Message boards : Graphics cards (GPUs) : BOINC 6.6.31 is out
| Author | Message |
|---|---|
|
https://boinc.berkeley.edu/dl/boinc_6.6.31_windows_intelx86.exe | |
ID: 10309 | Rating: 0 | rate:
 /   Reply
Quote
| |
 Beyond Send message Joined: 23 Nov 08 Posts: 1112 Credit: 6,162,416,256 RAC: 0 Level  Scientific publications    | |
|
Anyone know what's changed? | |
| ID: 10313 | Rating: 0 | rate:
/ Reply
Quote
| |
|
MarkJ  Volunteer moderator Volunteer tester Send message Joined: 24 Dec 08 Posts: 738 Credit: 200,909,904 RAC: 0 Level  Scientific publications    | |
Anyone know what's changed? Ageless posted this on the BOINC message boards under the BOINC 6 Versions change log. From reading the BOINC alpha mailing list there is a missing file under the new News Tab (which isn't meant to be there yet). I am not sure if its 6.7.5 or the 6.6.31 version that has this. There is supposed to be a couple of cuda related things in this version so might be worthwhile testing. ____________ BOINC blog | |
| ID: 10317 | Rating: 0 | rate:
/ Reply
Quote
| |
|
MarkJ Volunteer moderator Volunteer tester Send message Joined: 24 Dec 08 Posts: 738 Credit: 200,909,904 RAC: 0 Level Scientific publications | |
From reading the BOINC alpha mailing list there is a missing file under the new News Tab (which isn't meant to be there yet). I am not sure if its 6.7.5 or the 6.6.31 version that has this. I installed it on a couple of machines tonight. No sign of a news tab in this one. So far so good. Seems to have fixed the screen flicker under windows. Will see how it goes for a few more days before sticking it on my other machines. ____________ BOINC blog | |
| ID: 10323 | Rating: 0 | rate:
/ Reply
Quote
| |
|
MarkJ Volunteer moderator Volunteer tester Send message Joined: 24 Dec 08 Posts: 738 Credit: 200,909,904 RAC: 0 Level Scientific publications | |
|
Well i'm told there are no changes to the screen drawing so it should be the same as before. Also others are telling me theirs still flickers. | |
| ID: 10325 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level  Scientific publications | |
|
The main CUDA fix is most likely to only significantly help on SaH where you have high task turbulence with new tasks coming down with earlier deadlines than already running tasks. I am not saying that it will not affect us here, just that I don't think I have seen that issue here... | |
| ID: 10338 | Rating: 0 | rate:
/ Reply
Quote
| |
|
MarkJ Volunteer moderator Volunteer tester Send message Joined: 24 Dec 08 Posts: 738 Credit: 200,909,904 RAC: 0 Level Scientific publications | |
The main CUDA fix is most likely to only significantly help on SaH where you have high task turbulence with new tasks coming down with earlier deadlines than already running tasks. I am not saying that it will not affect us here, just that I don't think I have seen that issue here... Of course there will be another one tomorrow, thats half the fun of living on the leading edge :) I haven't managed to reproduce my crash, despite a few attempts to try and get it to do it again. As for cuda tasks getting preempted that will happen if you have more than one cuda project on the same machine and the tasks need to be swapped because of the TSI. ____________ BOINC blog | |
| ID: 10358 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Paul D. Buck Send message Joined: 9 Jun 08 Posts: 1050 Credit: 37,321,185 RAC: 0 Level Scientific publications | |
The main CUDA fix is most likely to only significantly help on SaH where you have high task turbulence with new tasks coming down with earlier deadlines than already running tasks. I am not saying that it will not affect us here, just that I don't think I have seen that issue here... Actually, CUDA work is run in EDF mode meaning if a task comes down that has an earlier deadline than the one running, the one running gets preepmpted. Even if the TSI is not met yet. Sad, but true ... this is one of the areas I have been banging my head on the wall in that the developers and JM VII cannot see that most of the time on newer systems this is not needed. For either GPU tasks or CPU tasks ... Bank Teller rules should apply ... But I cannot make them see this ... sadly ... In a perfect world there would be no risk to suspending a task ... sadly we are not in that perfect world and there is lots of risk to doing so in BOINC... many of the "bad things" that happen would not be happening if the Resource Scheduler would not arbitrarily preempt tasks without greater need... Anyway, the reason SaH is so prone to this is that tasks come down all the time with earlier deadlines than the tasks already in hand ... so ... preemption happens a lot more there ... | |
| ID: 10393 | Rating: 0 | rate:
/ Reply
Quote
| |
|
MarkJ Volunteer moderator Volunteer tester Send message Joined: 24 Dec 08 Posts: 738 Credit: 200,909,904 RAC: 0 Level Scientific publications | |
|
Well Paul your wish came true. 6.6.33 is now out, they didn't release 32. I'll start a new message thread for the latest and greatest :-) | |
| ID: 10410 | Rating: 0 | rate:
/ Reply
Quote
| |
Message boards : Graphics cards (GPUs) : BOINC 6.6.31 is out
