Who wants to cure disease for team 36362?

[Sparky]

In fahmon did you add the client? Right click on the empty white box on the left side and add new client, put in the name, and point it to where the GPU2 client is installed.
*EDIT* mike beat me. Oh you don't point it to the install folder? OK shows you what I know, all the others I've always pointed to the install folder ignore me

[MikeB12]

yeah gpu2 is different.. it confused me too when I tried it on my 2900 a couple months ago.. it's in the profile in apps I believe, which is hidden, so you have to show hidden folders if it's not already chesked.

[gallardo]

Wait, what? I'm doing 89 seconds per frame while coo-coo-clocker is doing 97 and he gets 4k ppds while I'm getting 1800? How does that work?

Oh, and... how is a 3850 faster than an 8800GTX? Bad drivers for nVidia or what?

[MikeB12]

why don't you config fahmon like we asked... then you can compare wu's and get accurate readings without asking people to dig thru your logs..


here's your other wu in that log. this is the last manual calc I'm doing for you. you know how to do this now from the formula above.
config fahmon, it's the tool you need to use for ppd calcs.....

wu 4727 = 192 points = 89sec/wu = 1863.89ppd


coo coo clocker is on a nvidia wu 5004 which is 479 points per wu. that's why he's getting more ppd with greater frame times..
your ati 4731 wu is a 193 point wu and 4727 is a 192 point wu...

so for every 5004wu he finishes he gets 479 points.
every 4731 or 4727 wu you finish, you get 192/3 points...

the 8800gtx is not slower, the wu is more intensive to process. the frames and frame times are not comparable..

[gallardo]

why don't you config fahmon like we asked... then you can compare wu's and get accurate readings without asking people to dig thru your logs...

Cause I'm at work right now. Will do so when I get home.
Didn't really ask you to calculate another wu for me... just wanted to know why I got less points for finishing faster. And now I got it. Different wu give different points and have different levels of difficulty. Got it. Thanks!

[MikeB12]

cool, sorry if I sounded harsh.. I just reread my reply....
btw you can see all the wu's and points values here (look for GROGPU2)
http://fah-web.stanford.edu/psummary.html

[gallardo]

Prefered and deadline days? What happens if we miss the deadline?

[MikeB12]

you shouldn't miss them with these gpu2 wu's.. they turn too quick...
if you miss the final, you dont get credit... you won't miss it though, unless you stop the client and restart it a couple days later on the same wu dl.
preferred is just what it says it is... preferred, no points loss if you miss it.

[Sparky]

Sounds like mike has you covered

[Sir_Loin]

you shouldn't miss them with these gpu2 wu's.. they turn too quick...
if you miss the final, you dont get credit... you won't miss it though, unless you stop the client and restart it a couple days later on the same wu dl.
preferred is just what it says it is... preferred, no points loss if you miss it.

I don't know why they even show a preferred deadline. Until they start handing out bonus credits for high performance factors, I could care less about the preferred deadline.

[MikeB12]

yeah I asked this ques last year at fah forums, and the response was..

if the preferred deadline is passed, then the wu goes out again... so the work gets duplicated by another machine...
so if everyone turned in wu's bt preferred and final, then all wu's would get sent out twice..

but that's not our fault as users, that's fah's policy. we can't really do anything about it except try to meet the preferred deadline.

this actually doesn't really affect the gpu2's, but it did the smp's. because many slower dual cores barely made the preferred deadline. in fact I was folding on a P4 HT that barely made the final deadline for a while, which means every wu that machine did created duplicate work... but I got points for it anyway, since it met the final deadline.

[gallardo]

Ok. Got home, got FahMon working and your calculation was spot on. It shows 1852 ppds.
One more question (if I'm starting to bug you, let somebody else answer it ):
Is the ppd count constant or does that depend on the project as well? Are the points for the project distributed in such a way that the ppd ends up the same for the same machine on different projects?
Cause.... 1800 ppds seems a bit low to me. I was expecting something closer to 10k.

[Sparky]

It is very dependant on the project. On the old standard CPU client I would get only like 60PPD with one project but like 300PPD on another. Current batch of units out there it is at about 160PPD.

Not to burst your bubble, but 10K PPD is on the high side for a single GPU.

[MikeB12]

I hate to bring this up because it's a sticky subject... but let's not forget there's some fah favoritism to nvidia right now. hopefully that will change and mhouston will come through on the promise to optimize ati gpu folding. but I'm not holding my breath...

that's really ashame with the history and dedication ati has shown to fah for so long.. but watcha gonna do, as they say... it is what it is...


edit: oh yeah, forgot to answer your ques... fah forum has some good ppd threads from users reporting..

ati's http://foldingforum.org/viewtopic.php?f=42&t=2522

nvidia http://foldingforum.org/viewtopic.php?f=43&t=3193

[Combatspace]

Now folding for team Xtreme!!! I feel part of a greater cause *sniffles*

[MaK2000]

Make sure you have your team number right. I can only find you under EOC team. Maybe you haven't got a WU in yet under 36362. Glad to have you!

[Combatspace]

Afraid not, got about 1k more frames on an amber core. Estimated to end at 3:30 pm tomorrow...

[dkjbama]

Joining up with a hd2900xt and q6600 @ 3.2

[Kingcarcas]

Curious how much PPD the 2900s get...

[dkjbama]

Curious how much PPD the 2900s get...

PPD?

I'm not sure for 2 reasons. First I can't get the card to kick up to 3d speeds. Neither atitool, atitraytools, nor rivatuner can adjust the speeds. Second, the folding display where it gives the stats,that slows the folding process down. From like 1000-1500 Iterations per sec down to about 300-400, but its the only place you get the stats.

I just started gpu folding, been on WCG since dec of last year and GIMPS (Prime95) for about 2-3 years before that.

[Marvin_The_Martian]

PPD?

I'm not sure for 2 reasons. First I can't get the card to kick up to 3d speeds. Neither atitool, atitraytools, nor rivatuner can adjust the speeds. Second, the folding display where it gives the stats,that slows the folding process down. From like 1000-1500 Iterations per sec down to about 300-400, but its the only place you get the stats.

I just started gpu folding, been on WCG since dec of last year and GIMPS (Prime95) for about 2-3 years before that.

Use Fahmon to monitor the client, it will not cause the iterations p/s slowdown. If the card is not folding on the right clocks post your driver version, posting that and a fahlog showing client version and core revision might help to. Thanks for participating in f@h

[Jamesrt2004]

meme just joined (y)

[Polizei]

[Marvin_The_Martian]

+1

Like the signature, happy folding

[dkjbama]

Use Fahmon to monitor the client, it will not cause the iterations p/s slowdown. If the card is not folding on the right clocks post your driver version, posting that and a fahlog showing client version and core revision might help to. Thanks for participating in f@h

At 2D speeds (503mhz) I'm getting PPD of 1194.75. Does PPD=parts per day?

I'm now using catalyst 8.6 and the speeds stay the same.

Here is the fahlog:

--- Opening Log file [July 21 00:25:04]


# Windows Console GPU Edition
Folding@Home Client Version 6.12beta8

http://folding.stanford.edu

[00:25:04] - Ask before connecting: No
[00:25:04] - User name: dkjbama (Team 36362)
[00:25:04] - User ID: 1505483C30D69AD3
[00:25:04] - Machine ID: 2
[00:25:04]
[00:25:05] Loaded queue successfully.
[00:25:05] Initialization complete
[00:25:05] + Benchmarking ...
[00:25:05]
[00:25:05] + Processing work unit
[00:25:05] Core required: FahCore_11.exe
[00:25:05] Core found.
[00:25:05] Working on queue slot 05 [July 21 00:25:05]
[00:25:05] + Working ...
[00:25:06]
[00:25:06] *------------------------------*
[00:25:06] Folding@Home GPU Core - Beta
[00:25:06] Version 1.03 (Tue Jun 3 11:31:16 PDT 2008)
[00:25:06]
[00:25:06] Compiler :
[00:25:06] Build host: amoeba
[00:25:06] Preparing to commence simulation
[00:25:06] - Looking at optimizations...
[00:25:06] - Files status OK
[00:25:06] - Expanded 41117 -> 231745 (decompressed 563.6 percent)
[00:25:06] Called DecompressByteArray: compressed_data_size=41117 data_size=231745, decompressed_data_size=231745 diff=0
[00:25:06] - Digital signature verified
[00:25:06]
[00:25:06] Project: 4731 (Run 2, Clone 20, Gen 119)
[00:25:06]
[00:25:06] Assembly optimizations on if available.
[00:25:06] Entering M.D.
[00:25:12] Will resume from checkpoint file
[00:25:12] Working on 544 p4731_fip35_ww_domain
[00:25:12] Client config found, loading data.
[00:25:12] Starting GUI Server
[00:25:18] Resuming from checkpoint
[00:25:18] Verified work/wudata_05.log
[00:25:18] Verified work/wudata_05.edr
[00:25:19] Verified work/wudata_05.trr
[00:25:19] Verified work/wudata_05.xtc