New larger WU (4742-4747 1254 atoms)

[pwnchu]

Are you guys getting better ppd or worse, and on what cards?
My 3650 is getting shafted over 20% ppd on the big ones :/
I'm finding p4740 to be excellent(1175ppd), and the 20s-30s to be ok (900-1000ppd), but these new ones are killer (770ppd)

[bruce]

Are you guys getting better ppd or worse, and on what cards?
My 3650 is getting shafted over 20% ppd on the big ones :/
I'm finding p4740 to be excellent(1175ppd), and the 20s-30s to be ok (900-1000ppd), but these new ones are killer (770ppd)

That's not surprising. The benchmark machine has a 3850 in it and it has 320 shaders whereas your 3650 only has 120. The small proteins didn't need all of the features of the mid-range cards or make effective use of their hardware so you've been getting a bonus. These larger proteins now use those extra shaders and everything below the 3850 is now earning a lower, more appropriate PPD. Sorry. This same trend will be pushing the 3850 harder so that the higher end cards will also start to pull away from the pack, as well.

[pwnchu]

Ah, I see. Hopefully, then, if small proteins will continue to be processed we will be able to choose which projects to crunch so that each gpu gets a fitting wu. Points dont bother me much; I'm just hoping my card is being used efficiently.

[bruce]

Ah, I see. Hopefully, then, if small proteins will continue to be processed we will be able to choose which projects to crunch so that each gpu gets a fitting wu. Points dont bother me much; I'm just hoping my card is being used efficiently.

The lower level card were being used efficiently with either small or large proteins. The change was in the mid- to high-range cards which will now be used more effectively. Because the efficiency of the benchmark has improved, your points will be lower but that doesn't mean the efficiency of your card changed. Like I said, you were getting a bonus because of the inefficiency of the benchmark hardware and now you will no longer get that bonus.

It will be interesting to see what happens when a similar change is made to the proteins being distributed to nVidia cards.

[fridgepants]

Are you guys getting better ppd or worse, and on what cards?
My 3650 is getting shafted over 20% ppd on the big ones :/
I'm finding p4740 to be excellent(1175ppd), and the 20s-30s to be ok (900-1000ppd), but these new ones are killer (770ppd)

I am running a 4870. With the new WU's (42 through 47 or so) I am getting a whopping 3600ppd. With the older ones (20s-30s) I am averaging about 2200ppd. A few of the others I've done are closer to 2600.

[EvilAlchemist]

I am running a 4870. With the new WU's (42 through 47 or so) I am getting a whopping 3600ppd. With the older ones (20s-30s) I am averaging about 2200ppd. A few of the others I've done are closer to 2600.

One of the good reasons the 48xx series will continue to shine.
They have an edge over the benchmark machine ... having the extra SP's .... so larger work units will continue to get more ppd .. or so we hope!

[dittopb]

.... !My friend still uses my old 3200+ to play wow, i think he needs a new video card LMAO

Tell your friend to try AoC, he will more likely to buy a faster video card.
BTW good job for the new WU. ATI donors need a little lovin after reading this article: http://www.tomshardware.com/news/Nvidia ... ,6254.html
ATI comes in third after PS3
Mhouston is our only hope. Please don't go to the dark side (Nvidia).

[fridgepants]

ATI comes in third after PS3

Yes and no. You're just comparing total TFLOPS which is redialy explained by the fact that there are far more nVidia clients operating than ATI clients. What is revealing is the TFLOPS/Client number which (strangely) is identical for the ATI and nVidia clients.

ATI: 404 tflops / 3677 Clients = .10987 tflops per client.
nVidia: 1428 tflops / 12982 Clients = .10999 tflops per client

It sure strikes me that PPD for the two brands should be more comparable. Perhaps there's data missing though (like science per tflop per brand).

[texaslabrat]

ATI comes in third after PS3

Yes and no. You're just comparing total TFLOPS which is redialy explained by the fact that there are far more nVidia clients operating than ATI clients. What is revealing is the TFLOPS/Client number which (strangely) is identical for the ATI and nVidia clients.

ATI: 404 tflops / 3677 Clients = .10987 tflops per client.
nVidia: 1428 tflops / 12982 Clients = .10999 tflops per client

It sure strikes me that PPD for the two brands should be more comparable. Perhaps there's data missing though (like science per tflop per brand).

The PPD difference comes from the huge differences in the dependency on the cpu between the two. If you had a liquid-nitrogen-cooled 8Ghz cpu, you could likely match/exceed the nvidia cards' PPD's with a 4870 on the smaller WU's. As the ati cores become less and less cpu-bound through mhouston's work, you'll see that the new WU's being issued for Nvidia will produce smaller and smaller PPD relative to the science being done since the benchmark machine will be doing more ns/day of simulation when the initial PPD mark is set with the more efficient core available at the time. At the end of the day, I believe we'll wind up seeing the PPD numbers matching what our expectations would be given the raw TFLOPS available and the observed efficiency rates of shader use. The Nvidia crowd should soak in the points while they can, because once the larger proteins start being crunched by them, their PPD will likely decrease drastically from what they are accustomed to now just by the fact that the ati client has become more efficient and is the benchmark reference by which everything else is compared to.

At the point where a given WU can be run on the benchmark machine in a non-cpu-bound condition is the point where a true apples-to-apples comparison between the two camps' gpu hardware (and stream computing software stack) can be made.

[deekey777]

Silly question: Why do I get no larger WUs?
http://fah-web.stanford.edu/cgi-bin/mai ... range=4000

Since tuesday I get small WUs only. So sad...

[nibbs]

Same here. Guess they are done "testing" larger WUs and now checking the results or just stopped over a bug they've found to solve before continuing to distribute those WU's.

[Viper666]

ATI comes in third after PS3

I never made that comment, how Is it someone quoted me, look back threw the thread WTF !
Actually IBM's cell comes in first, I guess everyone missed all the PS3 teams passing you by and the stats with in a week, of the ps3 ,it had out produced 5 years on Pc's. It does 32bit single p. also ! Whatever doesnt matter. Im not sure they get all of the cell.It maybe only %50 used.

[uncle_fungus]

ATI comes in third after PS3

I never made that comment, how Is it someone quoted me, look back threw the thread WTF !

It just some haphazard use of the quote tag. I'll fix it in the above posts.

[Foxbat]

Silly question: Why do I get no larger WUs?
http://fah-web.stanford.edu/cgi-bin/mai ... range=4000

Since tuesday I get small WUs only. So sad...

I picked one up (P4743R0C91G22) after midnight EDT, so they are still out there.

It certainly seems that the large atom WUs are in short supply. I'm lucky if I see one in a 24 hour period. My rig needs these LAWs to keep its HD3850 at 99%.

[SuperBORG]

Oh come on, life is so not fair. I've been folding for some time now. I've never come across these 474x WU. And when i get luck....

[05:22:34] Project: 4747 (Run 0, Clone 63, Gen 18)
[05:22:34]
[05:22:34] Assembly optimizations on if available.
[05:22:34] Entering M.D.
[05:22:40] Working on p4747_lam5w_300K
[05:22:40] Client config found, loading data.
[05:22:40] Starting GUI Server
[05:22:44] mdrun_gpu returned -1
[05:22:44] Going to send back what have done.
[05:22:45] logfile size: 0 info=0 bed=0 hdr=25
[05:22:45] - Writing 558 bytes of core data to disk...
[05:22:45] Done: 46 -> 85 (compressed to 184.7 percent)
[05:22:45] ... Done.
[05:22:45]
[05:22:45] Folding@home Core Shutdown: EARLY_UNIT_END
[05:22:48] CoreStatus = 72 (114)
[05:22:48] Sending work to server
[05:22:48] Project: 4747 (Run 0, Clone 63, Gen 18)


[05:22:48] + Attempting to send results [September 3 05:22:48 UTC]
[05:22:48] + Results successfully sent
[05:22:48] Thank you for your contribution to Folding@Home.
[05:22:52] - Preparing to get new work unit...
[05:22:52] + Attempting to get work packet
[05:22:52] - Connecting to assignment server
[05:22:53] - Successful: assigned to (171.64.65.103).
[05:22:53] + News From Folding@Home: GPU folding beta
[05:22:53] Loaded queue successfully.
[05:22:53] + Closed connections
[05:22:58]
[05:22:58] + Processing work unit
[05:22:58] Core required: FahCore_11.exe
[05:22:58] Core found.
[05:22:58] Working on queue slot 06 [September 3 05:22:58 UTC]
[05:22:58] + Working ...
[05:22:58]
[05:22:58] *------------------------------*
[05:22:58] Folding@Home GPU Core - Beta
[05:22:58] Version 1.10 (Tue Aug 12 10:03:11 PDT 2008)
[05:22:58]
[05:22:58] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[05:22:58] Build host: amoeba
[05:22:58] Board Type: AMD
[05:22:58] Core :
[05:22:58] Preparing to commence simulation
[05:22:58] - Looking at optimizations...
[05:22:58] - Created dyn
[05:22:58] - Files status OK
[05:22:58] - Expanded 45379 -> 246265 (decompressed 542.6 percent)
[05:22:58] Called DecompressByteArray: compressed_data_size=45379 data_size=246265, decompressed_data_size=246265 diff=0
[05:22:58] - Digital signature verified
[05:22:58]
[05:22:58] Project: 4715 (Run 8, Clone 10, Gen 58)
[05:22:58]
[05:22:58] Assembly optimizations on if available.
[05:22:58] Entering M.D.
[05:23:04] Working on p4715_supervillin_e1
[05:23:05] Client config found, loading data.
[05:23:05] Starting GUI Server
[05:24:13] Completed 1%
[05:25:12] Completed 2%
[05:26:05] Completed 3%
[05:26:56] Completed 4%
[05:27:47] Completed 5%
[05:28:38] Completed 6%
[05:29:29] Completed 7%