NVIDIA 6.12 beta 6 GPU client Released!

Come join the... uh... er... fold.

Moderators: just brew it!, farmpuma

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Thu Jun 26, 2008 8:15 am

i'm running on 64bit already. you just needed a modified inf file for the x64 driver.
mongoosesRawesome
Gerbil XP
Silver subscriber
 
 
Posts: 366
Joined: Mon May 10, 2004 5:36 pm
Location: Maryland, USA

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Thu Jun 26, 2008 9:15 am

mongoosesRawesome wrote:i'm running on 64bit already. you just needed a modified inf file for the x64 driver.


Only problem is that sometimes depending on the hardware you can get some instabilities. But for the most part it seems to run fine on 64-bit OS's. You can find the modded inf's at the folding at home forums usually.
CB5000
Gerbil XP
 
Posts: 365
Joined: Wed Mar 26, 2008 4:46 pm
Location: NW region

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Fri Jun 27, 2008 12:24 am

themattman wrote:Has there been any news on a version for those of us who run 64-bit operating systems?


Well, I've been running a 64bit OS throughout. I was running XP64 on my 8800GT and am running Vista64 and beta8 on my new 4850.

64 isn't 'supported', but it sure does work. Pande is just being timid/lazy WRT what works versus what they want to support.
Siglessness is boring.
Image - M4800-Eight1
Image - Vargr-Z97
Forge
Lord High Gerbil
 
Posts: 8010
Joined: Wed Dec 26, 2001 7:00 pm
Location: SouthEast PA

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Fri Jun 27, 2008 7:54 am

I have to say, this GPU client has upped my interest in folding again. It allowed my to be a bit competitive where my single machine was lagging way behind and getting passed up. Now with 2 GPU clients running on 2 machines along with CPU clients, it feels nice to pass 45 people in one day. What took me more than a year to hit 100K, will take me less than 7 days to get another 50K to get to 200K. Amazing change in PPD for me.

What is even more impressive is the TR's point output took a nice jump when several of us started running the GPU client.

What is SAD is how many people on TR who have dropped out of the race. We need to promote these largely unused GPU cycles!

Only drawback? No control for lowering the % of GPU time the client needs, and that really slows my GUI of Windows down a lot. I have to pause it while I am working.

-LS
liquidsquid
Minister of Gerbil Affairs
 
Posts: 2447
Joined: Wed May 29, 2002 10:49 am
Location: New York

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Sun Jun 29, 2008 8:44 pm

Lockheed_Tvr wrote:Ok! 177 drivers on an 8400GS:

Big improvement! Jumped from around 450PPD to about 560PPD both on the 98 point WUs. Good enough to scrap SMP, add a console client on one core and get a nice bump in points!

May I know what project that 450/560ppd came from? I have got some new data myself and I would like to compare notes, if it is at all possible.
Image
The Model M is not for the faint of heart. You either like them or hate them.

Gerbils unite! Fold for UnitedGerbilNation, team 2630.
Flying Fox
Gerbil God
 
Posts: 24357
Joined: Mon May 24, 2004 2:19 am

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 12:47 pm

Flying Fox wrote:
Lockheed_Tvr wrote:Ok! 177 drivers on an 8400GS:

Big improvement! Jumped from around 450PPD to about 560PPD both on the 98 point WUs. Good enough to scrap SMP, add a console client on one core and get a nice bump in points!

May I know what project that 450/560ppd came from? I have got some new data myself and I would like to compare notes, if it is at all possible.

I've got an 8800GT and an 8400GS both running (in separate boxes) on the p5002_supervillin_e1 project. Each percentage point on the 8800GT is taking around 83 seconds, whereas on the 8400GS it's taking around 215 seconds. I also have a 9500GS running in a third box and it's turning a percentage point every 150 seconds.

I'm not sure what this translates to in PPD as I"ve only been running these configs for a couple of days and have been futzing with the setups every so often. Also, I get a brain cramp when I try to figure out what's happening on a ppd/WU basis.
This problem was caused by Windows, which was created by Microsoft Corporation.
sluggo
Gerbil Jedi
Gold subscriber
 
 
Posts: 1540
Joined: Wed Feb 16, 2005 8:44 pm
Location: under the table and dreaming

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 12:53 pm

450 ppd is very low for a 8800GT. I pull 3k ppd with a 9600GT, and I will leave it running while playing games and watching videos. The new core (1.06) really makes a difference.
Sony a7
Sony Zeiss 55/1.8 SSM, 24-70/4 SSM
Minolta 17-35/2.8-4 D, 100-300 APO
TheEmrys
Minister of Gerbil Affairs
Silver subscriber
 
 
Posts: 2150
Joined: Wed May 29, 2002 8:22 pm
Location: Northern Colorado

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 1:02 pm

sluggo wrote:I've got an 8800GT and an 8400GS both running (in separate boxes) on the p5002_supervillin_e1 project. Each percentage point on the 8800GT is taking around 83 seconds, whereas on the 8400GS it's taking around 215 seconds. I also have a 9500GS running in a third box and it's turning a percentage point every 150 seconds.
Based on Stanford's project summary page, looks like you are running Project 5002 with a point value of 98. A bit of rough calculations (correct me if I'm wrong):

8400GS: 393.8ppd
9500GS: 564.5ppd (assuming it is running on the same project)
8800GT: 1020.1ppd

Looks like you are working on some sucky WUs. :cry:

sluggo wrote:I'm not sure what this translates to in PPD as I"ve only been running these configs for a couple of days and have been futzing with the setups every so often. Also, I get a brain cramp when I try to figure out what's happening on a ppd/WU basis.
You can get FahMon/FahSpy to do those ppd calculations for you. ;)

TheEmrys wrote:450 ppd is very low for a 8800GT. I pull 3k ppd with a 9600GT, and I will leave it running while playing games and watching videos. The new core (1.06) really makes a difference.
Re-read the posts above, the 450ppd was obtained with a 8400GS.
Image
The Model M is not for the faint of heart. You either like them or hate them.

Gerbils unite! Fold for UnitedGerbilNation, team 2630.
Flying Fox
Gerbil God
 
Posts: 24357
Joined: Mon May 24, 2004 2:19 am

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 2:33 pm

Flying Fox wrote:
Lockheed_Tvr wrote:Ok! 177 drivers on an 8400GS:

Big improvement! Jumped from around 450PPD to about 560PPD both on the 98 point WUs. Good enough to scrap SMP, add a console client on one core and get a nice bump in points!

May I know what project that 450/560ppd came from? I have got some new data myself and I would like to compare notes, if it is at all possible.


When I was writing those posts I was getting the 98 point work units (can't remember the project number.) Now I am getting 5004s (479 points) and it looks like I have been averaging a little less than 550 or so on those.
Nothing clever.
Lockheed_Tvr
Gerbil
 
Posts: 87
Joined: Thu Dec 23, 2004 1:30 pm
Location: San Diego

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 6:01 pm

Flying Fox wrote:You can get FahMon/FahSpy to do those ppd calculations for you. ;)


Ooh, thanks for that. I got it started and it's giving me numbers closer to what I'm seeing from the points servers. Yesterday was about 5800 points, and right now FaHMon has me at about 5100 for the 8800GT, 400 for an OC'd E2200, 300 for the 8400GS, 100 for a Celeron 356, and about 50 for my laptop (1.6GHz Turion).

Looks like I finished the bad WU's and am doing some other stuff now, so maybe that explains why the actuals don't match the calculated numbers.
This problem was caused by Windows, which was created by Microsoft Corporation.
sluggo
Gerbil Jedi
Gold subscriber
 
 
Posts: 1540
Joined: Wed Feb 16, 2005 8:44 pm
Location: under the table and dreaming

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 6:07 pm

sluggo wrote:400 for an OC'd E2200

Single instance of the single core client running on the dual core machine? Just confirming.
Image
The Model M is not for the faint of heart. You either like them or hate them.

Gerbils unite! Fold for UnitedGerbilNation, team 2630.
Flying Fox
Gerbil God
 
Posts: 24357
Joined: Mon May 24, 2004 2:19 am

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 7:34 pm

TheEmrys wrote:450 ppd is very low for a 8800GT. I pull 3k ppd with a 9600GT, and I will leave it running while playing games and watching videos. The new core (1.06) really makes a difference.

Same here My 9600GT is pulling in about 3500+ PPD according to FAHmon. It's overclocked to 700MHZ but it's not much of a boost in clock cycles.
CB5000
Gerbil XP
 
Posts: 365
Joined: Wed Mar 26, 2008 4:46 pm
Location: NW region

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 8:01 pm

Flying Fox wrote:
sluggo wrote:400 for an OC'd E2200

Single instance of the single core client running on the dual core machine? Just confirming.

That's correct, single instance of the 5.03 client. I tried running a second single core client on this CPU but it seemed to bog down the installed 8800GT so I dropped the second client. Is that a common occurance or is it possible to run a second core on a dual-core machine without affecting a GPU client?

The 9500GS GPU is at 1291 ppd. It's running on a machine with a 9750 quad-core running the SMP client, but FaHMon appears not to know what to do with its output - it just reports "N/A", even though the SMP client is definitely running (about 21 minutes per percentage point). Maybe I needs to do me some more reading about FaHMon.
This problem was caused by Windows, which was created by Microsoft Corporation.
sluggo
Gerbil Jedi
Gold subscriber
 
 
Posts: 1540
Joined: Wed Feb 16, 2005 8:44 pm
Location: under the table and dreaming

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Jun 30, 2008 10:14 pm

9500GS? I just discovered there's something called a 9600 GSO and for a couple of more hours it's available at ZZF for $104 - $20 MIR, and it makes me wonder if it's worth it for folding purposes. The 9600 GT seems to do much better than your GS, yet specs show 96 SPs for the GSO, and 64 for the GT. I are contused. :lol:
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5355
Joined: Sun Jan 20, 2002 7:00 pm

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Tue Jul 01, 2008 12:02 am

Ragnar Dan wrote:9500GS? I just discovered there's something called a 9600 GSO and for a couple of more hours it's available at ZZF for $104 - $20 MIR, and it makes me wonder if it's worth it for folding purposes. The 9600 GT seems to do much better than your GS, yet specs show 96 SPs for the GSO, and 64 for the GT. I are contused. :lol:

I sometimes get early access to hardware. This has been released, but I think it''s mostly an OEM kinda card.
This problem was caused by Windows, which was created by Microsoft Corporation.
sluggo
Gerbil Jedi
Gold subscriber
 
 
Posts: 1540
Joined: Wed Feb 16, 2005 8:44 pm
Location: under the table and dreaming

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Tue Jul 01, 2008 7:26 pm

Edit: FYI, FahMon lets you configure (in Preferences) how it updates for new WU's it doesn't have in its DB. I discovered that if you click on the particular client you're interested in, it will download the data for its WU if need be. Then you have to reload that client (option available from right clicking the highlighted client) or wait until it updates on its normal schedule and it will calculate PPD from then on instead of showing "N/A" there. Though maybe it has changed in later versions. I'm still running 2.3.2 since the next version (2.3.3) kept crashing on me and forcing me to delete and replace the config files.

I'm tempted to get that 9600 GSO (they extended the $20 rebate another month), since I won't be doing anything serious with it other than folding now and then if it isn't a power pig (which I'd discover when first testing out using that Kill-A-Watt thing I picked up from Newegg last year). Somewhere I read it has fewer... heck I forget now. But it's not just a smaller memory bus (192 bit vs. 256 for the GT), but some other kind of processor now that everybody changed what everything is called and I'm not very interested in finding out what means what. I think it may be RP, but can't recall what that is. 12 for the GSO, 16 for the GT.

No idea what that all adds up to, of course. I saw some talk on some forum that the 768 MB version are preferable, but the persons giving such advice were replying for gamers' purposes.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5355
Joined: Sun Jan 20, 2002 7:00 pm

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Tue Jul 22, 2008 10:36 pm

I did run my Kill-A-Watt on Sunday, and it shows the system I'm running my 8800GT-256 on using ~166 watts. Not bad, and it looks like the system uses around 120 watts before the video card starts folding, though that's not certain. The machine is as follows:

A64-3200+ @ ~2000 (or a bit more) MHz.
2 x 1 GB OCZ DDR RAM, I forget its speed but something like PC-4000, which is 50 MHz faster than the 200 MHz I run it at.
DFI NF4 Ultra-D LANparty motherboard.
1 Trendnet USB 802.11G wifi doohickey for network connection.
The aforementioned 8800GT, Gigabyte brand. 450 MHz GPU, I forget the rest off hand. :oops:

I'm generally getting > 4700 PPD from it, now and then a WU comes through that FahMon claims gives over 4900 PPD. I run it at stock speed.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5355
Joined: Sun Jan 20, 2002 7:00 pm

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Sat Jul 26, 2008 1:26 pm

I haven't been able to upload a WU for this client since before 10 AM Stanford time. Nor have I been able to download a new WU. So my GPU2 client is just sitting here idling.

They seem to be aware of the issue, but they think it's nothing. They upgraded the firmware on all their servers. Because they'd been working well, and we can't have that going on.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5355
Joined: Sun Jan 20, 2002 7:00 pm

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Sat Jul 26, 2008 10:09 pm

An odd thing has happened.

I ran RivaTuner 2.09 for a while but learned it was giving me bad information. So I decided to only use nVidia's tools through what they call their "Control Panel".

I ran nVidia's software, increased the shader clock to the nearest to 5% overclocked that it would allow, and all seemed well. Then I decided to stop the GPU2 client and see what would happen when I let nVidia's software optimize the overclock. I clicked on Optimize, and of course the system gave me a BSD within a minute or so.

I restarted. Now, however, when I run that software again, it gives me the same wrong numbers I was getting from RivaTuner, claiming a 700 MHz core clock and 1700 MHz memory.

I went to nVidia's site and downloaded ntune and tried to install that. It said there was already such software installed. So I uninstalled the system tools, and installed ntune. That software appears to have nothing in it the allows me to change the shader clock frequency.

Video card software can be rather annoying, I must say. :o
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5355
Joined: Sun Jan 20, 2002 7:00 pm

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Fri Aug 01, 2008 2:34 pm

Ragnar Dan, I finally got off my butt and installed the XFX 8800GS card. :P I have had it running under XP for about 15 hours now and FAHMon is reporting the ppd as slightly under 3000. That is quite a difference compared to an 8800GT (256 or 512) with 4700ppd, with a 16 sp advantage. The other difference is that the 8800GT is operating under Vista 64, but then again, I have an 8800GT 256 running under XP Pro and getting 4700ppd as well.. so that basically shot my theory on the OS having an influence on ppd to bits. :-? I don't know, but how can a difference in 16 sp have that much of an impact? I must be overlooking something here.
Fold! And I don't mean your clothes!

Do you have a favorite gerbil recipe? Please share with the TR community!
flybywire
Gerbil Jedi
 
Posts: 1883
Joined: Wed Jun 16, 2004 2:28 pm
Location: Springfield, VA - USA

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Sat Aug 02, 2008 12:40 am

I have no idea what it could be, but it's enough to make a significant difference. Maybe it's the memory bus width.

All I've seen about Vista is that people running the GPU client on there may be able to fold with their CPU's without harming GPU production because less CPU time is spent polling the GPU client.
Ragnar Dan
Gerbil Elder
Silver subscriber
 
 
Posts: 5355
Joined: Sun Jan 20, 2002 7:00 pm

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Sat Aug 02, 2008 7:36 pm

Isn't the 8800GS G80 instead of G92? I didn't have F@H then, but when I got my G92 8800GT I tested it out with some of the CUDA SDK samples and it positively beat the crap out of my buddy's G80 based 8800GTS.
Siglessness is boring.
Image - M4800-Eight1
Image - Vargr-Z97
Forge
Lord High Gerbil
 
Posts: 8010
Joined: Wed Dec 26, 2001 7:00 pm
Location: SouthEast PA

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Sat Aug 02, 2008 10:25 pm

If it is the same one in this review then the 8800GS is a G92 card. Just less shader units, rops and memory bus width than the 8800GT/GTS cards of the same family.
i3-530 | HR-01 Plus (passive) | DH55TC | 4GB Kingston DDR3 | Toshiba 250GB 2.5" HDD | Mini P180 | picoPSU 150-xt w/102w brick | 21w idle
Pegasus
Gerbil First Class
 
Posts: 161
Joined: Fri Jun 30, 2006 1:13 am

Re: NVIDIA 6.12 beta 6 GPU client Released!

Postposted on Mon Aug 04, 2008 10:04 am

Pegasus wrote:If it is the same one in this review then the 8800GS is a G92 card. Just less shader units, rops and memory bus width than the 8800GT/GTS cards of the same family.


Oh *right*! I forgot the 8800GS is to the 8800GT as the original 8800GTS was to the 8800GTX.

256bit cut to 192 versus 384bit cut to 320bit.

I find it hard to believe that the memory bus has much impact on this, since F@H at most uses a MB or so of VRAM here, and often none at all. It just streams stuff from memory to CPU to PCIe to GPU and back, no stops or buffering in VRAM.
Siglessness is boring.
Image - M4800-Eight1
Image - Vargr-Z97
Forge
Lord High Gerbil
 
Posts: 8010
Joined: Wed Dec 26, 2001 7:00 pm
Location: SouthEast PA

Previous

Return to TR Distributed Computing Effort

Who is online

Users browsing this forum: Yahoo [Bot] and 1 guest