Why does CreditFew Award Fewer Credits when my GPU is matched with a CPU than with another GPU?

Message boards : Number crunching : Why does CreditFew Award Fewer Credits when my GPU is matched with a CPU than with another GPU?
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5

AuthorMessage
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1935883 - Posted: 16 May 2018, 12:49:19 UTC - in response to Message 1933857.  


I don't care What the number is, huge, tiny, whatever. I just want it to accurately reflect some type of reality, and to be consistent, repeatable, and useful for comparisons purpose. To know the differences between different hardware, or between OS's, or even to indicate if something is configured wrong or broken. To me, that would be the end game for Credit Done.


. . Exactly, credit given is useless when on any given system working at its configured optimum the numbers can go up and down on a daily basis like a fiddlers elbow. In the end it means absolutely nothing. It needs that consistency which CreditScew lacks.

Stephen

:(
ID: 1935883 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22190
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1935886 - Posted: 16 May 2018, 12:54:16 UTC

It does, or should, work the same for aborted tasks

Currently "aborted" tasks are not considered as "Error" tasks, in the same way as "Download Failures" don't count against the user. Like you, I feel that "aborted" tasks should count against the user and result in a reduction in the daily allowance of tasks delivered in a cumulative manner - Something along the lines "abort x tasks, allowance reduced by y%" until back to "free flow", which can only be achieved if there are no more aborted (or error) tasks during the recovery period, more tasks aborted resulting in a reduction in the allowance, ending at one task per day until you stop.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1935886 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22190
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1935887 - Posted: 16 May 2018, 13:00:40 UTC

Essentially credit is awarded based on CPU performance. The number of "FLOPS" is estimated on a fairly simple calculation for a CPU, then scaled for a GPU. This scaling assumes that all tasks are the "same type", and thus the same scaling can be applied - and as we know, this is not correct, particularly for VLAR tasks that take proportionately longer on (nVidia) GPUs than CPUs, (and, I think, VHAR actually take proportionately less time on GPUs than CPUs).
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1935887 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1935892 - Posted: 16 May 2018, 13:21:30 UTC

I hope that if you folks come up with a proper solution to the CreditNew downward spiral that you can find somebody to code it and put in the fix.
The kitties are not holding their breath, but would welcome the repair.
It is rather disappointing when one upgrades a rig and is rewarded with it's RAC declining......LOL.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1935892 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5

Message boards : Number crunching : Why does CreditFew Award Fewer Credits when my GPU is matched with a CPU than with another GPU?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.