Folding@Home - #96276

Nvidia GPU client is out (BETA).
Have to use 174.55 drivers and 32bit Windows. It runs on all CUDA enabled Nvidia cards - list of cards with CUDA

this site is saying that you can use 177.35 drivers. Supposedly it runs faster with them.
"ForceWare 174.55 ~ 2500PPD
ForceWare 177.26 ~ 3500PPD
ForceWare 177.35 ~ 4500PPD"


This is pretty cool as well.
"Can I still use my GPU when the client is running?

Yes. Unlike the original GPU client, which interfered with many operations that used the GPU, the new GPU2 client does not. Playing videos and playing games either have no effect on the action of the GPU client other than a slow-down in processing, or cause a temporary suspension of folding. The new client will automatically back off whenever an application requests exclusive DirectX mode, although it is not reported in the client logfile. DirectX programs that do not request exclusive mode will cause the GPU client to slow down, and may in some instances have a detrimental effect on application performance. Full screen video is unaffected by the GPU client. "

(Current ATI cards also run GPU2 but the version for ATI cards. The 'GPU' version is no longer available)
 
I just tested it with my 8800GTS 512MB. It's an MSI card which is slightly OCed by default. I used FAHMON to monitor the work units and get an estimated PPD (points per day).

One unit was estimated to be going at the rate of 5000 ppd (meaning if I kept getting that more of the same unit that my points per day would be 5000).
Another unit I got was estimated at 2500 ppd.

My SMP client goes at ~1760 ppd. Most who have quad cores run 2 under vmware so that's 3520 ppd for a quad. To compare, if someone got a mix of both units for the GPU above then it would be 3750 ppd average. (2500+5000)/2

Only about 50% of one core was being used on my CPU.
 
wow...my 9600gt is folding this sucker really fast. also temperatures only raised like 3 degrees. cpu is running heavy though....why is that???

edit: nvm i was doing some other program, but yea it was running about 50% when folding.
 
I spilled water on my computer and it shorted out. ):
Running this on my very slow Celeron laptop...
 
Reading around forums, a lot of people are getting from 3500-5000 ppd which is equal to and more than running the SMP client on a quad. Although your also talking a lot more power used.

It seems the Nvidia Folding app is tied to the shader MHz and not really core or memory speeds. So you can underclock the core and memory to get lower temps. And if you want faster output folding then OC the shader MHz on the Nvidia GPU.
I spilled water on my computer and it shorted out. ):
That's too bad. :(
 
FYI
If you use FahMon to monitor your folding, to get it to see the GPU2 client you just point it to the folder under 'C:\Documents and Settings\USER\Application Data\Folding@home-gpu' rather than where you installed it to.
 
Is it just me or are the stats always unavailable? I constantly get that they are being updated and it should take 45 mins.

Anddd, i cant figure out how to access my own team's page that i just created.
 
I almost never go to the offical FAH stats page from Stanford.
I like checking out this site best for how they show the data but unfortunately our team isn't high enough for them to record individual user stats (team needs to be top 1000).
I use Kakao stats, what giant linked to, to view my own numbers.

FYI all my machines are back folding again. Hopefully HL2.net can hit 75k points per week.
It's hot in my room...
 
Still waiting on a new mainboard.
blankface.gif
 
heh, i dont see us getting over 1k anytime soon. but i run the gpu client everyday and knock out about 3-6 WU which are worth 479 each.
 
If everyone on the team ran the client for a week I bet we'd hit 1k very quickly. After I get my PC back online I should be able to double my output.
 
well we are at about 700,000 points as of now and the 1,000 ranked team has about 1,200,000 points. as of now we have 17k points per week. 7/23 active users;of course that doesn't count the computers which asus should be up and running soon. if we really did manage to pull at least 50k weekly then we should be ranked 1,000 in a month.
 
Ok, I have 5 clients going atm. And the AC is definitely staying on in this place. A GPU2, my PS3 and 3x SMP clients under linux through VMware.
HL2.net has put out 12k points so far in 24 hours.

I like running the Linux client for SMP better. It doesn't get stuck every time it wants to start a new WU and it always gets the quick 1760 pointers and not the 1900 ones. Plus it's a little faster since it's a 64bit client instead of 32bit like the Windows client.
 
Ugh, how do i find the team i made? I made a team for another website and cant for the life of me find out the number for it. :frown:

Sorry for the noob questions. :|
 
It's 96276. :p
OH, the other team... Forget about it. This thread is for HL2's team. :)

You don't get an email or anything when you create a team. They just...do it. It's all public with no user control. You have to know what it is to configure your client. And anyone can use any user name/team number when configuring their own client. I could change my user name on my FAH clients to 'giant384' if I wanted.
They have searches on the stat sites for team and user names but I'm not sure anything will come up if no WU's have been turned in under the team.
 
Got my MB. I'd be churning out some WUs right now but looks like my VC is not giving a picture to my LCD. >:(

Asus ATi 4850 time?
 
Unfortunately I have to put mine on hold for a few weeks. Going on a road trip D:
 
Got my MB. I'd be churning out some WUs right now but looks like my VC is not giving a picture to my LCD. >:(

Asus ATi 4850 time?
Hope ya get your stuff back together. What about 4850 time?

Unfortunately I have to put mine on hold for a few weeks. Going on a road trip D:
You'll be back though right? D:
And where are ya going for the road trip?
 
Flying to meet my friend in Colorado (he drove there a few days ago), driving down to Texas, visiting six flags and whatever else. Then we're driving across to California and making our way back up to BC finding stuff to do along the way. Mostly theme-parks...sight-seeing etc.

Leaving on the 9th. I should be back in about 3 weeks but I'll have a laptop with me :D
 
Hope ya get your stuff back together. What about 4850 time?

You'll be back though right? D:
And where are ya going for the road trip?

That's the card I'm going to get if it turns out mine is bad. It should really kill these proteins. Heck knows I don't need that much power for games.
 
ATM the 4800 series cards are not really being taken advantage of in the ATI GPU2 client. Might be a few versions until the work units start working them harder. Even last years 3800 cards are not 100% utilized with the in-house versions of the client.

5000 ppd - 8800GTS (512MB) or 9800GTX ($150-200)
4600 ppd - 8800GT (256mb and 512mb)
3000 ppd - 9600GT
2100 ppd - 4850 ($200)
1900 ppd - 3870
1500 ppd - 3850
1400 ppd - 8600GT

fyi as of June the GPU2 client got a point increase which covers both ATI and Nvidia clients. Base hardware for points, which is ATI's 3850, is currently set @ 1500 ppd. Was ~1k before, just keep that in mind when looking up numbers from forum posts.
Flying to meet my friend in Colorado (he drove there a few days ago), driving down to Texas, visiting six flags and whatever else. Then we're driving across to California and making our way back up to BC finding stuff to do along the way. Mostly theme-parks...sight-seeing etc.

Leaving on the 9th. I should be back in about 3 weeks but I'll have a laptop with me :D
3 weeks is a good amount of time. Lot of driving too.
w00t we are hitting the 75k mark we planned to hit.
@85k ppw :)

My AMD X2 4600+ system is running the SMP right now (~1000 ppd). But I just bought 2 8800GT 256MB cards to put in there and run two GPU2 clients instead of the SMP. Already have a board with 2 PCIexpress slots and a good enough PSU. $60 after rebate. That's a steal for 4.5k ppd each.
 
ATM the 4800 series cards are not really being taken advantage of ....................

Turns out the MB I bought is just defective, as was the one before it. I'm willing to bet the video card problems I have been having were not even stemming from the video card. Sure it was overheating, but after the last crash I found a huge wad of dust in the heatsink and removed it.
My previous failed MB: Asus P5B-E
My current defective MB: Asus P5K-E

Both have similar issues and 5 minutes of searching that I wish I did before buying reveals this is a common issue. Glad I didn't impulse buy that card. Think I'm going to stay away from Asus from now on...




No offense of course. :cheese:
 
Glad I didn't impulse buy that card. Think I'm going to stay away from Asus from now on...

Don't let the 4800 FAH situation turn ya off those cards for gaming. They are great. Although the GTX 260 did drop in price a lot recently...which runs FAH well. It just doesn't do quite as well in games.

No offense of course. :cheese:
Running ABIT, ASRock and MSI stuff atm. ;)
Although those cards I just ordered are Asus...
 
I just started. I had a heck of a time getting it to run on the older SMP clients, but I think this will do nicely.

edit - I just checked the team ranking... is this normal for HL2.net to have this increase? It's reassuring that we appear to be doing better than 4-chan.
 
this is the most points in a week this team has ever done, so that graph doesn't surprise me.
 
Assuming this is you?
It's funny how HardOCP is needing to recruit finally after getting a challenger and they are not #1 for PPD any more. Although they still have 3 years left for being #1 in total points.

I just started. I had a heck of a time getting it to run on the older SMP clients, but I think this will do nicely.

edit - I just checked the team ranking... is this normal for HL2.net to have this increase? It's reassuring that we appear to be doing better than 4-chan.
What do ya mean by 'older SMP clients'? Older FAH version or older CPUs? A little while ago I switched from running window SMP to running it virtualized in Linux. Meaning I run a VMware window and within that I run Linux with FAH running there. Often the Windows SMP version seem to freeze/crash when it got a new work unit. The Linux version doesn't freeze so there is no down time, it runs in 64bit (linux 64bit OS + 64bit CPU) = better points.

I think since you can now fold much better on Nvidia/ATI cards vs CPUs, gamers can get more points since they have the good cards. A lot of teams had a little increase with everyone trying out the Nvidia GPU2 beta.
 
Here is some info on how much performance vs watts used for FAH on my machines. PPD (points per day) per Watt.

A64 3000+ - 8800GTS (G92) - 2x1GB - ASRock K8NF6G-VSTA - Antec SU-430HT PSU (running XP)
124 Watts - Idle
205 Watts - GPU2 @ 4870 ppd **


Q9300 - 8800GTS (G92) - 2x2GB - Abit IP35 Pro - Corsair HX520 (running XP)
130 Watts - Idle
194 Watts - GPU2 @ 5000 PPD
170 Watts - VMware 2xSMP linux @ 3160 PPD *
216 Watts - GPU2 and VMware 2xSMP linux @ 7550 PPD ***

*18.6 ppd/watt for VMware and two 64bit SMP linux clients 100% on the quad.
**23.7 ppd/watt running the GPU2 client on my older CPU and an average PSU.
***35 ppd/watt when I moved the VMware over to NOT run on the 4th core and throwing a GPU2 client on there under WinXP.
 
hehe
I guess I do. Someone on another forum wanted to know so I did some quick testing. Thought I'd post it here too. They wanted to know so they could build a PC just for folding and what client/hardware would be the best for power usage and their electricity bill.
 
Should I be alarmed that I don't seem to be able to connect to their server? (see text below)
Code:
[01:17:06] Completed 145000 out of 500000 steps  (29 percent)
[01:32:56] Writing local files
[01:32:56] Completed 150000 out of 500000 steps  (30 percent)
[01:46:55] + Attempting to send results
[01:47:26] Couldn't send HTTP request to server (wininet)
[01:47:26] + Could not connect to Work Server (results)
[01:47:26]     (171.64.65.64:8080)
[01:47:26] - Error: Could not transmit unit 01 (completed July 10) to work server.
[01:47:26] + Attempting to send results
[01:47:58] Couldn't send HTTP request to server (wininet)
[01:47:58] + Could not connect to Work Server (results)
[01:47:58]     (171.64.122.86:8080)
[01:47:58]   Could not transmit unit 01 to Collection server; keeping in queue.
[01:48:20] Writing local files
[01:48:20] Completed 155000 out of 500000 steps  (31 percent)
I can ping their server, and am online as I type this, so the connection seems okay.
 
why is it sending results at 30%?

if it doesnt do anything after awhile then i would just delete your work folder in the FAH directory.
 
Seems like you just can't connect to the server atm. Just let it be. Looks like it is working on a new unit while trying to turn the old one in at the same time.
 
Since the SMP client needs to run on a Windows account with a password, I had to create a new one just for this, and I just switch users (so they're both logged in). Perhaps if the SMP account isn't the 'active' one (even though it's logged in and running), it doesn't have access to the Internet connection?
 
Back
Top