I've been reading around on the forums for the past fortnight or so having just got boinc (with einstein@home) back running on my pc.
My current pc has an 8800 GT and is CUDA capable afaik, it seems to run the seti@home cuda app fine, my question is how do I definitivly know if einstein@home is using my GPU?
I have looked for an answer and in general all i can ascertain is that if I'm running > version x.x.x it just works, but how do I check this? For instance seti@home has a very conveyant "cuda" in the taks description.
for refrence:
BOINCE v 6.6.38
Microsoft windows 7 64 bit
nvidia 191.07
apoligies if this has been covered and I somhow missed the information.
Copyright © 2024 Einstein@Home. All rights reserved.
How to check if CUDA is actually working?
)
Shouldn't there be something like "Running (0.x CPU, 1 CUDA)" in the Status column of the Tasks tab? I can't tell for sure, because I don't have CUDA.
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
RE: My current pc has an
)
You'd have to install the application by hand, from the Beta Applications page. You'd have to exit BOINC, unzip the information in the zip file into your BOINCData\projects\einstein.phys.uwm.edu\ directory and restart BOINC.
All quite memorable things to do.
Einstein isn't giving out CUDA work through their own scheduler yet, such as Seti does.
Not in their tasks description, but in the application description. Just as The Einstein application has the description in the science application's name. As does any other project if they give out the same application version for CPU and GPU.
And else it'll show on their applications page.
Thanks Ageless. I must
)
Thanks Ageless.
I must have missed that the cuda app is still beta.
Hi Jord, I think I have
)
Hi Jord,
I think I have something off the same.
That is what going wrong at my place, after deleting the app_info.xml file I guess.
Everything is running fine now with regularly new work. S5R6a work and ABP 3.11 work.
However now new CUDA work, but some messages:
19/10/2009 18:32:19 Einstein@Home Sending scheduler request: To fetch work.
19/10/2009 18:32:19 Einstein@Home Requesting new tasks for GPU
19/10/2009 18:32:23 Einstein@Home Finished upload of h1_1094.60_S5R4__1599_S5R6a_0_0
19/10/2009 18:32:24 Einstein@Home Scheduler request completed: got 0 new tasks
19/10/2009 18:32:24 Einstein@Home Message from server: No work sent
19/10/2009 18:33:24 Einstein@Home Sending scheduler request: To fetch work.
19/10/2009 18:33:24 Einstein@Home Reporting 1 completed tasks, requesting new tasks for GPU
19/10/2009 18:33:29 Einstein@Home Scheduler request completed: got 0 new tasks
19/10/2009 18:33:29 Einstein@Home Message from server: No work sent
Requesting GPU, that is the graphics board and thus CUDA?
Did I understand this wrong again?
Thanks.
Greetings from
TJ
RE: Requesting GPU, that is
)
CUDA is the software that makes work NVIDIA GPUs.
Once the server sends you work you will have "Running (0.x CPU, 1 CUDA)"
Right now I have two units running on my GTX260 and GT 9800 GPUS. Actually it's " 1 CPU and 1 GPU" each.
Einstein CUDA is Beta, so might be still have problems running... but if we don't test it we will never make it work :o)
HOW TO - Full installation Ubuntu 11.10
If my GTX275 is now doing
)
If my GTX275 is now doing Einstein work why aren't the work units being cranked out faster than before? It says on an X3360 four core cpu,cpu,cpu,cpu+gpu but they are still at about 5 hours each which is the same as without the GTX275.
Robert
Robert In Jan. the newer
)
Robert
In Jan. the newer ABP2 thingies have been optimised for CUDA GPUs and are taking 55 minutes now compared to 5 hours or so for ABP1s. (X3360, 4gb, GTX275)
GPU-Z showed that my GTX275 Nvidia GPU was used little on ABP1s but 55% on ABP2s. That's not what I bought it for but things are moving in the right direction.
Robert