huge credit/time ratio disparity between short/long WUs

Message boards : Number crunching : huge credit/time ratio disparity between short/long WUs

Author	Message
JLConawayII Send message Joined: 31 May 10 Posts: 48 Credit: 28,893,779 RAC: 0 Level Scientific publications	Message 23067 - Posted: 22 Jan 2012 \| 0:52:10 UTC Last modified: 22 Jan 2012 \| 1:08:05 UTC
	I'm a bit baffled by what I'm seeing in my completed tasks list. I have 3 long run WUs and 5 standard ACEMD2 units, all of which took between 30-38k seconds to complete on my GTX 260 machine. However, even though they all took around the same time to finish, I only earned 7-13k credits for the short WU's, vs 30-35k for the long runs. They all had a <48h turnaround, so they should have bonus credits factored in. Even so, should the long runs be worth 3x the credits for the same run time? Or a better question might be, should the standard WUs be taking anywhere near that long to complete in the first place?
	ID: 23067 \| Rating: 0 \| rate: / Reply Quote

skgiven Volunteer moderator Volunteer tester Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level Scientific publications	Message 23068 - Posted: 22 Jan 2012 \| 1:38:58 UTC - in response to Message 23067.
	On your system NATHAN tasks tend to take between 32K and 38K sec. These are released as Long runs (8-12 hours on fastest card) v6.15 (cuda31). Some supposedly shorter tasks take around the same time, for example, KASHIF_HIVPR and TONI_AGGMI, both use ACEMD2: GPU molecular dynamics v6.15 (cuda31) ____________ FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help
	ID: 23068 \| Rating: 0 \| rate: / Reply Quote

JLConawayII Send message Joined: 31 May 10 Posts: 48 Credit: 28,893,779 RAC: 0 Level Scientific publications	Message 23070 - Posted: 22 Jan 2012 \| 2:00:01 UTC
	I will just deactivate short runs then, if they aren't going to give credit for the amount of work done.
	ID: 23070 \| Rating: 0 \| rate: / Reply Quote

raTTan Send message Joined: 17 Mar 11 Posts: 7 Credit: 28,985,881 RAC: 0 Level Scientific publications	Message 23429 - Posted: 12 Feb 2012 \| 23:46:42 UTC - in response to Message 23068. Last modified: 12 Feb 2012 \| 23:51:29 UTC
	Your reply didn't answer the OPs question as to why the credit/time is so much lower for short runs. Is it intentional that there is such disparity? I came to notice this issue as well and further noticed that many people seem to have the short runs disabled, I can only assume for this reason. Are the people in charge here trying to discourage people from running short runs for some reason?
	ID: 23429 \| Rating: 0 \| rate: / Reply Quote

skgiven Volunteer moderator Volunteer tester Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level Scientific publications	Message 23434 - Posted: 13 Feb 2012 \| 2:32:42 UTC - in response to Message 23429.
	There was a lean towards over-rewarding for long tasks as these benefit the project the most. With the addition of badge systems and the need to facilitate at least one project that requires the 'normal' run time (steps) to remain the same, for scientific reasons, a suggestion has been made that will hopefully go some way towards addressing the concerns if it's implemented. In the long run it's anticipated that most projects will have both short and long tasks, so people with average cards can contribute to the same projects as people with high end cards. ____________ FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help
	ID: 23434 \| Rating: 0 \| rate: / Reply Quote

Post to thread

Message boards : Number crunching : huge credit/time ratio disparity between short/long WUs

	About	Science	Volunteers	Performance	Forum	Join us	Donate