I tried the hosted files and it seems to be working for me. No failures anymore.

For my stock GTX 480, It's finishing a WU (175 credit) in just over 4000 seconds. I'm new to CUDA crunching...is this normal?

I stuck with the .33CPU dedication for the GPU computing, what is optimal?

One thing I noticed when watching GPU usage is that it rarely used 100% of the GPU during its run. It looked like a saw going for mid 60s% to 90% each second.

You're been doing VLAR tasks, they are very ineffiently done on GPU's, i suggest you get the ReSchedule1.9.exe program and rebrand the VLAR's to your CPU.

Claggy

I tried the hosted files and it seems to be working for me. No failures anymore.

For my stock GTX 480, It's finishing a WU (175 credit) in just over 4000 seconds. I'm new to CUDA crunching...is this normal?

I stuck with the .33CPU dedication for the GPU computing, what is optimal?

One thing I noticed when watching GPU usage is that it rarely used 100% of the GPU during its run. It looked like a saw going for mid 60s% to 90% each second.

You're been doing VLAR tasks, they are very ineffiently done on GPU's, i suggest you get the ReSchedule1.9.exe program and rebrand the VLAR's to your CPU.

Claggy

Thanks for the info. I'll DL this and see if I can figure it out.

---

Todd,

Thank you for your assistance. I am now up and running with my Geforce 480 running Seti CUDA tasks. Much appreciated.

---

This worked great! WUs are flying now on the GPU.

One last question regarding GPU usage. Using MSI Afterburner to watch usage it's doing that saw pattern where most of the time is in the 68% area spiking into the 70s and occasionally the 90s.

Definitely not full on GPU usage like I see with the CPU in task manager.

---

Wow! I'm glad that everyone is working so well with these new GPU's. Between myself, Rasitmer and Jason G it took a bunch of effort to get this done. Everyone using this build should give them a BIG HIGH 5!

One last question regarding GPU usage. Using MSI Afterburner to watch usage it's doing that saw pattern where most of the time is in the 68% area spiking into the 70s and occasionally the 90s.

This is normal from what I have seen in my own tests with the MSI Afterburner utility - this app is far from optimized so as we go forward the applications should become more optimized - people are working on this - from my understanding.

You're been doing VLAR tasks, they are very ineffiently done on GPU's, i suggest you get the ReSchedule1.9.exe program and rebrand the VLAR's to your CPU.

I highly recommend using this utility to get the most out of your GPU's - it has changed my output level dramaticly! Before using this app I was hard pressed to break 200k RAC and now I am over 300k - this made a huge change.

My average WU time on this machine is about 8m35s with .33 CPU per GPU - I have played with this setting a ton - back and forth and this seems to be the best level while still allocating more cores to cpu tasks. These GPU's still need to be fed well to perform.

Don't bother to run more than one app instance per GPU - the performance just sucks right now - wait until the app gets optimized and I'm sure with the these cards it can be done, just not now.
Todd

---

David Anderson has just fixed the Beta site, re-installing the Fermi app as v6.10 with the correct DLLs. All seems to be running well as an official download on my 470, but we could do with a few more testers to let some Beta tasks through and check that validation etc. is OK too.

Then we should be able to get the official app transferred to the main site, and the generation who follow the intrepid pioneers shouldn't need to sweat so hard over these private drops.

Well, so far so good, but things are not really fast yet. I may have messed something up, but at least no errors yet.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

David Anderson has just fixed the Beta site, re-installing the Fermi app as v6.10 with the correct DLLs. All seems to be running well as an official download on my 470, but we could do with a few more testers to let some Beta tasks through and check that validation etc. is OK too.

Then we should be able to get the official app transferred to the main site, and the generation who follow the intrepid pioneers shouldn't need to sweat so hard over these private drops.

Awesome work gentlemen. You have a community about to turn blue, we're breathing again. Thanks a lot.

http://www.nasa.gov/wav/62283main_landing.wav

Regards,

---

Well, so far so good, but things are not really fast yet. I may have messed something up, but at least no errors yet.

Steve

If I understood it correct, the V12b CUDA app crunch only on ~ 50 % of the Fermi chip.

If I look to the app output of a GTX480, it have only 15 CUDA cores.
My GTX260-216 have 27 CUDA cores. The GTX275, 285, 1/2 295 have 30 CUDA cores.
So either the app don't detect correct, or it's true that the whole Fermi chip isn't used complete.

If the 6.10_cuda_fermi app use the whole Fermi chip, maybe you should go to SETI@home BETA and crunch a few WUs, or if it's allowed you could use this app here after manually DL and installation.
[http://setiweb.ssl.berkeley.edu/beta/apps.php]


------------------------------------------------------------------


BTW. This 6.10_cuda_fermi app is really only for the Fermi chip, or it will replace the 6.08/6.09 apps in near future - so it will work on all GPUs?

---

Wow! I'm glad that everyone is working so well with these new GPU's. Between myself, Rasitmer and Jason G it took a bunch of effort to get this done. Everyone using this build should give them a BIG HIGH 5!

High fiiiiveee! :D

Regards Vyper

---

_________________________________________________________________________
Addicted to SETI crunching!
Founder of GPU Users Group

Wow! I'm glad that everyone is working so well with these new GPU's. Between myself, Rasitmer and Jason G it took a bunch of effort to get this done. Everyone using this build should give them a BIG HIGH 5!

High fiiiiveee! :D

Regards Vyper

Hi Five from Steve as well. Also a mighty thank you for your hard work!

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

Hi Five from Steve as well. Also a mighty thank you for your hard work!

Steve

No problem.. Todd and i are gonna make your machine purr like a cat and produce massive amounts of work..

Keep up the good spirit.

Regards Vyper

---

_________________________________________________________________________
Addicted to SETI crunching!
Founder of GPU Users Group

Hi Five from Steve as well. Also a mighty thank you for your hard work!

Steve

No problem.. Todd and i are gonna make your machine purr like a cat and produce massive amounts of work..

Keep up the good spirit.

Regards Vyper

I can't thank Raistmer, Jason, Todd, and yourself enough! The result of your hard work is amazing! My Rigs RAC jumped 1000 points in 7 hours. I feel like I just launched a rocket. The crazy thing is that my rig is running with such low temps. There is a whole lot more room for tweaking, but this thing is definitly putting out work! Thank you! Thank you! Thank you!

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

[...]
The crazy thing is that my rig is running with such low temps.
[...]

My OCed GTX260-216 make shorties in ~ 2 m : 15 s with stock 6.09 .
A normal (AR 0.44x) WU in ~ 8 m : 50 s.
The EVGA brothers. The GIGABYTE is little bit faster.

If you have the same time, you will see a booster if you'll get a CUDA app which use the whole Fermi chip.
And then the temps will increase also.

---

[...]
The crazy thing is that my rig is running with such low temps.
[...]

My OCed GTX260-216 make shorties in ~ 2 m : 15 s with stock 6.09 .
A normal (AR 0.44x) WU in ~ 8 m : 50 s.
The EVGA brothers. The GIGABYTE is little bit faster.

If you have the same time, you will see a booster if you'll get a CUDA app which use the whole Fermi chip.
And then the temps will increase also.

Sutaru, thank you also! My old GTX 295 did nothing but create errors and crash my rig. Those times yoiu are getting are similar to what I am getting on the 480's. CPU WU's are going in about 20 minutes. If the app is only using half the chip, then I can't imagine the boost with the whole chip. I am wanting to OC the 480's and see what happens, but I may just leave them alone for a bit to see where they go.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

Sutaru, thank you also! My old GTX 295 did nothing but create errors and crash my rig. Those times yoiu are getting are similar to what I am getting on the 480's. CPU WU's are going in about 20 minutes. If the app is only using half the chip, then I can't imagine the boost with the whole chip. I am wanting to OC the 480's and see what happens, but I may just leave them alone for a bit to see where they go.

Steve

You are welcome!


You can't imagine the booster? ;-)
I guess.. only guessing, you would see at least halving of crunching times with a real CUDA_Fermi_app. Double RAC!

The nVIDIA site say, the GTX480 have 480 CUDA cores.
But this are the shader cores. 480 shader cores, are 60 CUDA cores.

So if the CUDA V12b app detect only 15 CUDA cores and only use 15 CUDA cores, then theoretically you have 45 unused CUDA cores.
Then it would mean you would see a quarter crunching times. Fourfold RAC..

But I don't want arouse too much anticipation.. :-)


If current the Fermi chip is only used to 50 or 25 %, you could imagine that the temps will increase later with 100 % usage, or? ;-)


Maybe someone with more knowledge about the CUDA V12b app and the Fermi chip could jump in here? Which performance (RAC) would be to expect with a real CUDA_Fermi_app and 100 % usage of the Fermi chip?

---

You are welcome!


You can't imagine the booster? ;-)
I guess.. only guessing, you would see at least halving of crunching times with a real CUDA_Fermi_app. Double RAC!

The nVIDIA site say, the GTX480 have 480 CUDA cores.
But this are the shader cores. 480 shader cores, are 60 CUDA cores.

So if the CUDA V12b app detect only 15 CUDA cores and only use 15 CUDA cores, then theoretically you have 45 unused CUDA cores.
Then it would mean you would see a quarter crunching times. Fourfold RAC..

But I don't want arouse too much anticipation.. :-)


If current the Fermi chip is only used to 50 or 25 %, you could imagine that the temps will increase later with 100 % usage, or? ;-)


Maybe someone with more knowledge about the CUDA V12b app and the Fermi chip could jump in here? Which performance (RAC) would be to expect with a real CUDA_Fermi_app and 100 % usage of the Fermi chip?

The only thing I can relate as far as temperature goes, is when I ran Furmark. That easilly stressed the GPU's up t0 100%, and my max temperature was only 58C. It is very nice to not have to worry about temperature at all. My GTX 480's are definitly running at less than 50%, with temps of 45C or less.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

Update to my geforce 480 testing...

after finally suspending my other project and running Todd's files for Seti exclusively, I have had three system crashes in the last 8 hours (windows 7 x64).

I have now disabled Seti and am running my other project over the weekend to see if the crashes go away.

Before I went to exclusively SETI earlier today, I had been running only Einstein since I got the 480 with no crashes running their cuda app on the 480.

---

Before I went to exclusively SETI earlier today, I had been running only Einstein since I got the 480 with no crashes running their cuda app on the 480.

Einstein using GPU only occasionally, for FFT. That is, if some suspiction for hardware stability exists, try to process some another project with bigger GPU usage (SETI MB has good GPU usage, but not full indeed).
Then you could see if its hardware failures or smth wrong with software.

Is there any chance that one of you guys that are up and running with your 470s / 480s, could post an idiots guid to installing Fermi friendly SETI. I bought two 470s before I found out that they don't run to well on regular SETI (6.10.43). I have tried following the threads but end up with nothing. Please help. Thanks.

---