• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Anyone phasing out a working GPU? Deal has been made!

Rudy Toody

Diamond Member
Fellow Crunchers--

My on-board chip is producing video artifacts. It is soon to expire🙁

If you are upgrading your system and want to pick up some cash, I need any NVIDIA GPU that has double precision floating point OR an ATI from this list:
23Fzi.png


In addition to crunching, I will be using it for Mathematica processes and maybe some C code development.

PM me with a price (including shipping.)

Thanks,

--Fred
 
Last edited:
I have a pair of GTX460 1GB OC (715Mhz) cards with the Gigabyte WindForce dual-fan cooler. Sadly, one fan out of each pair is going. Temps rose about 5C, still fairly acceptable temps under heavy GPU load. The one I currently have in the machine gets to 71C, in a 70F A/C room.
 
I have a pair of GTX460 1GB OC (715Mhz) cards with the Gigabyte WindForce dual-fan cooler. Sadly, one fan out of each pair is going. Temps rose about 5C, still fairly acceptable temps under heavy GPU load. The one I currently have in the machine gets to 71C, in a 70F A/C room.
My NVIDIA list shows only the GTS460. What's the difference between the GTS and GTX?

I don't know what any of the listed cards have, except they all have double precision floating-point.
 
My NVIDIA list shows only the GTS460. What's the difference between the GTS and GTX?

I don't know what any of the listed cards have, except they all have double precision floating-point.
i'm not sure where you got that nVidia list Fred, but the first 4 entries are nonexistent (there is no GTS 480, 470, 465, or 460...only the GTX 480, 470, 465, and 460). the different designations GTS and GTX refer to different GPU dies. the GTX 4xx series GPUs uses the fully enabled GF100 die and the almost fully enabled GF104 die, while the GTS 4xx series uses the even more disabled GF106 and GF108 dies. for more details, check out this nVidia GPU listing at Wiki and scroll down to the GeForce 400 Series chart. unfortunately that chart doesn't show FP64 (double precision) performance figures, let alone which GPUs in that series is FP64 capable in the first place. but rest assured that every GTX 4xx card (the 460, 465, 470, and 480) is FP64 capable, and will far outperform any of their GTS bretheren.
 
The list is from Wolfram Research. It is supposed to be double precision, openCL and CUDA.

I will go to NVIDIA to check them out.
 
Ok, my list is correct. Those GTS cards are OEM versions. The GTX cards don't have double precision floating-point or CUDA. The ATI cards don't have CUDA, but have the double-precision. All cards can do openCL.
 
Here's a list from Nvidia of cuda cards along with compute capability designation.

http://developer.nvidia.com/cuda/cuda-gpus

Cards with compute capability 1.3 or later (2.x) are capable of double precision floating point operations according to this document from Nvidia.

http://developer.download.nvidia.co.../Floating_Point_on_NVIDIA_GPU_White_Paper.pdf

The above document came from this list of Cuda developer information:

http://developer.nvidia.com/cuda/nvidia-gpu-computing-documentation

The cuda wiki also has some nice tables:

http://en.wikipedia.org/wiki/CUDA#cite_note-16
 
Thanks, biodoc. This looks like too much work!

But, I will wait a few days to see if anyone wants to get rid of an old gpu. Meanwhile, I have an old NVIDIA card ready for installation if the on-board chip fails completely.
 
Ok, my list is correct. Those GTS cards are OEM versions. The GTX cards don't have double precision floating-point or CUDA.

I've never ever heard nor seen of a GTS 480, 470, 465, or 460. Surely they would have popped up in the VC&G forum if they existed.

And I know for certain that the GTX variants of same DO have DP FP ops.

I can run MW@Home OpenGL on my GTX460, and it requires DP FP.
 
it was news to me too that those cards existed...but as OEM versions, they're not available to us anyways...unless you have access to Dell's (or some other major computer seller) inventory lol.
 
And I know for certain that the GTX variants of same DO have DP FP ops.

I can run MW@Home OpenGL on my GTX460, and it requires DP FP.

That is correct! My statement intended to mean: those gpu didn't have dp fp or they didn't have CUDA. Per biodoc's fine post above, they don't have CUDA. Also, per that post, CUDA causes such a production hit when using dp fp, that for my purposes, I would use openCL when I use dp fp.

I wanted a CUDA card because Mathematica uses CUDALink for best results. OpenCL is the best for using C++ to create code to be distributed to others, because you can do it for ATI, too.

Bottom line, disregard the CUDA. See the OP!
 
Last edited:
I'm not sure how much postage is San Jose -> Seattle, but I have a Sapphire HD5850 Toxic Edition that I'd be ok to let go to a happy home. Crossfire on a secondary system feels a little like overkill 🙂 I'll PM you some details...
 
well then VirtualLarry's GTX 460's should be more than sufficient for your needs. so should my GTX 560 Ti...but i don't know what your budget is, and a GTX 460 or a GTX 560 Ti might be overkill for your needs...unless you put it to work on the right DC project...then either could be a PPD monster!
 
I'm not sure how much postage is San Jose -> Seattle, but I have a Sapphire HD5850 Toxic Edition that I'd be ok to let go to a happy home. Crossfire on a secondary system feels a little like overkill 🙂 I'll PM you some details...

don't hold out on us Fred! what kind of GPU did you end up getting?

I took the one from jibberegg, who made me a deal I couldn't refuse!
 
very nice 😎...my 6950 is from the Sapphire Toxic series too, and i love it!

*EDIT* - any idea which project(s) you might put this bad boy to work on Fred?
 
Last edited:
That is correct! My statement intended to mean: those gpu didn't have dp fp or they didn't have CUDA. Per biodoc's fine post above, they don't have CUDA. Also, per that post, CUDA causes such a production hit when using dp fp, that for my purposes, I would use openCL when I use dp fp.

I wanted a CUDA card because Mathematica uses CUDALink for best results. OpenCL is the best for using C++ to create code to be distributed to others, because you can do it for ATI, too.

Bottom line, disregard the CUDA. See the OP!

I agree, OpenCL seems like the way to go. 🙂
 
OK, I got the card. Had to move my ethernet card to make room. Got everything up and running. POP! My 1250W psu died. Replaced it with a 650W. Recovered the zapped tracks on the HD. Everything runs except the internet. Troubleshooting indicates that the ethernet card was also zapped. Got a new one today. Everything is Hunky-Dory!

Boinc doesn't recognize the gpu. Is there anything special I need to do?
 
Boinc doesn't recognize the gpu. Is there anything special I need to do?
are you familiar with using a cc_config.xml file? if so, try adding the following line of code to it:

Code:
<use_all_gpus>1</use_all_gpus>
if you aren't familiar w/ that file, open notepad (not MS Word or any other text editor, as they tend to add invisible characters from time to time), and add the following lines of code to it:
Code:
<cc_config>
<options>

<use_all_gpus>1</use_all_gpus>

</options>
</cc_config>
save the file, and then change the .txt extension to .xml. then place the file in your BOINC data directory, restart BOINC, and see if the new GPU is recognized.
 
Last edited:
Back
Top