Advanced search

Message boards : Wish list : Choosing WU's

Author Message
Profile sTnatic
Send message
Joined: 14 Apr 09
Posts: 3
Credit: 52,564
RAC: 0
Level

Scientific publications
watwat
Message 9595 - Posted: 10 May 2009 | 16:43:44 UTC

I have just nVidia 9600GSO. A lot of people have old cards, and can't run BOINC 24/7. WUs like KASHIF_HIVPR have to be computed for 40hours for card like mine. It's hard to complete it in time. WUs like GIANNI, IBUCH give less credit but they are processing almost 2 times faster. I'd like to see option that allows me to choose which WUs I'd like to compute.

Profile DoctorNow
Avatar
Send message
Joined: 18 Aug 07
Posts: 83
Credit: 100,483,297
RAC: 2,536
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 9710 - Posted: 13 May 2009 | 12:48:15 UTC - in response to Message 9595.
Last modified: 13 May 2009 | 12:50:07 UTC

I'd like to see option that allows me to choose which WUs I'd like to compute.

I second that.
The KASHIFs are a pain for older cards. My 9600GT just finished a KASHIF with approx. 50 hours crunch time with - obviously - no extra credit. And I had to kick it three times manually, because it was on hold for some reason. Not to speak about the upload file of about 56 MB, took me 1 1/2 hour to upload with my slow line. ;-)
____________
Member of BOINC@Heidelberg and ATA!

Profile sTnatic
Send message
Joined: 14 Apr 09
Posts: 3
Credit: 52,564
RAC: 0
Level

Scientific publications
watwat
Message 9802 - Posted: 15 May 2009 | 16:44:10 UTC

But you can see that anybody isn't interested in it. Nobody of GPUGRID staff has written so far. End isn't that possible to make some use of checkpoints? If my task crushes at 95% - 20h it's kind of annoying. If I had a success checkpoint at 94% is there no way to load it after compute error?

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1919
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 9805 - Posted: 15 May 2009 | 18:16:55 UTC - in response to Message 9802.

For the larger units (not all the Kashifs are large), it was just a mistake.
We would like to finish the batch in order not to waste all the WUs computed so far.

Choosing WU is not possible, but we try to keep them as homogeneous as possible in terms of length and data upload.

gdf

Profile robertmiles
Send message
Joined: 16 Apr 09
Posts: 434
Credit: 493,761,146
RAC: 606,193
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 12043 - Posted: 22 Aug 2009 | 20:26:40 UTC - in response to Message 9805.
Last modified: 22 Aug 2009 | 20:36:14 UTC

duplicate message deleted

Profile robertmiles
Send message
Joined: 16 Apr 09
Posts: 434
Credit: 493,761,146
RAC: 606,193
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 12045 - Posted: 22 Aug 2009 | 20:35:10 UTC - in response to Message 9805.

For the larger units (not all the Kashifs are large), it was just a mistake.
We would like to finish the batch in order not to waste all the WUs computed so far.

Choosing WU is not possible, but we try to keep them as homogeneous as possible in terms of length and data upload.

gdf


I'd like to see the addition of what looks like ads to the bottom of threads modified so that it does not overwrite the last message or two in the thread. I couldn't read the above message until I started a reply quoting it.

I'd also like to see your procedure for deciding who gets what workunits modified so that the people who return workunits quickly are more likely to get the ones that take the most computation, and people with fast communications connections get the ones expected to take the most time downloading and uploading.

Post to thread

Message boards : Wish list : Choosing WU's