Optimal CPU - GPU comparison table [moderate to larger WUs]

[X1900AIW]

(no change in system configuration) > Influence of the new BIG WU in Dual-GPU-configuration (again:) I have no need for them (with HD3850).
2x HD3850/512MB
gpu 0 @796 / 999 OC / gpu 1 @796 OC (PCIe x16 / x4)

GPU usage 99% / 45%
Project 4742 / 4715
2332 PPD / 2451 PPD (= 4783 PPD)

GPU usage 80% / 99% (no more need for OC with small WU in "gpu 0")
Project 4715 / 4715
2395 PPD / 2451 PPD (= 4846 PPD)

/ /

Edit1: Another Big WU: 4743
(gpu 0 @796 / 999 OC / gpu 1 @796 OC - no change)
> each Big Wu is "grabbing performance", second card drops in usage-%, regardless which PCIe-slot

GPU usage 30-35% / 99%
Project 4727 / 4743
1863 PPD / 2113 PPD (= 3977 PPD)



Edit2: normal WUs 2x 4729 (no more memory OC)
gpu 0 @796 OC / 829 stock
GPU usage 75-80% (+7 degree Celsius)
4729 (Run 3, Clone 182, Gen 1)
2150 PPD

gpu 1 @796 OC / 829 stock
GPU usage 90-95%
4729 (Run 3, Clone 193, Gen 0)
2150 PPD

> exactly identic performance at different PCIe-lines

[legoman666]

http://www.newegg.com/Product/Product.a ... 6814121082

HD2600XT for $19.99 isn't a bad deal for a low budget folding machine. Averages ~ 1000PPD.

[legoman666]

Ponies!

ftp://legoman666.dyndns.org:47024/scores_ati.zip

[jrweiss]

HD3850
Core 1.10

Proj 4727
1975 PPD

Proj 4735
2297 PPD
[Edit: Added new WU]

Proj 4745
2143 PPD
[Edit: Added yet another]

Proj 4724
1799 PPD
[Edit: Added yet another]

Proj 4742
2133 PPD
[Edit: Added yet another]

Proj 4743
2132 PPD
[Edit: Added yet another]

[SKeptical_Thinker]

HD3200 (IGP@950MHz, 500 stock)
P4740
533PPD
Core 1.10
Catalyst 8.6

[almcdsmi]

Been seeing some improvement lately with 1.10 core

C2D E4400 @ 3.2ghz
Windows XP SP3
Catalyst 8.8
Client 6.20 System tray
core 1.10

HD3850 @ 770/1008

Project 4715
PPD 2922

Project 4727
PPD 2300

Project 4731
PPD 2311

Project 4738
PPD 2802

Project 4742
PPD 2240

Project 4743
PPD 2240

[pwnchu]

ati 3650 sapphire 512mb ddr3 (750gpu/825memory), 8.7 driver
AthlonX2 4600+ (2.41ghz )
6.20 console/v1.10

p4728-911ppd
p4730-999ppd
p4740-1175ppd
p4742-774ppd
p4744-774ppd
p4746-775ppd
p4747-771ppd

edit: p4737:1168ppd
p4715:1086ppd
p4724:884ppd

[legoman666]

HD4850
p4738: 2246
p4739: 2297
p4740: 2084
p4741: 2197
p4744: 3156
p4745: 3135
p4746: 3156

hd3870
p4742: 2265
p4743: 2223
p4744: 2265
p4746: 2276

Cat 8.8
core 1.10
client 6.20
Q6600 @ 3.1

[devpao]

Athlon 64 X2 2GHz win XP
HD3850 catalyst 8.8

p4744 2142 (99%)
p4731 1016 (61%)
p4724 1200 (62%)

[Karamiekos]

HD4850 @ 690/1125
p4739: 2128
p4740: 2063
p4741: 2001
p4742: 3507
p4743: 3507
p4744: 3507
p4745: 3430
p4746: 3456
p4747: 3507

Core 1.10
Cat 8.8
Windoze XP SP3
X2 3800+ @ 2.5Ghz

Comparing mine to LegoMan's shows were the better CPU helps, and where the OverClock kicks in. (Nice CPU BTW Lego.)

[GenTarkin]

I cant recall what project number off the top of my head, but its one of the higher 4000 ones.
I just got my 4850 and coupled with an opteron 165 @ 2.8ghz Im doing nearly 4k ppd with the card.
Its like 3950PPD...
The card is clocked @ 685 / 1100 right now.
Amazing improvement over my 3850 that was clocked @ 864 / 1030 =)

[ashleylist]

Any chance you could setup one of these summary tables for CPUs for the SMP client/graphical client?

Ash

[Samsonic]

Hi fellow folders i've got some new stats for my machine

Cpu: e8400 @ 3600 mhz
fsb: 1600 mhz
Gpu: Ati 3870 @ 825 mhz core & 2250 mhz mem
Gpu driver vers.: 8.8 catalyst
Gpu @ 99 %
Client version: 6.20
Os: Win Xp Sp3

Project: 4715 - Ppd: 3194

Project: 4724 - Ppd: 2317

Project: 4728 - Ppd: 2513

Project: 4729 - Ppd: 2932

Project: 4737 - Ppd: 3017

Project: 4741 - Ppd: 3063

Project: 4744 - Ppd: 2391

Project: 4745 - Ppd: 2391

Project: 4746 - Ppd: 2391

Interesting to see that the new projects generate the same amount of points.

I also have a question for mr Pande, i would like to know what is being done with the enourmous amount processing power, i mean the folding at home supercomputer is about 2.4 times faster then IBM's roadrunner S.C.. When will there be more results released, are we getting closer in eradicating those horribul diseases like cancer and parkinsons etc. I'm really anxious to know these things, i know that the data that you eceive from us folders is ginormous but i really want some more info and i think i'm not the only one. Keep up the good work.

P.s. Are u also working with any Dutch Universities ?

Folding 4 ever......

[bruce]

Any chance you could setup one of these summary tables for CPUs for the SMP client/graphical client?

Ash

Why don't you use http://fahinfo.org/? It allows everybody to report times with various projects on their own hardware and there are some excellent tools to summarize the results.

[X1900AIW]

Q6600@3,0GHz, (2x GPU2 @all cores, 3x BOINC @all cores, without BOINC the cards use 30% of all cores)
WindowsXP/SP3, Catalyst 8.6, Core Version 1.10
Edit: Catalyst 8.8, BOINC deactivated
Edit2: not finished yet, 3x BOINC activated meanwhile
Edit3: Project-mix again, surprising performance, incl. 4x BOINC
2x HD3850/512MB
gpu 0 @stock / gpu 1 @stock (PCIe x16 / x4)
gpu 0 @stock / gpu 1 @796 OC because of 99% without OC @PCIe x4
GPU usage fluctuating 30-80%, mostly between 50-70% (> no need for OC)
GPU usage 94% / 99%
GPU usage 70-99% / 90-99% - fluctuating again
GPU usage 85-95% - both GPUs the same range, fluctuating up to 99% >gpu 0 will be oced too

Project 4746 / 4742 (first time two big WUs with 1254 atoms)
Project 4742 / 4742 (Run 4, Clone 12, Gen 9) vs. (Run 9, Clone 45, Gen 8)
1083 PPD / 1049 PPD
1220 PPD / 1236 PPD
1461 PPD / 1417 PPD - increasing trend ...
2309 PPD / 1345 PPD Trying to understand this strange balancing ...

Can´t be happy with ~2100 PPD using two cards ...
Sum about ~2900 PPD sounds much better.
Can be seriously happy with ~3600 PPD increasing trend similar, especially @PCIe x4; update (50% done, L3F): 2113 PPD / 2113 PPD = 4227 PPD

Descriptions in the project summary are not updated, see 4742-4746 "Points and deadlines".