![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
New GPU client - Open Beta
- Thread starter Rebel44
- Start date
rmdashrrootsplat
Gawd
- Joined
- Apr 2, 2008
- Messages
- 921
Anyone have some results from an integrated AMD 780G chipset? Is it supported/feasible? I know the numbers won't be that impressive but the cost to build one is. 
Anyone have some results from an integrated AMD 780G chipset? Is it supported/feasible? I know the numbers won't be that impressive but the cost to build one is.
http://foldingforum.org/viewtopic.php?f=10&t=1833&start=0&st=0&sk=t&sd=a&
I don't think 40 SPs is what they're looking for with this client.
rmdashrrootsplat
Gawd
- Joined
- Apr 2, 2008
- Messages
- 921
Yeah I read that, was hoping for someone else to be the guinea pig.
*SIGH* I looked at the original GPU option with the X1950XTX's about 2 years ago and just didn't see the overall benefit. As a gamer it makes it impossible to game AND run the GPU client full-time.
I have only read parts of the first few pages in this thread, but are you able to game without issue while running the new GPU client?
I have only read parts of the first few pages in this thread, but are you able to game without issue while running the new GPU client?
Theoretically, yes, but at least one person recommended a restart after gaming to prevent EUEs. Probably just a quirk from the beta.
Yeah I read that, was hoping for someone else to be the guinea pig.
I don't think there will be many guinea pigs with 40 SPs.
- Joined
- Mar 30, 2005
- Messages
- 2,205
Update: I let one WU run last night. It only took about 2 hours and it netted me 97 points. So IF that point/hour ratio would hold, I guess I would be looking at about 1200 points per day with a 3600+ X2 @ 2.4GHz and a stock 3870. Of course as I noted in my previous post my GPU held steady at about 45 - 50% utilization as reported in Rivatuner.
I will have more time to do some testing tomorrow.
[H]ugh_Freak - How did you get these averages? Just curious as I wouldnt mind figuring these as well for comparisons sake. Thanks!
I will have more time to do some testing tomorrow.
[H said:
[H]ugh_Freak - How did you get these averages? Just curious as I wouldnt mind figuring these as well for comparisons sake. Thanks!
Crazy Larry
[H]ard|Gawd
- Joined
- Oct 12, 2004
- Messages
- 1,137
[H]ugh_Freak;1032348719 said:---Q6600 @ 3.4GHz---
---3870 @ 875core/1322mem---
GPU2 client giving me over 1700PPD on project 2799
steps are taking on average 48 seconds
GPU activity is 69/83/78% min/max/average
What is the ppd difference running the SMP client vs. the SMP client and the GPU client?
- Joined
- Oct 12, 2004
- Messages
- 15,022
The key here is that the 4000 PPD is with 2 SMP clients. When you must free one core, it become impossible to run 2 SMP clients so you must add the efficiency loss of a single SMP + GPU. The net gain won't be like 700 PPD but likely around 200-300.
- Joined
- Oct 12, 2004
- Messages
- 15,022
This is why my stance still stand with the usefulness limited to a computer meant for gaming or HTPC. For a pure folding box, best to get a quad core and be done.
This is why my stance still stand with the usefulness limited to a computer meant for gaming or HTPC. For a pure folding box, best to get a quad core and be done.
You've done nothing but bash the new client through the entire thread. Always the pessimist and quite a bit premature.
97 point WUs are very small. Small chunks of data are proabably inefficient to parallize between 320 Stream Processors. They likely won't saturate a high end GPU, so you won't know the full performance of the GPU2 client until we get larger WUs.
And as always, this is an EARLY look at an early beta client. There is a lot of room for tweaking and performance improvements. Making any godlike decrees about how worthless the client will be is putting the cart before the moose.
- Joined
- Oct 12, 2004
- Messages
- 15,022
Lol, this is coming from someone who bash on those who like to tweak for the most PPD. My comment is also made with the power efficiency in mind.
Keep your opinion to yourself and let us decide with numbers, not ramblings.
Keep your opinion to yourself and let us decide with numbers, not ramblings.
Sunin, what is the GPU utilization on your AGP 3850 when you're running another 3D app like rthdribl? Does it hit 100%?
![[H]ugh_Freak](/data/avatars/m/65/65895.jpg?1580525148){kind=avatar}
- Joined
- Apr 30, 2003
- Messages
- 1,133
small copy of my logfile... roughly 48 seconds per %
[10:49:07] Completed 14%
[10:49:56] Completed 15%
[10:50:43] Completed 16%
[10:51:31] Completed 17%
[10:52:19] Completed 18%
[10:53:07] Completed 19%
[10:53:56] Completed 20%
[10:54:43] Completed 21%
[10:55:31] Completed 22%
[10:56:19] Completed 23%
[10:57:07] Completed 24%
[10:57:55] Completed 25%
100*48sec/60 = 80 minutes per WU
1440 min/day / 80min per WU = 18WUs/Day
18*97 = 1746 PpD
so around 1700PpD or so.. does take a few minutes to upload/download new WUs
Only about 200PpD in favor of one SMP client + GPU2 rather then 2 SMPs.. also.. GPU2 client seems hella more stable then the SMP client..
edit: and I forgot to get a Kill-A-Watt after work.. I'd forget my head if it wasn't attached
- Joined
- Oct 12, 2004
- Messages
- 15,022
Thanks for putting the numbers, this is in line with my prediction
It looks nice but probably not enough to go buy a lot of cards for the farm (7im, folding isn't about having the biggest and varied farm but to use the unused cycles. Read the F@H introduction !).
It looks nice but probably not enough to go buy a lot of cards for the farm (7im, folding isn't about having the biggest and varied farm but to use the unused cycles. Read the F@H introduction !).
The GPU client is also considerably more consistent in PPD because the variety of WUs is much smaller. So, that estimated 200PPD advantage could just as easily turn into a deficit dependent on the SMP WU mix.[H]ugh_Freak;1032351708 said:
- Joined
- Apr 30, 2003
- Messages
- 1,133
true it could turn into a deficit, or even larger.. my PpD figures are using project 2653 from the SMP client
- Joined
- Oct 12, 2004
- Messages
- 15,022
Don't blame the deficit on the GPU2 client tho. The problem is with some of the stupid WU actually.
Yes, I was comparing the variance of WU PPD with the SMP client. The GPU client is very predictable and I expect the GPU2 client to be much the same. We'll see when it develops a little more. So far it's looking surprisingly good.
- Joined
- Oct 12, 2004
- Messages
- 15,022
This mean there is some room for improvement...
Maybe something is choking somewhere with the GPU2 client.
Maybe something is choking somewhere with the GPU2 client.
You mean one GPU and one CPU client? That's probably the ideal setup with HT processors and a GPU client.
How do you see how far/fast it's doing the current WU? When I right-click on it and select Display, it says that "The new viewer is not yet packaged with this beta." And when I check the fahlog.txt, the file ends with:
How do you check the progress of the WU? I know it's running, since one of my cores is maxed out...
Code:
[21:29:29] Core found.
[21:29:29] Working on Unit 05 [April 11 21:29:29]
[21:29:29] + Working ...
Folding@Home Client Shutdown.Do you have two cards in your system?
Nope, but since it was only using 43% of my 1 gpu it had in essence had capacity
- Joined
- Mar 30, 2005
- Messages
- 2,205
Maybe I will set my CPU back to stock and see how much it affects the performance of my GPU. That might shed some more light on if the GPU2 client is somewhat CPU limited or if its due to the current work units.
I could run the SMP client with my X2 but I think the GPU2 client would net me more points. I was quite excited about the new GPU client merely for the fact that all these 500GFlop GPUs laying around could potentially provide a ton of processing power to be used for folding research.
I guess I'd rather run which client is going to provide more processing power to the Folding project not necessarily which client will get me more points.
[H]ugh_Freak - Thanks for the tip!
I could run the SMP client with my X2 but I think the GPU2 client would net me more points. I was quite excited about the new GPU client merely for the fact that all these 500GFlop GPUs laying around could potentially provide a ton of processing power to be used for folding research.
I guess I'd rather run which client is going to provide more processing power to the Folding project not necessarily which client will get me more points.
[H]ugh_Freak - Thanks for the tip!
- Joined
- Apr 30, 2003
- Messages
- 1,133
I could run the SMP client with my X2 but I think the GPU2 client would net me more points. I was quite excited about the new GPU client merely for the fact that all these 500GFlop GPUs laying around could potentially provide a ton of processing power to be used for folding research.
I guess I'd rather run which client is going to provide more processing power to the Folding project not necessarily which client will get me more points.
[H]ugh_Freak - Thanks for the tip!
My X2 4400 @ 2.6, (939flavor), only gives me around 980PpD running just the SMP. Using a GPU2 + standard console client might give the best numbers out of it.... your welcome
Using my Q6600 @3.4 only gives me ~80% gpu utilization so not sure how well you can get a P4 or similar to use more GPU.
It sure is fun to watch it fly through the WU in FahMon though. Every time it updates its gone up by 8 or 10%. Only takes it about an hour and 20 minutes total.
Indeed. That's what I wanted to see. There could be something to what 7im was saying with the complexity of the work units not being up to snuff yet. The GPU is bored right now with a P4 but that doesn't mean Stanford won't increase the complexity and take the CPU out of the equation.
Indeed. That's what I wanted to see. There could be something to what 7im was saying with the complexity of the work units not being up to snuff yet. The GPU is bored right now with a P4 but that doesn't mean Stanford won't increase the complexity and take the CPU out of the equation.
I was actually wonder if it has to do with the pre handling and post handling and the P4 not being able to load the gpu fast enough because other machines with AGP are getting over 1200PPD
