Allocation of BRP-Cuda WU's per day to increase ?

dunx
dunx
Joined: 13 Aug 10
Posts: 119
Credit: 53470527
RAC: 0
Topic 195536

Having fixed my W7 scheduler fiasco, I have reduced my BRPS_CUDA WU duration from 10 hours down to 50 minutes...

So I could process >=28 BRPS_Cuda WU's a day if the project would feed me...

Is this likely to be allowed in the future ?

Or should I spread my GTX 460 across another BOINC project ? Even if I'm not sure I could get it working.

dunx

archae86
archae86
Joined: 6 Dec 05
Posts: 3145
Credit: 7024924931
RAC: 1806140

Allocation of BRP-Cuda WU's per day to increase ?

How many are allowed now?

BRP3 is pretty new--my little flotilla of (non-CUDA) PCs has so far seen only one WU. Is it possible your low feeding rate is initial work shortage?

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686219689
RAC: 550782

I would expect the project to

I would expect the project to supply CUDA BRP tasks to satisfy demand once the problems with the new apps are fully worked out. Having said that, it's always wise to have a backup project configured.

CU
HBE

Jeroen
Jeroen
Joined: 25 Nov 05
Posts: 379
Credit: 740030628
RAC: 456

Is this 28 WU limit for each

Is this 28 WU limit for each host system or a limit for each CUDA enabled GPU? Are you getting a notice is the message log stating a limit has been reached? Thanks.

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686219689
RAC: 550782

RE: Is this 28 WU limit for

Quote:
Is this 28 WU limit for each host system or a limit for each CUDA enabled GPU? Are you getting a notice is the message log stating a limit has been reached? Thanks.

This 28 WU figure was the estimated max throughput if you assume 50 minutes per WU.

BOINC allows the project to define a limit for task download per day, which is based on the number of cores and GPUs and gets decreased if a host submits error results. This "quata" is meant to prevent unreliable clients to excessively drain the task pool without returning good results.

You will see a message about an exceeded quota in the messages tab of BOINC manager if it hits you.

CU
HBE

Ascholten
Ascholten
Joined: 20 Oct 10
Posts: 63
Credit: 2848108
RAC: 0

For what it's worth, if you

For what it's worth, if you find yourself out of Einstein Cuda apps milky way has been pretty reliable at feeding them recently. Im tagging between both of them to keep my GPU busy. After the money I just spend on the thing (we won't mention the power supply I cooked and had to replace) it better earn it's keep.

Aaron

If god meant for us not to BOINC he'd have made our #$%^%^ shorter!!

dunx
dunx
Joined: 13 Aug 10
Posts: 119
Credit: 53470527
RAC: 0

24 hours * 60 minutes / 50

24 hours * 60 minutes / 50 minutes per WU = 28.8 per day...

A week's worth of BRP WU's were done by 8AM this morning.

It may be a case of trying to get Milkyway to co-exist on my set-up, but I'm no BOINC expert...

I could do with a beginner's guide to setting up multiple projects to share resources.

( Just to add that I received 12 more BRPS WU this morning, so that's me happy till teatime... )

dunx

P.S. If I buy another 460 GTX in the sales, that would need feeding too !

P.P.S. My local problem is the mains supply, far too often my PC's are killed when the supply goes down, this week my Q6600 died of overwork/old age...

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

I am running 6 BOINC

I am running 6 BOINC projrcts, all equally shared, on my Linux box, no GPU, plus a VirtualBox running 2 virtual machines, one OpenSolaris Express and a SETI app, and the second is a BOINC project. BOINC 6.10.58 manages all this stuff with no problem, I rarely have to intervene.
Tullio

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.