NVIDIA GPU 0: GeForce GTX 295 (driver version 0, compute capability 1.3, 895MB, est. 106GFLOPS)
NVIDIA GPU 0: GeForce GTX 275 (driver version 0, compute capability 1.3, 896MB, est. 120GFLOPS)

---

11.10.2009 15:37:01 ATI GPU 0: ATI Radeon HD 3800 (RV670) (CAL version 1.3.186, 512MB, 509GFLOPS)
no CUDA ;-(

BOINC 6.10.13 version

http://boinc.berkeley.edu/download_all.php

11.10.2009 15:37:01 ATI GPU 0: ATI Radeon HD 3800 (RV670) (CAL version 1.3.186, 512MB, 509GFLOPS)
no CUDA ;-(

BOINC 6.10.13 version

http://boinc.berkeley.edu/download_all.php

6.10.13 wouldn't help him.. if SETI@home have no ATI app..

Maybe he meant that he have no nVIDIA GPU?

;-)


BTW.
My 4 manufacturer OCed GTX260-216 have 112 GFLOPS/each.

If the ATI GPUs have ~ 1,000+ GFLOPS, if an ATI app would exist - the calculation time would be ~ 10x faster? Woohoo..!

---

NVIDIA GPU 0: GeForce GTX 295 (driver version 0, compute capability 1.3, 895MB, est. 106GFLOPS)
NVIDIA GPU 0: GeForce GTX 275 (driver version 0, compute capability 1.3, 896MB, est. 120GFLOPS)

275 faster than 295? Is it OCed?

---

NVIDIA GPU 0: GeForce GTX 295 (driver version 0, compute capability 1.3, 895MB, est. 106GFLOPS)
NVIDIA GPU 0: GeForce GTX 275 (driver version 0, compute capability 1.3, 896MB, est. 120GFLOPS)

275 faster than 295? Is it OCed?

Don't know about this particular case, but that's the way they're sold at my local computer store.

The GTX 275 has a faster clock speed. With the GTX 295, you get two of them in the same box, but they run slower and cost more than twice as much.

BTW.
My 4 manufacturer OCed GTX260-216 have 112 GFLOPS/each.

If the ATI GPUs have ~ 1,000+ GFLOPS, if an ATI app would exist - the calculation time would be ~ 10x faster? Woohoo..!

According to the NVidia marketing people quoted at GPUGrid, a standard GTX260/216 runs at 804.80 GFlops. I suspect ATI measure their cards using the same scale.

That would almost a thousend times faster then a P4 CPU! Or 300 -500 times the todays CPU !
sure hope that isn't too much, without errors, I mean.

Hi, stil looking for some suitable CUDA Capable Cards, use only one
a 9800GTX, the 8500GT in my VISTA(Home Premium;32BIT) I don't even use.

The 9800GTX is in WIN XP64 Pro cruncher, MoBo with X38 Chipset & QX9650 CPU @ 3,42 GHz.
Odd work-fetch in BOINC 6.6.36 :

12-10-2009 10:14:14 SETI@home Reporting 2 completed tasks, requesting new tasks for GPU
12-10-2009 10:14:20 SETI@home Scheduler request completed: got 0 new tasks
12-10-2009 10:14:20 SETI@home Message from server: (Project has no jobs available)
12-10-2009 10:15:32 SETI@home Sending scheduler request: To fetch work.
12-10-2009 10:15:32 SETI@home Requesting new tasks for GPU
12-10-2009 10:15:38 SETI@home Scheduler request completed: got 11 new tasks
12-10-2009 10:15:40 SETI@home Started download of 11se09ae.3864.4162.12.10.63
12-10-2009 10:15:40 SETI@home Started download of 02se09ae.18287.72.9.10.249
12-10-2009 10:15:45 SETI@home Finished download of 11se09ae.3864.4162.12.10.63
12-10-2009 10:15:45 SETI@home Finished download of 02se09ae.18287.72.9.10.249
12-10-2009 10:15:45 SETI@home Started download of 11se09ae.3864.4162.12.10.39
12-10-2009 10:15:45 SETI@home Started download of 08se09ac.12644.481.14.10.170
12-10-2009 10:15:50 SETI@home Finished download of 11se09ae.3864.4162.12.10.39
12-10-2009 10:15:50 SETI@home Finished download of 08se09ac.12644.481.14.10.170
12-10-2009 10:15:50 SETI@home Started download of 02se09ae.18287.72.9.10.248
12-10-2009 10:15:50 SETI@home Started download of 02se09ae.18287.72.9.10.254
12-10-2009 10:15:54 SETI@home Sending scheduler request: To fetch work.
12-10-2009 10:15:54 SETI@home Requesting new tasks for GPU
12-10-2009 10:15:55 SETI@home Finished download of 02se09ae.18287.72.9.10.248
12-10-2009 10:15:55 SETI@home Finished download of 02se09ae.18287.72.9.10.254
12-10-2009 10:15:55 SETI@home Started download of 11se09ae.3864.4162.12.10.84
12-10-2009 10:15:55 SETI@home Started download of 08se09ac.12644.481.14.10.167
12-10-2009 10:15:59 SETI@home Scheduler request completed: got 6 new tasks
And so on, absolutely NOTHING against it ;)
And so I'll never be out of work.
For the time being, that is. ;^)

Still wandering why SETI GPU crunching takes the CUDA card almost twice
the current(=Watt's), compaired to GPU-grid crunching, only by far more computations/sec and/or threads?

CUDA= 9800GTX; 512MByte; compute capability 1.1; est. 85 GFLOPS; 2.3CUDA Lib & Drivers{190.38}

[OFF TOPIC]Google Earth, tempted me to download a BETA version which woudn't obviously be compatible with XP64 or the 2.3 CUDA Dll's?

---

This thread prompted me to finally download EVGA Precision to oc my card.
So a lowly 8800 GTS 512 ended up with 91 GFLOPS (up from 77):

CUDA device: GeForce 8800 GTS 512 (driver version 19062, compute capability 1.1, 512MB, est. 91GFLOPS)

---

NVIDIA GPU 0: GeForce GTX 295 (driver version 0, compute capability 1.3, 895MB, est. 106GFLOPS)
NVIDIA GPU 0: GeForce GTX 275 (driver version 0, compute capability 1.3, 896MB, est. 120GFLOPS)

275 faster than 295? Is it OCed?

Don't know about this particular case, but that's the way they're sold at my local computer store.

The GTX 275 has a faster clock speed. With the GTX 295, you get two of them in the same box, but they run slower and cost more than twice as much.

I guess his PC [http://setiathome.berkeley.edu/show_host_detail.php?hostid=5011059] have 2x GTX295 and 1x GTX275.
Every GTX295 have 2x 106 GFLOPS = 212 GFLOPS/whole GTX295.
So GTX295= 212 GFLOPS and GTX275= 120 GFLOPS

Hmm..
To now nobody sold OCed GTX295.
Since the one PCB style.. for example EVGA sell SC and FTW editions:
[http://www.evga.com/articles/00494]

---

BTW.
My 4 manufacturer OCed GTX260-216 have 112 GFLOPS/each.

If the ATI GPUs have ~ 1,000+ GFLOPS, if an ATI app would exist - the calculation time would be ~ 10x faster? Woohoo..!

According to the NVidia marketing people quoted at GPUGrid, a standard GTX260/216 runs at 804.80 GFlops. I suspect ATI measure their cards using the same scale.

Thanks for the URL! :-)

AFAIK, a stock GTX260-216 have 106 GFLOPS. This mean ~ 805 GFLOPS.
My OCed with 112 GFLOPS should have then mathematically ~ 851 GFLOPS.


Just curious..
A GTX285 have 1,063 GFLOPS.
A 9800GX2 have 2x 576= 1,152 GFLOPS.
So it would be better to go with 9800GX2 instead of GTX285?
But, after short look.. nobody sell the old 9800GX2 in Germany. :-(

---

NVIDIA GPU 0: GeForce GTX 295 (driver version 0, compute capability 1.3, 895MB, est. 106GFLOPS)
NVIDIA GPU 0: GeForce GTX 275 (driver version 0, compute capability 1.3, 896MB, est. 120GFLOPS)

275 faster than 295? Is it OCed?

Don't know about this particular case, but that's the way they're sold at my local computer store.

The GTX 275 has a faster clock speed. With the GTX 295, you get two of them in the same box, but they run slower and cost more than twice as much.

I guess his PC [http://setiathome.berkeley.edu/show_host_detail.php?hostid=5011059] have 2x GTX295 and 1x GTX275.
Every GTX295 have 2x 106 GFLOPS = 212 GFLOPS/whole GTX295.
So GTX295= 212 GFLOPS and GTX275= 120 GFLOPS

Yes, that makes sense. Thanx, Sutaru!

---

CUDA device: GeForce GTS 250 (driver version 19107, compute capability 1.1, 1024MB, est. 84GLOPS
CUDA device: GeForce GTS 250 (driver version 19107, compute capability 1.1, 1024MB, est. 84GLOPS

The odd thing is that I actually have a 260 and a 250, it appears something is mislabling or perhaps misidentifing my cards.

I also have see ppl talk about the new .DLLs ,I installed using the lunatics installer. Do I have them ?

---

And after the spanking comes...



Ni !

I also have see ppl talk about the new .DLLs ,I installed using the lunatics installer. Do I have them ?

No, they are not part of the installer. Get 'em here and copy to your SETI-projects folder.

---

Where would I get the new .dlls? and are they neccesary for improved performance?

---

And after the spanking comes...



Ni !

OPs, didn't see the link. Thanx. I'll do it when kwsn comes back online

---

And after the spanking comes...



Ni !

GeForce GT 120 (driver version 19107, compute capability 1.1, 1024MB, est. 15GFLOPS)

---

Where would I get the new .dlls? and are they neccesary for improved performance?

AFAIK, if the GPUs which are insert in the PC case are 'very' different, only one of them are used.
Or like this.. or similar.. or other.. ;-)

To make sure all GPUs are used, you could make a cc_config.xml with:
<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
</options>
</cc_config>

[http://boinc.berkeley.edu/wiki/Client_configuration]


Also you could look, if the tasks lists are available again ;-( , how the result output look.
Also in your TaskManager how much CUDA_apps are running.


Also be aware, that you need nVIDIA_driver_190.38+ for CUDA_V2.3 .
CUDA_V2.2 to CUDA_V2.3 -> ~ 30 % speed up.

---

CUDA device: GeForce GTS 250 (driver version 19107, compute capability 1.1, 512MB, est. 84GFLOPS) 

It's interesting that the 512MB version has the same estimated speed as the 1024MB GTS 250...

CUDA device: GeForce GTS 250 (driver version 19107, compute capability 1.1, 512MB, est. 84GFLOPS) 

It's interesting that the 512MB version has the same estimated speed as the 1024MB GTS 250...

The amount of RAM doesn't affect the raw computational ability of the cards. In a situation where the task required more than 512MB of memory, the 1GB card would have a performance advantage as it could keep everything in its own RAM instead of having to page in and out of main memory.

---

CUDA device: GeForce 9800 GT (driver version 19107, compute capability 1.1, 512MB, est. 68GFLOPS)Zotac 9800 GT AMP edition Factory OC'd 700/1700/1000

---