ready for some Fermi numbers? GTX 480

actually not yet lol. but i did order it and the package is on the way. i've never bought a video card over $100 before since i'm not a gamer, and for some reason i had the balls to blow almost $600 on a gtx480. as a matter of fact, TWO of them haha. what a great way to spend my last paycheck :rolleyes:

i'm surprised there's no crunching/folding numbers on these yet. will post results as soon as i get them, so stay tuned.

edited to include Fermi GTX 470 and 480 guide and results
================================================== =======

GTX 470 & 480 Setup Guide

i'm assuming you already have BOINC installed and know how to install/setup all the hardware. this is also for windows, and i don't know anything about linux haha, although they should be similar.




#1
download the specific drivers.

i noticed that nvidia has only one driver for all their desktop cards, EXCEPT the new fermi cards, which has their own drivers. here is the LINK for the download. make sure you select the correct OS. as of writing this guide, the latest driver for fermi is 197.41 (make sure it's not 197.45, because that's the drivers for all other cards).

http://img.photobucket.com/albums/v2...ermi/480-1.png

http://img.photobucket.com/albums/v2...ermi/480-2.png




#2
uninstall old video drivers, and install new drivers.

uninstall the existing video drivers either from the control panel or device manager. restart your computer, and install the new drivers you just downloaded. if you're anal about it, you can run ccleaner first to wipe all remaining bits of the old nvidia driver, then install the new one. if you're lazy, you can just skip uninstalling it altogether and just install the new drivers over the old one. after the new drivers are installed, restart again.

http://img.photobucket.com/albums/v2...ermi/480-3.png




#3
next, you need to add/change a profile to use ACEMD beta version v6.23 (cuda30) only, because any other version will give an instant error in the WU.

after logging in to gpugrid.net, go to Your account, and click on GPUGRID preferences. on this page, you need to select any of the separate preferences profiles (home, school, or work) to add/edit. in my case, i choose "school", but it really doesn't matter which one you pick, as long as all the computers with fermi cards use this profile.

in the Run only the selected application option, make sure you tick ONLY the ACEMD beta, and untick the other two boxes. in the Run test application option, select Yes. if the If no work for selected applications is available, accept work from other applications option, make sure it's unticked so you don't get any non-beta WU. after it's set similar to mine below, click Update preferences.

http://img.photobucket.com/albums/v2...ermi/480-4.png


#4
now you need to apply one of the computers to the profile just set.

go back to Your account, and this time select View your Computers on this account. click Details on the computer that has the gtx480/470, and at the very bottom next to Location, select the profile that you created/edited to run only the beta applications. click update.

http://img.photobucket.com/albums/v2...ermi/480-5.png

http://img.photobucket.com/albums/v2...ermi/480-6.png




#5
go to the BOINC manager, select the Projects tab, highlight the GPUGRID project, and click update. you're now ready to rock n' roll!!

http://img.photobucket.com/albums/v2...ermi/480-7.png


================================================== =======

Points and Results

if you have results on these cards, please post them. things that greatly affect the output are overclocks, OS, dedicating a full thread or not, hyperthreading, and whether you also game while crunching or not. i had a 6k/day boost in points by having max stable OC, and also dedicating a full thread to GPUgrid. here are my results:

points: 32k/day
clocks: stock
OS: win7 64bit
dedicated thread: no
HT: yes
gaming: no

points: 38k/day
clocks: 800/1600
OS: win7 64bit
dedicated thread: yes
HT: yes
gaming: ~1 hour/day

points: 77k/day
clocks: 775/1550
OS: win XP 32bit
dedicated thread: yes
HT: no
gaming: none
comments: WCG at 100% load also

Just bought one today also. How come so much...$600? Seems like gpugrid will have the issues ironed out within the next week.

Quote:

---

Originally Posted by =[PULSAR]=

Just bought one today also. How come so much...$600? Seems like gpugrid will have the issues ironed out within the next week.

---

im in california, and there's just so many computer stores here lol. price gouging, and on top of that, tax is a bioch on things that costs a couple hundred.

what issues has gpugrid been having?

On 30 March they released a beta for Fermi but has had some issues because the creators of gpugrid haven't had a 480 or 470. GPUgrid did just receive a 480 and 470 on 23 April and they are ironing the kinks out as we well type.

I'm really interested to know how they do with GPUGrid and F@H, could you please post numbers for each? Thanks :)

I am typing from a PC with Win7 64 that already has a GTX480 installed, just waiting on GPUGrid.
pwolfe has 2x 470s installed on a linux box and waiting also.

Quote:

---

Originally Posted by Oj101

I'm really interested to know how they do with GPUGrid and F@H, could you please post numbers for each? Thanks :)

---

I'll give it a shot on F@H for ya and let you know the stats. It has been a while since I've used it but we will see I know different WU's have different projected ppd rates.

If you could run it for a week and tell me the average PPD... :D You'd be helping a good cause anyway :D

ack ... cross posting :-(

Quote:

---

Originally Posted by Oj101

If you could run it for a week and tell me the average PPD... :D You'd be helping a good cause anyway :D

---

Alright, I do it for the super computer week and see what the average is like. Wasn't F@H suppose to release a new GPU2 client for Fermi architecture?

I don't know, I kind of gave up on F@H after my HD2900s as I found it was very touchy and needed to be babied. It could've just been my Windows installation at the time but I never got back into it and swapped to WCG instead :)

I think F@H is still working things out as well
I hope these perform as everyone hopes

Quote:

---

Originally Posted by INFRNL

I think F@H is still working things out as well
I hope these perform as everyone hopes

---

yes, 3.5 times the amount of what high end cards output now is a pretty big claim. i'm hoping it can just get close to these numbers.

Quote:

---

Originally Posted by Snow Crash

I am typing from a PC with Win7 64 that already has a GTX480 installed, just waiting on GPUGrid.
pwolfe has 2x 470s installed on a linux box and waiting also.

---

ah yes i just noticed you had one on the grid. not sure if you noticed, but you got all errors today with the gtx480. i hope it's from the bugs and not your overclocks.

GPUGrid is clear on the fact that their app crashes on GTX480 but I thought I would try anyway :)

To this test card's limits I'm running a couple other projects.
I'm not going to do any *optimized app* installations as I am a simple kind of cruncher.

SETI works good with shaders up to 1732 (1803 is a little shakey and 1823 grey screens).
It is hilarious to watch WUs crank through in 15-17 SECONDS instead of hours, the uploads and downloads can hardly keep up :rofl:

<edit> Seems like this is actually erroring without telling me, finishing and sending up bad results. I may test again with a lower clock to see if I can get some valid reutrns</edit>

SETI has some WUs that are calculating signals from low angles and they run slower on a GPU than on a CPU.
It's a known issue. There is an opti app that will switch to your cpu when these WUs are detected.

Working out OC for Colltaz right now, Stock is good, 1532 is good, 1632 is good, 1732 good.

Interesting temps and fan speed.
Collatz @ 1732 shaders is pushing the GPU usage at 85% which is in about the same ballpark as GPUGrid (Win7 HP 64, i7-920 w/HT on, 4.2 GHz)
I manually set the fan at 70% and temp is ... wait for it ... 73c :shocked: Ambient is ~23-24 c.
It's looking pretty good to me so far, temps and fan noise are well within my tolerance levels :up:
I have not put an extra 120 mm fan blowing on the backside (like I do on the GTX295) because It just doesn't need it.

I have no intention of running any of these projects as soon as GPUGrid is ready, just trying to get a feel for the card's limits.

That's great Snow Crash, so all your doing now is waiting for GPUGRID to give the go ahead-hope that happens soon. Glad it seems to be working with the other projects might get an 470 soon myself

guess this gives me a little time before i bite. no sense it getting one if we can't fold yet. although i need to sell off another gtx275 or 2 before those lose value an i can use that money to help fund the 480. i hope a 480 can produce well more than 2 275's can otherwise would not be worth the investment

Quote:

---

Originally Posted by INFRNL

guess this gives me a little time before i bite. no sense it getting one if we can't fold yet. although i need to sell off another gtx275 or 2 before those lose value an i can use that money to help fund the 480. i hope a 480 can produce well more than 2 275's can otherwise would not be worth the investment

---

I was thinking that no doubt once everything is sorted out the 480's will produce some good results-but hardware in numbers will not hurt either:D. Though to what you said about waiting- I hope prices for the 470/480's don't go up as time goes on

I only have one basis for equating the 480 for CUDA ... on Collatz sftp (the major dude who contributed a 470 and a 480 to GPUGrid) said his 295 would crunch a WU in 18minutes. My 480 is cranking them out in 7 minutes. That is not to say GPUGrid will do better or worse in optimizations but as the only piece of evidence I can point to it looks pretty good to me. Lastest post from GPUGrid regarding fermi is they expect to have a linux version ready tomorrowthey are now focused on optimizations and they may have a Windows version ready also. Hold on to your hats, I'm gonna put up some serious numbers (well for me anyway :sofa:)

The guys at GPUGrid really do move along quite quickly. If you follow all the different tech they try out and the papers they publish, top it off with the public project ... they are Xtreme :up:

So Fermi cards are now working properly and producing validated WU's?

i have a chance to trade my GTX 480 + $200 cash for two GTX 470. is this worth it for crunching outputs? i figured the GTX480 has 480 shader processors clocked at 1400mhz, and two GTX470 has 896 shader processors combined ( 448 x 2) at 1215mhz, so two 470's should give much more output correct?

Quote:

---

Originally Posted by Snow Crash

I only have one basis for equating the 480 for CUDA ... on Collatz sftp (the major dude who contributed a 470 and a 480 to GPUGrid) said his 295 would crunch a WU in 18minutes. My 480 is cranking them out in 7 minutes. That is not to say GPUGrid will do better or worse in optimizations but as the only piece of evidence I can point to it looks pretty good to me. Lastest post from GPUGrid regarding fermi is they expect to have a linux version ready tomorrowthey are now focused on optimizations and they may have a Windows version ready also. Hold on to your hats, I'm gonna put up some serious numbers (well for me anyway :sofa:)

---

this is very exciting to hear. now i'm wondering if the processor can keep up if it's also loaded 100% with WCG.

responded to your Q here in the other thread.
Like all GPU projects you will take a little hit on both sides if you are running with HT on.
If no HT then gpu usage is only slowed down by 1-2% at most.

Quote:

---

Originally Posted by Snow Crash

I only have one basis for equating the 480 for CUDA ... on Collatz sftp (the major dude who contributed a 470 and a 480 to GPUGrid) said his 295 would crunch a WU in 18minutes. My 480 is cranking them out in 7 minutes. That is not to say GPUGrid will do better or worse in optimizations but as the only piece of evidence I can point to it looks pretty good to me. Lastest post from GPUGrid regarding fermi is they expect to have a linux version ready tomorrowthey are now focused on optimizations and they may have a Windows version ready also. Hold on to your hats, I'm gonna put up some serious numbers (well for me anyway :sofa:)

The guys at GPUGrid really do move along quite quickly. If you follow all the different tech they try out and the papers they publish, top it off with the public project ... they are Xtreme :up:

---

if this is wht we can expect on the grid as well, then this is excellent news:up:

then another question is like white dragon asked...load up with 470's or 480's:shrug::ROTF: 2x470 is approx $200 more, but is the production worth it. only time will tell; hopefully we make the right decision. my pln is 480's but 470's might be the way to go and save some $ or wait for the next rev of these cards; hopefully soon:shrug:

Posted by the people of gpugrid.

Quote:

---

So far, the performance of a GTX480 is only 30% faster than a GTX 285.
Fermi is completely different from a G200 chip, it's not just plug&play for a code optimized on the G200 as ACEMD.

We are trying to understand what's the problem. A factor two (100% faster) should be really the minimum for the hardware.

I'll keep you posted.
gdf

---

crap. only 30% faster than gtx285, but price difference is more than 30% :(

Quote:

---

Originally Posted by WhiteFireDragon

crap. only 30% faster than gtx285, but price difference is more than 30% :(

---

But should soon be 100% faster. :yepp:

Quote:

---

Originally Posted by Snow Crash

I only have one basis for equating the 480 for CUDA ... on Collatz sftp (the major dude who contributed a 470 and a 480 to GPUGrid) said his 295 would crunch a WU in 18minutes. My 480 is cranking them out in 7 minutes. That is not to say GPUGrid will do better or worse in optimizations but as the only piece of evidence I can point to it looks pretty good to me. Lastest post from GPUGrid regarding fermi is they expect to have a linux version ready tomorrowthey are now focused on optimizations and they may have a Windows version ready also. Hold on to your hats, I'm gonna put up some serious numbers (well for me anyway :sofa:)

The guys at GPUGrid really do move along quite quickly. If you follow all the different tech they try out and the papers they publish, top it off with the public project ... they are Xtreme :up:

---

That seems like good news:up: But are you sure that his 295 crunches a WU in 18 minutes in one of the two GPUs or both of the GPUs? I don't really know how Collatz works, though:shrug:

Quote:

---

Originally Posted by WhiteFireDragon

crap. only 30% faster than gtx285, but price difference is more than 30% :(

---

:down::down::down::down: And I woulda have bought a few of the output was more of what we were expecting.

Quote:

---

Originally Posted by trn

:down::down::down::down: And I woulda have bought a few of the output was more of what we were expecting.

---

They are still tweaking the issues. They discovered they have to rewrite a lot of things because everything was based on the older architecture. The 30% increase was with using Fermi on the older GTX2xx architecture.

From GDF

Quote:

---

Current performance on the GTX480 is 35% faster than the fastest version of ACEMD on a GTX285.

There is little more we can do I think. The cuda3 compiler seems much worse at optimizing the code than the 2.2 version. If this is the case, then 3.1 could bring some improvement.

Tomorrow, we clean up and submit the new application in beta.

gdf
PS: no doubts this is much less than expected

---

Very disappointed in the performance figures quoted by GDF :shakes: Looks like the 295 will still be the king of cards for a while. Glad I waited before ordering now:rolleyes:

yes, having blown over $1k on two of these, i sort of regret spending on so much. i was just so excited when i saw it in stock. well to minimize buyer's remorse, i'm going to be optimistic and know that they'll be optimizing the WU's for these cards within the next few weeks

Update

Quote:

---

we have now a good performance model of the application. The maximum we should have expected from fermi is a reduction in time of 60% compared to a gtx275. Actually, we now think that this will be achievable in the future. The problem is that cuda 3 is slower even on gtx200 by almost 15%. Indeed, we are 60% faster compared to both applications compiled for cuda3 but compared to cuda2.2 only 35%.

On our machine the running temperature is 91 degrees. Equivalent to a gtx275.
The only real problem is the price. Later i will report the time of the gtx470 and running temperature.

Gdf

---

Quote:

---

Originally Posted by =[PULSAR]=

Update

---

if this is true, then fermi is not worth the cost, right? just trying to figure out if my head is working correctly.:rofl:
if the 480 is max 60% faster, 60% more cost would mean the 480's should only cost $400

Quote:

---

Originally Posted by WhiteFireDragon

crap. only 30% faster than gtx285, but price difference is more than 30% :(

---

Well, that's hardly ever not the case with new hardware.

Although, to be fair, Fermi was looking amazing for math. I still think the potential is there, though. How can it not be with those @Home benches? :shrug:

Quote:

---

Originally Posted by INFRNL

guess this gives me a little time before i bite. no sense it getting one if we can't fold yet. although i need to sell off another gtx275 or 2 before those lose value an i can use that money to help fund the 480. i hope a 480 can produce well more than 2 275's can otherwise would not be worth the investment

---

PM me. I didn't think you had any left. :rofl:

bob

Okay well fermi is up and running now just waiting on the gpugrid to release beta they said sometime tomorrow.

@Pulsar ... have you run any Collatz to get a feel for your OC? It runs a higher on the GPU utilization than GPUGrid so I think we will be stable if we max Collatz. Nice way to get a feel for the card in general, it's OC capabilities and the temps. Currently I am rock solid at 1672 but can push it to 1732 at stock v. I have not had much sucess OCing higher with more volts but I am sure I will be working on that again once GPUGrid is ready for us.

I'll give it a shot later on this evening trying to hammer out the last kinks in my WCG farm, almost had it finish and then HDD died. Back to using a prehistoric IDE :(

Isn't it a better idea to buy used GTX295s off eBay? Some of them are going under 300 dollars already.

Quote:

---

Originally Posted by lkiller123

Isn't it a better idea to buy used GTX295s off eBay? Some of them are going under 300 dollars already.

---

I have 2 gtx 295's and you can't run more than 1 per machine on evga MB's for gpugrid so byebye 295's

Quote:

---

Originally Posted by =[PULSAR]=

I have 2 gtx 295's and you can't run more than 1 per machine on evga MB's for gpugrid so byebye 295's

---

Is it only for EVGA MBs? I am thinking of getting some used 295s in the summer. I bet their price will fall like crazy by summer:)

Quote:

---

Originally Posted by lkiller123

Is it only for EVGA MBs? I am thinking of getting some used 295s in the summer. I bet their price will fall like crazy by summer:)

---

Yeah tried it on both E760 and E758 tried it on an older s775 and AM2 with no issues. I'm looking at getting rid of mine shortly probably within the next month or two. Want me to contact you when I do?

Quote:

---

Originally Posted by =[PULSAR]=

Yeah tried it on both E760 and E758 tried it on an older s775 and AM2 with no issues. I'm looking at getting rid of mine shortly probably within the next month or two. Want me to contact you when I do?

---

It will be no earlier than Mid June/Early July before I have time to do builds. You know how fierce high school is:p: I will be happy if you do contact me, though:up:

just got mine today and been really frustrated with it :mad:. look at all these errors it's giving. i've been trying to fix it for over an hour already and still no luck.

it seems that nvidia only has one driver to cover all cards except for the new gtx400 series, which has it's own drivers. i updated it, and still doesn't work. it keeps going through a loop of downloading the WU, and instantly report it back as error, and DL's a new one, and same thing happens. any idea how i can just get it running or is fermi still not yet supported?

GDF says it is beta is supported for linux. He says it may or may not work on windows yet, some users are getting it to work on XP. I tried running it today on win7 and no luck.

are you getting an infinite loop of error as well? i'm not sure if BOINC even detects mine. this is on win7

yeah running win7 64 here and as soon as a unit start it errors and loads another.

maybe you could for the time being run gpugrid on linux using vmware?

I'm currently to addicted to the supersonic sled demo haha

Quote:

---

yeah running win7 64 here and as soon as a unit start it errors and loads another.

---

That was happening to me also on Win7 64 ... then I turned my OC down until I started processing properly. I think I'm at high 15XX and stable.
I also set the environmental variable they posted (SWAN_SYNC=0) which forces the GPUGrid app to use a full core per WU, they say you get a substantial (20-30%) performance increase.

Wierd I'm running the card at stock speeds and its still doing it. What drivers are you using?

ack ... I'm at work now but I think 197.55? ... the version that said they were expanding to quad SLI support ... not that I am doing that but it might help you find the version number.

Quote:

---

Originally Posted by Snow Crash

That was happening to me also on Win7 64 ... then I turned my OC down until I started processing properly. I think I'm at high 15XX and stable.
I also set the environmental variable they posted (SWAN_SYNC=0) which forces the GPUGrid app to use a full core per WU, they say you get a substantial (20-30%) performance increase.

---

is it a full core or a full thread? if it's a full core, i'm not sure i'm willing to dedicate 1core/2thread to the GPU, as that's 25% on one machine and 50% loss from another machine. not sure if this is worth the hit to WCG right now

1 thread ... neither GPUGrid or I am saying you have to do this, I was just saying what I have my system set to and that it works. If they have any beta WUs left when I get home maybe I will try running without the dedicated thread ...
:doh: :am: maybe that was what why I was erroring in the beginning and not the OC like I *assumed*. Heck, I have a backup from last night ... I AM going to test this out.

i want to test this out right now too, but i'm getting the max quota message. i have to wait a full day before they will give me more :shakes:

Should roll over at 12:00 UTC (that's in 30 minutes ... at least that is how WCG works). If you are really desperate you can try changing the machine name and then after it's all set and crunching to sucess you can tell GPUGrid to merge the PC with the old name.

I've been trying to download the ACEMD beta work units and it keeps telling me there is no work available. I posted on their forums asking to increase the number of ACEMD beta units available. I've got 2 480 GTX cards ready to roll.

hmmm ... I've been getting a steady steam of them for hours now. Just uploaded one, another downloaded immediatley. Are you sure you have all the GPUGrid preferences set correctly? check beta box and make sure the "Run test applications" is also selected.

Hmm for some reason BOINC is saying my 480 194 GFLOPS peak? Whats everyone elses say?

Okay well just to give you guys an idea a GTX480 850mhz core and 1700mhz shader is giving me 32.8k PPD.

Quote:

---

Originally Posted by =[PULSAR]=

Okay well just to give you guys an idea a GTX480 850mhz core and 1700mhz shader is giving me 32.8k PPD.

---

so you got at least one WU to complete with a 480?? did you do anything different than before to make it work? or did gpugrid just released beta drivers or WU?

i'm hesitant to try again because if it keeps cycling through errors again, i'll be locked out of grpugrid again on this computer, even if i put back the good GPU

I got quite of few done. GPUgrid is throwing me odds WU's though some are cuda 6.03 which error out, then theres the cuda 3.0 beta v6.23 which so far work perfectly. You have skype or something I'll walk you through it there are some things you need to change.

thanks for the offer of personal help! i actually don't have skype, but i have AIM. looks like you have the exact opposite... ICQ and skype haha. i think it'll be better if you posted the steps in public, that way others can view it too, and it looks like several people can't get their fermi's working either.

i also sent you a PM

All WUs EXCEPT beta will error if executed on a 470 or a 480.

Quote:

---

Originally Posted by =[PULSAR]=

... GPUgrid is throwing me odds WU's though some are cuda 6.03 which error out ...

---

To turn on/ off delivery of the specific GPUGrid apps ...

http://www.gpugrid.net/prefs.php?subset=project

Click Edit on either the Primary or your specific profile as appropriate and change the "Run only the selected applications" to only have checked the ones you want.

In order to run betas you also need to make sure the "Run test applications" is checked yes.


WFD - all your errors were non-betas. Once you change your setting like I described above you should be rockin and rollin :up:

Quote:

---

Originally Posted by Snow Crash

All WUs EXCEPT beta will error if executed on a 470 or a 480.



To turn on/ off delivery of the specific GPUGrid apps ...

http://www.gpugrid.net/prefs.php?subset=project

Click Edit on either the Primary or your specific profile as appropriate and change the "Run only the selected applications" to only have checked the ones you want.

In order to run betas you also need to make sure the "Run test applications" is checked yes.

---

Thats how mine is set I only have beta checked but I'm still getting the old v6.03 WU's also :confused:

Make sure those settings are on the same profile that your machine is set up to use ... see the Location field at ...

http://www.gpugrid.net/hosts_user.php

Snow that link bring any of us to our own machines

ummm ... that was the point ... I am helping Pulsar make sure the "Location" of his machine with the 480 was set to the same "Profile" that he set up specifically for beta. Profile=Location, same thing, two different names. Dumb as it is, all boinc do this. Seeing as he said he already had the "Profile" set up I wanted him to check his Location. While he could look at his machine details and get the same info the link I used works fine. Setting up generically like I did makes the post useful for other people, not just him. Is there something you think I should do differently?

Turns out I had this option checked to "Yes" - If no work for selected applications is available, accept work from other applications?

Glad to hear you found the issue ... looks like between us we have covered all the bases ... now to see if we can get Otis to write it up nice and clean and add it to one of the great stickies he has already put together. :clap:

No I figured the link pointed each person to their profiles- I was just making sure I was not crazy

Still wondering why its taking me 12 minutes to do a WU and other people are doing them in 6 minutes?

People doing 6 minutes are Linux which is known to run faster than Windows (for now).
reasons you might be slow ...
Running stock
Not turned on SYNC_SORT
Running HT ON
Gaming
Win7 or Vista (XP is better)

Quote:

---

Originally Posted by =[PULSAR]=

Okay well just to give you guys an idea a GTX480 850mhz core and 1700mhz shader is giving me 32.8k PPD.

---

if this is true; not good or they need to figure something out. a gtx 275 can produce this much:shakes:

at least you are making progress and getting some production done:up:

i also had my fermi up and running the whole day. i was at some friend's place and wasn't suppose to come back until tomorrow morning, but i've been dying to see how my machine is doing so i went home tonight :lol:

at stock, it's crunching with fan 50%, GPU usage 50%, and temps at a steady 75C. one weird thing is that there is a huge variance in how long it takes to complete a WU, ranging from 8min to 13.5 minutes. all WU are worth 281 points. any ideas to the inconsistencies?

ok, i updated the original post to now include a guide to put these cards to work. while i was writing the guide, the computer crashed, and i had to rewrite the whole thing :cussing:. my OC'ed CPU was on the edge of stability, just barely enough to be stable with the old video card. with the new 480 plugged in, it somehow became unstable, and i have no idea why since it's just switching video cards. i decreased the cpu by 150mhz (temporarily), and the system seems stable.

so i'm trying to compare my 480 to the gt240 and gtx260 that i have to see relative performance. this was a little hard since setting my other machines to only accept the exact beta WU's that the 480 crunches at because BOINC only allowed these beta WU for fermi cards (ACEMD beta version v6.23 (cuda30)) and would not let me download them. when i downloaded the WU on the fermi machine, i suspended the task, uninstalled the video drivers, plugged in my gt240, and reinstalled the correct drivers since the 480 and gt240 uses different drivers.

after my gt240 competed the WU for the fermi, i was a little surprised when i saw the results. it only took twice the amount of time to finish the WU (23.5minutes). at this rate, it's a projection of 16.5k points/day, which is exactly half of the gtx280. when crunching on the other beta WU (ACEMD beta version v6.22 (cuda)), my gt240 even gave more points, at a projection of 18k points/day. all parameters of the test was the same.

to be honest, i'm a little disappointed with the outputs of these fermi's. it's sad to see that two $50 cards will give the same, or even more computational power than a $500 card at a much lower power consumption :(

http://www.gpugrid.net/results.php?hostid=69365

Here are my beta work units being completed with 2 GTX 480s running at 1550 shader with an i7 920 with HT on at 3.2ghz and SWAN_SYNC set to 0 to allow each work unit to utilize an entire thread.

"no access" for me

Link to Dangers' host with the 480 rather than his personal ID.

http://www.gpugrid.net/results.php?hostid=69365

I don't think something is right ... here is a link to my 480 at stock speeds on an i7-920 HT On / 3.285 GHz / SYNC_SORT=0
http://www.gpugrid.net/results.php?hostid=31780

I'm coming in much faster, typically around ~675 seconds while yours is at ~1100.
The only difference I can see is that I am running Win7 64 Home Premium and you have Ultimate which should not make a difference.

Are you sure you have SYNC_SORT as the environment var with a value of 0 and have you completly stopped BOINC and restarted it? Sometimes BOINC does not stop even when you tell it to so I rebooted.
Do you have your BOINC folder excluded from any AV scanning?

Danger30Q, something looks very off with your results when it's taking over 1000sec (17min) . it shouldn't take that long to finish a WU. mine is at stock clocks, no SWAN_SYNC, and win7 OS completes in about 13.5minues. are you also gaming with it while crunching? here are my latest results:

http://img.photobucket.com/albums/v2...completion.png

Just finally got my machine built and configured last night. Attempted to pull some WU's but I hadn't changed my pref's to only pull the beta app. I'll know tonight if it is all working on my end. 2 470's and Ubuntu 10.04. Currently using 195.16.18 until the x.x.24 drivers hit the ubuntu repo's or ppa's.

I know the SWAN_SYNC set to 0 is working properly b/c each unit uses 13% of my i7 with HT meaning it's using an entire thread. I'm not gaming and I've restarted BOINC many times with system restarts. I have the Nvidia control panel power management set to maximum performance as well as my Win7 power management to high performance.

Each GPU is only at 40% usage.

EDIT: Is SWAN_SYNC and SYNC_SORT the same thing? I only have SWAN_SYNC set to 0 as a System variable.

I'm using MSI Afterburner to overclock the card. I'm going to try EVGA Precision 1.9.2 later tonight. For some reason, GPUGrid is only showing that my card is running at 0.81ghz. I'm hoping I'll have better luck with Precision, and I'm going to update to BOINC 6.10.51 released on May 2.

Quote:

---

Originally Posted by Danger30Q

I know the SWAN_SYNC set to 0 is working properly b/c each unit uses 13% of my i7 with HT meaning it's using an entire thread. I'm not gaming and I've restarted BOINC many times with system restarts. I have the Nvidia control panel power management set to maximum performance as well as my Win7 power management to high performance.

EDIT: Is SWAN_SYNC and SYNC_SORT the same thing? I only have SWAN_SYNC set to 0 as a System variable.

I'm using MSI Afterburner to overclock the card. I'm going to try EVGA Precision 1.9.2 later tonight. For some reason, GPUGrid is only showing that my card is running at 0.81ghz. I'm hoping I'll have better luck with Precision, and I'm going to update to BOINC 6.10.51 released on May 2.

---

if the "acemdbeta" process in the task manager shows 13% CPU usage, then you have it the SWAN_SYNC correct. it's very odd why your performance is so bad. only reason i can think of, is that if you overclocks fail, your core and shaders will downclock. DL the EVGA precision (don't DL from official nvidia, theirs is not updated) or GPUz to check the gpu usage and clocks when it's crunching to make sure it's correct. at stock, your core usage should be 50%, and a little less when overclocked

I'm just getting no work is available for acemd beta version / acemd beta version is not available for your type of computer. 2 470s - linux

Did you make sure to select yes to the option "Run test applications?" in the GPUGrid preferences? I made this mistake for about a day of being told there were no ACEMD beta work units to download.

Sorry for the confusion I caused on the environmental var ... you are all correct ... it is SWAN_SYNC.
The only time I see WU that slow is after a driver crash and the GPU goes into a low power mode ... I have precision 1.9.2 and it will show what speed it is actually running, not just what it is set to. Once the driver resets itself the only way to back up to full speed is to reboot :-(

I came home today to find that there were no more beta work units available. Hopefully they refill the beta work unit pool asap.

I'll be trying out Precision tonight instead of Afterburner and hopefully that improves the speed of completion.

Quote:

---

Originally Posted by Danger30Q

I know the SWAN_SYNC set to 0 is working properly b/c each unit uses 13% of my i7 with HT meaning it's using an entire thread. I'm not gaming and I've restarted BOINC many times with system restarts. I have the Nvidia control panel power management set to maximum performance as well as my Win7 power management to high performance.

Each GPU is only at 40% usage.

EDIT: Is SWAN_SYNC and SYNC_SORT the same thing? I only have SWAN_SYNC set to 0 as a System variable.

I'm using MSI Afterburner to overclock the card. I'm going to try EVGA Precision 1.9.2 later tonight. For some reason, GPUGrid is only showing that my card is running at 0.81ghz. I'm hoping I'll have better luck with Precision, and I'm going to update to BOINC 6.10.51 released on May 2.

---

this is not right is it? 40%, you should be way higher, no? thats like 2d :shrug:
i hope they get this figured out. i am not impressed at all with fermi, i know these cards are supposed to kick arse

just for clarification, when setting the SWAN_SYNC, is the vairable name "SWAN_SYNC=0" and the variable value is "1", or is the variable name "SWAN_SYNC" with the value "0"? win7 forces me to enter a value and i don't know what number to put. so left pic or right pic?

http://img.photobucket.com/albums/v2.../swansync0.png http://img.photobucket.com/albums/v2.../swansync1.png

btw, how did you guys figure out this trick about SWAN_SYNC? i tried to google it, and nothing came up. what does swan_sync mean and how does it work?

Quote:

---

Originally Posted by Danger30Q

I came home today to find that there were no more beta work units available. Hopefully they refill the beta work unit pool asap.

---

yeah sometimes it doesn't give beta WU, but that only happened once to me. during this time that you didn't get them, i still received them fine.

did you ever figure out has to why your WU's take over 1000+ sec? your GPU at 40% is a little low, unless you had an extremely high overclock. it should be around 50%.

well its been almost 24 hours of no work available messages for me. I just want 2 wu's to verify that everything is working.

Quote:

---

Originally Posted by WhiteFireDragon

just for clarification, when setting the SWAN_SYNC, is the vairable name "SWAN_SYNC=0" and the variable value is "1", or is the variable name "SWAN_SYNC" with the value "0"? win7 forces me to enter a value and i don't know what number to put. so left pic or right pic?

btw, how did you guys figure out this trick about SWAN_SYNC? i tried to google it, and nothing came up. what does swan_sync mean and how does it work?

yeah sometimes it doesn't give beta WU, but that only happened once to me. during this time that you didn't get them, i still received them fine.

---

SWAN_SYNC is the variable and the value is 0.

There was so much clamor about using a whole CPU thread to process a GPU WU in the past that they gave this to us as an option rather than force it on us.

http://www.gpugrid.net/forum_thread....rap=true#16666
(For those that do not know, GDF is the head honcho over at GPUGrid)


My betas have dried up for now also but I have Collatz set up with a resource percent of 1 so it doesn't interfer with GPUGrid but will pull a WU when GPUGrid runs dry. I hate being idle :rofl:

With HT off, I'm getting around 65% GPU load with my 480 and full CPU thread, and around 8 minutes for a WU.
With HT on, sharing CPU cycle with 12 threads of WGC, I'm getting around 60% GPU load, and 9 minutes for a WU.

My 480 is overclocked to 850Mhz though, so that's got to contribute to reduce the work time.

Getting a steady stream of WUs from GPUGRID at the moment, with pauses of around 10 minute max before receiving a new WU to crunch.

I don't understand why I am not getting WUs when other people are. Here are my pref's for that computer. Everything looks in order to me... Is it because I'm on Linux?


Resource share
If you participate in multiple BOINC projects, this is the proportion of your resources used by GPUGRID 100
Use NVIDIA GPU if present
(enforced by 6.10+ clients) yes
Run test applications?
This helps us develop applications, but may cause jobs to fail on your computer yes
Is it OK for GPUGRID and your team (if any) to email you?
Emails will be sent from ps3grid@gmail.com; make sure your spam filter accepts this address. yes
Should GPUGRID show your computers on its web site? yes
Default computer location home
Maximum CPU % for graphics
0 ... 100 75
Run only the selected applications ACEMD: no
ACEMD ver 2.0: no
ACEMD beta: yes
If no work for selected applications is available, accept work from other applications? no
Use Graphics Processing Unit (GPU) if available yes
Use Central Processing Unit (CPU) yes

You only have betas checked?
Are there wu's available?

Quote:

---

Originally Posted by pwolfe

I don't understand why I am not getting WUs when other people are. Here are my pref's for that computer. Everything looks in order to me... Is it because I'm on Linux?


Resource share
If you participate in multiple BOINC projects, this is the proportion of your resources used by GPUGRID 100
Use NVIDIA GPU if present
(enforced by 6.10+ clients) yes
Run test applications?
This helps us develop applications, but may cause jobs to fail on your computer yes
Is it OK for GPUGRID and your team (if any) to email you?
Emails will be sent from ps3grid@gmail.com; make sure your spam filter accepts this address. yes
Should GPUGRID show your computers on its web site? yes
Default computer location home
Maximum CPU % for graphics
0 ... 100 75
Run only the selected applications ACEMD: no
ACEMD ver 2.0: no
ACEMD beta: yes
If no work for selected applications is available, accept work from other applications? no
Use Graphics Processing Unit (GPU) if available yes
Use Central Processing Unit (CPU) yes

---

i posted a guide on the very first post of this thread. you left out one important part in step #3. i highlighted it bold and red for you. you currently have it set to "no", it should be "yes"

I looked at a lot of accounts on GPUGrid and only found 470 running linux and they have not got any new betas in the last two days.
GPUGrid set the beta up originally on Linux and us Windows types are trying to find out how we can match (much less beat) their runtime?

Quote:

---

Originally Posted by WhiteFireDragon

i posted a guide on the very first post of this thread. you left out one important part in step #3. i highlighted it bold and red for you. you currently have it set to "no", it should be "yes"

---

Then you should update the image in your howto as it shows only the beta radio button selected, not the acemd.

thanks though! we'll see if that has any effect

Quote:

---

Originally Posted by pwolfe

Then you should update the image in your howto as it shows only the beta radio button selected, not the acemd.

thanks though! we'll see if that has any effect

---

what do you mean "beta radio" button? i have something wrong on there? did you get yours crunching yet?

on the gpugrid preferences page screenshot, you only have

ACEMD - no check mark
ACEMD - ver 2.0 - no check mark
ACEMD beta - checked

which is what I had originally as my set up, but you said I need to change ACEMD to yes, unless I am misunderstanding you.

But no, still unable to pull any WUs, but that may very well be becasue there are none available.