i was wondering if i could use a geforce 260 with a old amd 3000+ 1.8 cpu i got laying around with a pci express slot i would just be useing it for gpu crunching or should i get like the 9800 gt or should i just build another amd phenom ii x4 940 hehe thanks for any help you can give

use what you have... a machine crunching is better than no machine crunching...

if it will work as is, do it...

would my 3000+ be any good at crunching it self or should i add a video card seti will be its only job i got a phenom ii x4 940 and Athlon 64 X2 4800+ crunching now

any cruncher is a good cruncher. the 3000+ can be a bit slower(a lot) than the phenom but if you are looking to save money then stick with what you have.

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

can the 3000+ do two wu at a time or one

would my 3000+ be any good at crunching it self or should i add a video card seti will be its only job i got a phenom ii x4 940 and Athlon 64 X2 4800+ crunching now

At least as good as my 3000+

It is no speed demon, but it turns in finished, valid work.

---

that's what i was wondering if it would get the bigger ap wu in on time thanks for your help ill get it running after i get a power source and clean it up a bit :) at least then it will be of use

Hi Nojnagrom,

I believe your 3000+ is only a single core, so it would be 1 wu at a time. If the mobo you're installing it onto has a PCI-E slot it should work, but you might find that it is PCI-E 1.1 not a 2.0. The difference is in the bandwidth you get but the card should work fine as long as the PSU has the connectors to power it properly. You'll probably find that the RAC of the GPU will be much more than the CPU, but I'd only do it if I already had the stuff lying around.

I wouldn't buy a card specifically for putting into that system.

regards, Gizbar.

---

A proud GPU User Server Donor!

ya im gonna let it crunch on its own for now and when seti uses ati cards ill have a bunch that would work i have never been a fan of nvidia hehe have had ati since 3dfx stopped making cards heh

that's what i was wondering if it would get the bigger ap wu in on time thanks for your help ill get it running after i get a power source and clean it up a bit :) at least then it will be of use

An Athlon XP 3000+ will have no problems meeting deadlines. I only crunch about 14 hours a day on mine and it has no issues at all.

It also does benefit nicely from optimized apps., especially for A/P.

---

The GTX 260 is now capable with CUDA 2.3 dll's + drivers to get over 10000 RAC
Huge speedup with 2.3 (I'd estimate 12k-14k RAC if left on 24/7 and with minimal problems and slowdowns)

Bob

---

Do you Good Search for Seti@Home? http://www.goodsearch.com/?charityid=888957
Or Good Shop? http://www.goodshop.com/?charityid=888957

@ Nojnagrom

I have the current top_host_#2.. ;-)
http://setiathome.berkeley.edu/top_hosts.php
AMD Phenom II X4 940 BE @ 4 x 3.0 GHz with 4 x OCed GTX260-216.

IIRC, my old 'AMD Athlon64 3200+' mobo have only a AGP slot.
So no CUDA crunching.

For every GPU one 'PCIe 1.0 x16' or 'PCIe 2.0 x8' slot on mobo should be available for max. CUDA crunching performance.

If it will be possible, I wouldn't crunch with an ATI GPU with heatsink without fan.
This GPUs have no big performance.
Like my old GeForce 6200, without fan and no CUDA possible.. ;-)

---

my 3000+ finishes AP v5 in about 100 hours using the opt app
it would be much worse using the standard app.

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

@ Nojnagrom

I have the current top_host_#2.. ;-)
http://setiathome.berkeley.edu/top_hosts.php
AMD Phenom II X4 940 BE @ 4 x 3.0 GHz with 4 x OCed GTX260-216.

IIRC, my old 'AMD Athlon64 3200+' mobo have only a AGP slot.
So no CUDA crunching.

For every GPU one 'PCIe 1.0 x16' or 'PCIe 2.0 x8' slot on mobo should be available for max. CUDA crunching performance.

If it will be possible, I wouldn't crunch with an ATI GPU with heatsink without fan.
This GPUs have no big performance.
Like my old GeForce 6200, without fan and no CUDA possible.. ;-)

Congratulations on reaching #2! I see some of your WUs are being aborted because of the use of the VLARkiller app. Have you looked at the Reschedule1.9 app available at Lunatics? This will move the VLAR tasks to the CPU instead of aborting them. This will help with your work shortage you have reported elsewhere also. You never know you might make it to #1 then ...............

---

GPU Users Group

@ Nojnagrom

I have the current top_host_#2.. ;-)
http://setiathome.berkeley.edu/top_hosts.php
AMD Phenom II X4 940 BE @ 4 x 3.0 GHz with 4 x OCed GTX260-216.
...

Congratulations on reaching #2! I see some of your WUs are being aborted because of the use of the VLARkiller app. Have you looked at the Reschedule1.9 app available at Lunatics? This will move the VLAR tasks to the CPU instead of aborting them. This will help with your work shortage you have reported elsewhere also. You never know you might make it to #1 then ...............

Sutaru isn't doing any CPU work, he's decided that having all 4 cores of his Phenom II instantly available at all times to tell the GPUs what to do next is more productive. Considering his host is ahead of others with an equal number of GPUs of similar or even better ratings, his judgement seems good. I don't know if he has tested all combinations of number of CPUs kept mostly idle, though.

                                                              Joe

Yes, I let run only CUDA WUs.

The CPU is only for support of the GPUs.

If you have a big WU cache.. you have very big [25 % CPU - 100 % Core] and long boinc.exe peaks in TaskManager.
Every ~ 5 sec. for ~ 5 sec.
This increase with much ULs/DLs.
Also the 'System' in TaskManager have high, up to ~ 13 % CPU.

If I would crunch also on the CPU, this boinc.exe/System peaks [normal priority] disturb all which have lower priority. CPU and GPU tasks.
Because of this, the GPUs would idle/stop from time to time.

I made the test with 4 CPU and 4 GPU tasks.
And the performance of the GPUs were very very bad.. maybe ~ 1/2 - 1/4 performance.

And the RAC of my GPU cruncher would be much much less than now.

After new nVIDIA_driver_190.38 and CUDA_V2.3 .dll's the performance increased ~ 30 %.

I guess the final RAC could be ~ 52,000 for my GPU cruncher.

Current : AR=0.44x/91Cr./10min./GPU - 13,100 RAC/GPU - 52,400 RAC/4 GPUs



Small calculation..

I calculate with the old short MB WUs on CPU.

Old MB AR=0.44x WU on CPU = ~ 60 min.
Same WU on GPU was 6m:45s , now longer WU [same AR].. ~ 10 min.

On the CPU it should be now ~ 120 min.

This mean the GPU is 12 x faster than one CPU Core.
And this are 4 x OCed GTX260-216 with AMD Phenom II X4 940 BE @ 4 x 3.0 GHz.



I would recommend for PCs with GTX2xx series:
4 GPUs - 4 thread CPU - only CUDA
4 GPUs - 8 thread CPU - 2, max. 4 CPU WUs.

Everybody should make the test and will find the best settings for his PC. AMD/Intel CPU - 8/9xxx /GTS/GTX GPU series.
Also I know the CPU WUs are faster on Intel.. ;-)

---

BTW.

Vyper with his current top_host_#1.. I guess he could reach ~ 100,000 RAC with his GPU cruncher.

Very very well.. the highest PC/RAC in the whole SETI@home history, or? :-)


Ahh.. now they are 3 AMDs @ 1st , 2nd and 3rd place.. ;-D

---

an i7 with 2 gtx 295's will get up there if it were dedicated to seti....

---

I recommend Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
Go Georgia Tech.

Have you tried with just one CPU in use by BOINC say.

Preferences 'On multiprocessors, use at most 25 % (instead of 100%) of the processors'.

You could move all the VLARs to be processed by this one core and you would not then run out of tasks when there is a shortage of work because they have been aborted. If you have no work then your RAC will suffer anyway and this would help stop you running out of work.

I don't have multi GPU machines so perhaps don't see the problems you have. I see only a small amount of CPU used to transfer data to and from the GPU - all procesing is done on the GPU.

---

GPU Users Group

I have two GTX 260-216 running with 7/8 "cores" of my i7 also crunching. My GPUs are about 2 minutes apart (suspended one of them), takes about 1m for the CPU to load the GPU, so that one spare CPU seems to be able to feed both GPU without any problems.