So what is blocking the card? HDD cage? Is it a structural part of the case?
Printable View
All is well!!! :)
Nothing some pliers and a sheet metal cutter couldn't handle.
I was butting up against the drive bays (mostly unused), which were soldered to the case. Not very eloquent, but I had to do it else I wouldn't be able to sleep tonight.
Now to see what she can do. Take her out number 1...
Right on man! Lets see them numbers fly!
It's always good to get another mega cruncher going. If we had a few more people as dedicated, well......I guess I needn't say it....
Yo-
So, I'm finishing up a few tasks for moo! and gpugrid before going solo on PG.
Interesting, Moo will actually use both gpus for a single task and about .91 cpu.
But when it switches to gpugrid, only 1 gpu is used and .35 cpu, while the second gpu goes idle. I would think it would attach to another task.
I suspended Moo and gpugrid and noticed that 2 PG tasks were each using a gpu as expected.
weird.
Hmm, As a thought, try suspending any/all your CPU tasks and see if that makes your second GPU kick in on gpugrid.
Yo-
Welcome to the world of the MM cruncher... ;)
Moo and DNETC will both run only one task at a time, no matter now many GPUs you have. Currently, I have four GPUs in my system (2x5970) and run only 1 Moo task at a time (and one DNETC task when I was on that). And yeah, Moo CPU utilization is high - you'll probably have to dedicate 1 full core to it.
Yo hit upon an interesting issue I have yet to try - dedicating a full core to GPUGrid. I've heard it helps, but does NOT get GPU utilization up to near 100%. I have not tried that myself, though. I've also heard running 2 GPUGrid WUs/GPU helps (again, something I've not tried).
And when you say GPUGrid only runs on 1 GPU - does that mean you have 1 WU running on 1 GPU, and 1 WU not crunching while the other GPU sits idle? If so, that's not right. If you only have 1 GPUGrid WU downloaded, then what you described is normal.
Seriously, for the time being, crunch the crap out of PG. It's fun...:cool:
Okay, so now you have me confused.....are you saying that GPUGrid is still only running on 1 GPU or both after suspending all other tasks? Or, are you saying that PG runs on both GPUs while GPUGrid does not after suspending any/all other tasks?
Max hit on something also...is the GPUGrid project giving you more than 1 WU to crunch?
If you only have 1 in your que, then you can only crunch 1 at a time. If this is true, then maybe there's a way to, "fake out" the project into giving you more than WU 1 at a time, but I don't know about that yet.
I would personally like to know this info as I was planning on running 3 GPUGrids at once, on 1 box, but if I cannot do this, then they will fall by the way-side.
Yo-
Seems to be working now.
There was only a single gpugrid task to crunch. The problem was when boinc switched to the gpugrid task, it only used a single gpu while the other sat idle. I expected that other gpu to work on something else, PG for example.
Not sure what happened but now that same task (long wu) is working along side my PG task. So, both gpus are working now. Don't know what happened.
About the 'more than one wu at a time' question. How would I tell this? I just assumed 1 task/wu per line in the boinc manager.
Everything looks pretty normal now.
One thing I do notice is that gpugrid does not make very efficient use of my gpu compared to PG. PG pegs gpu 1 usage at 99%, while gpugrid has gpu 2 at around 30%.
One more note about GPUGRID: You get a credit bonus if for each task you return quickly. I think within 24 hours, which starts at D/L, NOT when you start crunching it. So you don't want to D/L a task and have it sitting in your queue burning the clock, until you are close to starting to crunch it. So keep a very short queue setting for any GPUGRID machines. I use .5 days or less.
Okay, straight forward question....Are you able to run 2 GPUGrid tasks at the same time, one running on each of your GPUs?
Sweet, I'm getting "tingly" all over.