleted file=00/wudata_01.ckp
16:24:20:WU00:FS00:0x15:- Created dyn
16:24:20:WU00:FS00:0x15:- Files status OK
16:24:20:WU00:FS00:0x15:sizeof(CORE_PACKET_HDR) = 512 file=<>
16:24:20:WU00:FS00:0x15:- Expanded 58293 -> 257358 (decompressed 441.4 percent)
16:24:20:WU00:FS00:0x15:Called DecompressByteArray: compressed_data_size=58293 data_size=257358, decompressed_data_size=257358 diff=0
16:24:20:WU00:FS00:0x15:- Digital signature verified
16:24:20:WU00:FS00:0x15:
16:24:20:WU00:FS00:0x15roject: 8070 (Run 158, Clone 7, Gen 12)
16:24:20:WU00:FS00:0x15:
16:24:20:WU00:FS00:0x15:Assembly optimizations on if available.
16:24:20:WU00:FS00:0x15:Entering M.D.
16:24:22:WU00:FS00:0x15:Tpr hash 00/wudata_01.tpr: 2493796616 2046815114 2030391350 2713380791 2571875720
16:24:22:WU00:FS00:0x15:GPU device id=0
16:24:22:WU00:FS00:0x15:Working on Gallium Rubidium Oxygen Manganese Argon Carbon Silicon t= 158.00000
16:24:22:WU00:FS00:0x15:Client config unavailable.
16:24:22:WARNING:WU00:FS00:FahCore returned: GPU_MEMTEST_ERROR (124 = 0x7c)
16:25:19:WU00:FS00:Starting
16:25:19:WU00:FS00:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/John/AppData/Roaming/FAHClient/cores/www.stanford.edu/~pande/Win32/AMD64/NVIDIA/Fermi/Core_15.fah/FahCore_15.exe -dir 00 -suffix 01 -version 703 -lifeline 4924 -checkpoint 15 -gpu 0 -gpu-vendor nvidia
16:25:19:WU00:FS00:Started FahCore on PID 9048
16:25:19:WU00:FS00:Core PID:32
16:25:19:WU00:FS00:FahCore 0x15 started
16:25:20:WU00:FS00:0x15:
16:25:20:WU00:FS00:0x15:*------------------------------*
16:25:20:WU00:FS00:0x15:Folding@Home GPU Core
16:25:20:WU00:FS00:0x15:Version 2.25 (Wed May 9 17:03:01 EDT 2012)
16:25:20:WU00:FS00:0x15:Build host AmoebaRemote
16:25:20:WU00:FS00:0x15:Board Type NVIDIA/CUDA
16:25:20:WU00:FS00:0x15:Core 15
16:25:20:WU00:FS00:0x15:
16:25:20:WU00:FS00:0x15:Window's signal control handler registered.
16:25:20:WU00:FS00:0x15reparing to commence simulation
16:25:20:WU00:FS00:0x15:- Looking at optimizations...
16:25:20:WU00:FS00:0x15eleteFrameFiles: successfully deleted file=00/wudata_01.ckp
16:25:20:WU00:FS00:0x15:- Created dyn
16:25:20:WU00:FS00:0x15:- Files status OK
16:25:20:WU00:FS00:0x15:sizeof(CORE_PACKET_HDR) = 512 file=<>
16:25:20:WU00:FS00:0x15:- Expanded 58293 -> 257358 (decompressed 441.4 percent)
16:25:20:WU00:FS00:0x15:Called DecompressByteArray: compressed_data_size=58293 data_size=257358, decompressed_data_size=257358 diff=0
16:25:20:WU00:FS00:0x15:- Digital signature verified
16:25:20:WU00:FS00:0x15:
16:25:20:WU00:FS00:0x15roject: 8070 (Run 158, Clone 7, Gen 12)
16:25:20:WU00:FS00:0x15:
16:25:20:WU00:FS00:0x15:Assembly optimizations on if available.
16:25:20:WU00:FS00:0x15:Entering M.D.
16:25:22:WU00:FS00:0x15:Tpr hash 00/wudata_01.tpr: 2493796616 2046815114 2030391350 2713380791 2571875720
16:25:22:WU00:FS00:0x15:GPU device id=0
16:25:22:WU00:FS00:0x15:Working on Gallium Rubidium Oxygen Manganese Argon Carbon Silicon t= 158.00000
16:25:22:WU00:FS00:0x15:Client config unavailable.
16:25:22:WU00:FS00:0x15:Finished fah_main status=59
16:25:22:WU00:FS00:0x15:mdrun_gpu returned 59
16:25:22:WU00:FS00:0x15:GPU memtest failure
16:25:22:WU00:FS00:0x15:
16:25:22:WU00:FS00:0x15:Folding@home Core Shutdown: GPU_MEMTEST_ERROR
16:25:22:WARNING:WU00:FS00:FahCore returned: GPU_MEMTEST_ERROR (124 = 0x7c)
16:25:22:WARNING:WU00:FS00:Too many errors, failing
16:25:22:WU00:FS00:Sending unit results: id:00 state:SEND error:FAILED project:8070 run:158 clone:7 gen:12 core:0x15 unit:0x0000000d6652edb45122dfdaee1de7c0
16:25:22:WU00:FS00:Connecting to 171.67.108.36:8080
16:25:23:WU00:FS00:Server responded WORK_QUIT (404)
16:25:23:WARNING:WU00:FS00:Server did not like results, dumping
16:25:23:WU00:FS00:Cleaning up
16:40:00:WU01:FS01:0xa4:Completed 3500000 out of 25000000 steps (14%)
16:57:11:WU01:FS01:0xa4:Completed 3750000 out of 25000000 steps (15%)
17:14:23:WU01:FS01:0xa4:Completed 4000000 out of 25000000 steps (16%)
17:31:31:WU01:FS01:0xa4:Completed 4250000 out of 25000000 steps (17%)
17:48:41:WU01:FS01:0xa4:Completed 4500000 out of 25000000 steps (18%)
18:05:48:WU01:FS01:0xa4:Completed 4750000 out of 25000000 steps (19%)
18:22:48:WU01:FS01:0xa4:Completed 5000000 out of 25000000 steps (20%)
18:39:52:WU01:FS01:0xa4:Completed 5250000 out of 25000000 steps (21%)
i don't have a clue

 

[loft]

16:25:22:WU00:FS00:0x15:Folding@home Core Shutdown: GPU_MEMTEST_ERROR
16:25:22:WARNING:WU00:FS00:FahCore returned: GPU_MEMTEST_ERROR (124 = 0x7c)

There's your issue. Is it a factory overclocked videocard? Sometimes the factory OC'd cards are unstable.

 

[jfalbanese]

There's your issue. Is it a factory overclocked videocard? Sometimes the factory OC'd cards are unstable.

so i can't go in and dial it down? i guess.installing the new liquid cooler in a day or 2 . i should be able to get more bang for my buck on the cpu.

 

[jfalbanese]

went to the forum @ f@h. it say's it might be driver issue. downgraded to the last known working driver, and see how it goes.

 

[jfalbanese]

so far when i at idle seems to be working. nvidia users can download V309.67 drivers

 

[loft]

so i can't go in and dial it down? i guess.installing the new liquid cooler in a day or 2 . i should be able to get more bang for my buck on the cpu.

You can with msi afterburner or evga precisionx. I use afterburner on mine.

Sent from my SAMSUNG-SGH-I317 using Tapatalk 2

 

[jfalbanese]

You can with msi afterburner or evga precisionx. I use afterburner on mine.

Sent from my SAMSUNG-SGH-I317 using Tapatalk 2

definitely was the driver issue.folding with both when my comp goes into standby.

 

[Starlyte2]

I don't use Nvidia. I have never and never will use Windows. I use big FPU's from ATI. Theyare great. But would not work with your workings. Sorry.