Welcome to MilkyWay@home

ATI application updated again

Message boards : News : ATI application updated again
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 . . . 5 · Next

AuthorMessage
Matt Arsenault
Volunteer moderator
Project developer
Project tester
Project scientist

Send message
Joined: 8 May 10
Posts: 576
Credit: 15,979,383
RAC: 0
Message 47532 - Posted: 11 Apr 2011, 1:52:35 UTC

I've updated the new ATI application to 0.59. This fixes the 100% CPU problem, and adds a command line flag -r or --responsiveness-factor . This number just multiplies the estimate for how long things are expected to take to decide how to break things into smaller pieces to allow the screen a chance to redraw. Numbers greater than 1 should make things more responsive. 0 ignores any need for responsiveness for maximum speed.

I'm not exactly happy with how it decides to break the problem down to keep the screen responsive; I think how it's done now can still cause slowdowns on some GPUs (and is kind of fragile), and you shouldn't have to do anything to keep things responsive. I'm not really sure the best way to do it. If you feel you actually need to use this, complaints here with what GPU you have might be useful.

ID: 47532 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JAMC

Send message
Joined: 9 Sep 08
Posts: 96
Credit: 336,443,946
RAC: 0
Message 47533 - Posted: 11 Apr 2011, 1:59:04 UTC

Are the old .57 WU's going to get credit or do we abort them... they show as 'Not in DB' under Application.
ID: 47533 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Matt Arsenault
Volunteer moderator
Project developer
Project tester
Project scientist

Send message
Joined: 8 May 10
Posts: 576
Credit: 15,979,383
RAC: 0
Message 47534 - Posted: 11 Apr 2011, 2:01:02 UTC - in response to Message 47533.  

Are the old .57 WU's going to get credit or do we abort them... they show as 'Not in DB' under Application.
Just the application changed. The workunits aren't actually associated with the version so it should work fine.
ID: 47534 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile arkayn
Avatar

Send message
Joined: 14 Feb 09
Posts: 999
Credit: 74,932,619
RAC: 0
Message 47535 - Posted: 11 Apr 2011, 2:19:48 UTC
Last modified: 11 Apr 2011, 2:31:39 UTC

Will the command line options from 0.23 work with the new version?


One can configure the app through some parameters in the supplied app_info.xml.
This is done by editing the line

<cmdline></cmdline>

There are several options one can put in there, the order does not matter. If the same argument
is given twice, the last one counts. Remember that for editing the app_info.xml one should take
the same precautions as for the installation of the application, i.e. one should stop BOINC
completely and restart it afterwards.

Options:

kernel frequency: f (default 30)
The application determines the size of the work packages sent to the GPU in a dynamic manner. It
tries to get the number of executed packages per second above the value specified here by
splitting them to smaller ones. The OS is completely blocked from accessing the graphics card for
the duration of one package leading to a somewhat sluggish behaviour of the user interface.
Limiting the size and therefore reducing the execution time per package creates more opportunities
for the OS to slip in between two work packages and to react to user input. The default value of
30 Hz is tuned for usability of the system. Reducing it could increase the throughput of the app
slightly.

Example:
<cmdline>f10</cmdline>
The app will try to run 10 work packages per second (limits the execution time to 100ms or shorter).


Wait factor: w (default 1.0)
The app tries to release the CPU during the the GPU computations. It does so by predicting
the runtime for the GPU calls and send the CPU thread to sleep for that time. One can manually
correct the prediction of the application with this factor. When raising it, the CPU thread
sleeps longer, decreasing the value will lead to some faster polling. Setting this value to 0.0
turns off the release of the CPU. Default is of course a value of 1.0.
If you see a low GPU load, you can try to decrease it (if you have a load of 80%, set it to 0.8).
If you see a high CPU load, a slight increase of this value may help. Increase it too much and it
will affect the crunching speed. You could use this for throttling of the GPU in case the high
GPU load leads to a very sluggish behaviour of the user interface or even VPU recover events.
Setting w1.1 could improve the situation (see also the f, b, and p options).

Example:
<cmdline>w1.3</cmdline>
This would increase the sleep time by 30% in relation to the prediction.


Priority: p (default 2)
All BOINC applications normally run with the lowest possible priority to not disturb other
applications. This can lead to a low GPU load, as it may be not possible to fire up the next tasks
if all cores of the CPU are under load. Raising the priority may help here. BOINC recommends a
slightly increased priority for GPU applications. Default priority is normal. Possible values:
p0: idle priority, used by CPU BOINC applications
p1: normal priority in idle priority class (below normal), this is recommended for BOINC GPU
applications, but apears to be not enough to enable millisecond polling of the GPU with Vista
p2: normal priority in normal priority class, the default
p3: normal priority in high priority class, use with care!
The priority will affect how much time it takes for the the app to get back control if it releases
its time slice (see also option b).

Example:
<cmdline>p3</cmdline>
This raises the priority to high (not recommended).


Polling behavior for the GPU within the Brook runtime: b (default 0)
See the option w for starters. If that time has elapsed, the GPU polling starts. This can be done
by continously checking if the task has finished (b-1), enabling the fastest runtimes, but potentially
creating a high CPU load (a bit dependent on driver version). Second possibility is to release the time
slice allotted by the OS, so other apps can run (b0). This is the default value. The catch is that there
is some interaction with the priority. The time slice is only released to other tasks of the same
priority. So raising the priority effectively disables the release and the behavior is virtually
identical to setting this parameter to -1. If a raised priority and a low CPU time is wanted, one may
set it at to 1. This suspends the task for at least 1 millisecond, enabling also tasks of lower priority
to use the CPU in the meantime. This can also help to get a smoother system behaviour. One can use b2 or
b3 if one wants reduce the polling frequency even more.
Possible values:
b-1: busy waiting
b0: release time slice to other tasks of same priority (default)
b1, b2 or b3: release time slice for at least 1, 2, or 3 milliseconds respectively
See also options p and w.

Example:
<cmdline>b-1</cmdline>
Enable busy waiting (no release of the time slice during polling).


The options can be combined in arbitrary order. Therefore the following argument list is also valid:
<cmdline>f40 w1.1 p1 b1</cmdline>
(frequency 40Hz, wait factor 1.1, priority set to below normal, and CPU friendly polling of the GPU)

ID: 47535 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Zydor
Avatar

Send message
Joined: 24 Feb 09
Posts: 620
Credit: 100,587,625
RAC: 0
Message 47536 - Posted: 11 Apr 2011, 2:22:46 UTC
Last modified: 11 Apr 2011, 2:24:57 UTC

2x5970s (790/500) on a 1090T @3.7Ghz - first thoughts after a few.

I am running some PRPNet on the CPUs - same as I was with 0.23 - so its a sort of level playing field, as such.

The GPU utilisation appears to be way down - runs at about 50-60%. Heat is no issue, probably because of the low utilisation, but its the coolest I've ever seen with MW GPU WUs.

I had no screen lag at all, and also loaded a couple of browsers with no noticeable effect re lag - so that seems fine, but bare in mind its against 2x5970s....

Overall speed compared to 0.23 seemed to be about 25 - 30% slower, I ran it at the same gpu/memory settings as I used for 0.23 so a comparison could be seen. It maybe the new app needs higher gpu/memory, dont know yet.

Looks like its still cpu bound(ish) as useage is way higher than 0.23, at least with 5970s set with CPUs running PRPNet (PRPNet took a 30% hit slowing down while 0.57 was running).

Regards
Zy
ID: 47536 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Beyond
Avatar

Send message
Joined: 15 Jul 08
Posts: 383
Credit: 729,293,740
RAC: 0
Message 47537 - Posted: 11 Apr 2011, 2:22:51 UTC - in response to Message 47532.  

I've updated the new ATI application to 0.59. This fixes the 100% CPU problem, and adds a command line flag -r <some number> or --responsiveness-factor <some number>. This number just multiplies the estimate for how long things are expected to take to decide how to break things into smaller pieces to allow the screen a chance to redraw. Numbers greater than 1 should make things more responsive. 0 ignores any need for responsiveness for maximum speed.

1) What's the default?
2) And would it be something like: <cmdline>-r0</cmdline> in an app_info.xml for maximum speed?

And thanks!


ID: 47537 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Matt Arsenault
Volunteer moderator
Project developer
Project tester
Project scientist

Send message
Joined: 8 May 10
Posts: 576
Credit: 15,979,383
RAC: 0
Message 47538 - Posted: 11 Apr 2011, 2:23:02 UTC - in response to Message 47535.  

Will the command line options from 0.23 work with the new version?
This is a complete rewrite since we never had the source for that. I've never seen these options for the old one before, but I'll look into it. For right now there's only this quick little thing I just added.
ID: 47538 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Matt Arsenault
Volunteer moderator
Project developer
Project tester
Project scientist

Send message
Joined: 8 May 10
Posts: 576
Credit: 15,979,383
RAC: 0
Message 47539 - Posted: 11 Apr 2011, 2:33:23 UTC - in response to Message 47537.  

1) What's the default?
1

2) And would it be something like: -r0 in an app_info.xml for maximum speed?
Yes
ID: 47539 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Matt Arsenault
Volunteer moderator
Project developer
Project tester
Project scientist

Send message
Joined: 8 May 10
Posts: 576
Credit: 15,979,383
RAC: 0
Message 47540 - Posted: 11 Apr 2011, 2:35:17 UTC - in response to Message 47536.  

Overall speed compared to 0.23 seemed to be about 25 - 30% slower, I ran it at the same gpu/memory settings as I used for 0.23 so a comparison could be seen. It maybe the new app needs higher gpu/memory, dont know yet.
It's been about the same speed for me, but I just ran a few units myself in BOINC and noticed this slowdown on the same GPU.
ID: 47540 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Zydor
Avatar

Send message
Joined: 24 Feb 09
Posts: 620
Credit: 100,587,625
RAC: 0
Message 47541 - Posted: 11 Apr 2011, 2:38:38 UTC - in response to Message 47540.  
Last modified: 11 Apr 2011, 2:41:45 UTC

..... It's been about the same speed for me, but I just ran a few units myself in BOINC and noticed this slowdown on the same GPU.


Same 0.23? or roughly the speed I noticed?

I'm just trying to guage what the "new" baseline is to setup the 5970s for 0.57

EDIT: uppps - sorry typo, meant 0.59

Regards
Zy
ID: 47541 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile kashi

Send message
Joined: 30 Dec 07
Posts: 311
Credit: 149,490,184
RAC: 0
Message 47542 - Posted: 11 Apr 2011, 2:40:04 UTC

0.59 still using one full CPU core per task on my computer.

Xeon 3520, HD 5970, Win 7 64, Cat 11.4
ID: 47542 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
KWH*
Avatar

Send message
Joined: 24 Aug 10
Posts: 181
Credit: 83,101,286
RAC: 0
Message 47543 - Posted: 11 Apr 2011, 2:43:30 UTC

Well, now all my boxes are working and not returning errors. :)
Looks like CPU usage is still a little higher than the old but GPU temps are down.
Times are looking good as well.
Keep up the good work guys.
ID: 47543 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
HalfEmpty

Send message
Joined: 16 Nov 10
Posts: 17
Credit: 17,999,210
RAC: 0
Message 47544 - Posted: 11 Apr 2011, 2:49:33 UTC

The new app takes care of the CPU over utilization, thank you. It doesn't hammer the GPU as hard as 0.23 did resulting in a little longer run time, but a more responsive system and a cooler card. Fair trade off, I even think it's better for a non-dedicated cruncher.

I'm still having the problem with the app bullying and beating its way to the top of the queue mentioned in post 47501. I enabled <coproc_debug> and following is a snip of the output. Any ideas?

4/10/2011 9:55:03 PM | SETI@home | [coproc] Assigning ATI instance 0 to 17fe11ac.25226.11519.15.10.5_0
4/10/2011 9:55:03 PM | SETI@home | Restarting task 17fe11ac.25226.11519.15.10.5_0 using setiathome_enhanced version 610
4/10/2011 9:55:06 PM | Milkyway@home | Sending scheduler request: To fetch work.
4/10/2011 9:55:06 PM | Milkyway@home | Reporting 1 completed tasks, requesting new tasks for ATI GPU
4/10/2011 9:55:07 PM | Milkyway@home | Scheduler request completed: got 1 new tasks
4/10/2011 9:55:09 PM | Milkyway@home | [coproc] Assigning ATI pending instance 0 to de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:55:09 PM | SETI@home | [coproc] ATI device 0 already assigned: task 17fe11ac.25226.11519.15.10.5_0
4/10/2011 9:55:10 PM | Milkyway@home | [coproc] Assigning ATI instance 0 to de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:55:10 PM | Milkyway@home | [coproc] Assigning ATI instance 0 to de_separation_10_3s_fix10_1_500170_1302486546_1
...
  >> This line repeats several hundred times <<
...
4/10/2011 9:55:10 PM | Milkyway@home | [coproc] Assigning ATI instance 0 to de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:55:10 PM | Milkyway@home | [coproc] Assigning ATI instance 0 to de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:55:10 PM | Milkyway@home | Starting de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:55:10 PM | Milkyway@home | Starting task de_separation_10_3s_fix10_1_500170_1302486546_1 using milkyway version 59
4/10/2011 9:55:38 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:56:09 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:56:25 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:57:25 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:57:38 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:58:09 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:58:25 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:58:32 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:58:55 PM | Milkyway@home | [coproc] ATI instance 0: confirming for de_separation_10_3s_fix10_1_500170_1302486546_1
4/10/2011 9:59:01 PM | Milkyway@home | Computation for task de_separation_10_3s_fix10_1_500170_1302486546_1 finished
4/10/2011 9:59:01 PM | SETI@home | [coproc] Assigning ATI instance 0 to 17fe11ac.25226.11519.15.10.5_0
4/10/2011 9:59:01 PM | SETI@home | Restarting task 17fe11ac.25226.11519.15.10.5_0 using setiathome_enhanced version 610
4/10/2011 9:59:03 PM | Milkyway@home | Sending scheduler request: To fetch work.
4/10/2011 9:59:03 PM | Milkyway@home | Reporting 1 completed tasks, requesting new tasks for ATI GPU
4/10/2011 9:59:04 PM | Milkyway@home | Scheduler request completed: got 1 new tasks
4/10/2011 9:59:07 PM | Milkyway@home | [coproc] Assigning ATI pending instance 0 to de_separation_13_3s_free_1_500595_1302486884_1
4/10/2011 9:59:07 PM | SETI@home | [coproc] ATI device 0 already assigned: task 17fe11ac.25226.11519.15.10.5_0
4/10/2011 9:59:07 PM | Milkyway@home | [coproc] Assigning ATI instance 0 to de_separation_13_3s_free_1_500595_1302486884_1
4/10/2011 9:59:07 PM | Milkyway@home | [coproc] Assigning ATI instance 0 to de_separation_13_3s_free_1_500595_1302486884_1
...
  >> And the process repeats for every WU <<

ID: 47544 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Zydor
Avatar

Send message
Joined: 24 Feb 09
Posts: 620
Credit: 100,587,625
RAC: 0
Message 47545 - Posted: 11 Apr 2011, 2:59:57 UTC

Ran a second batch 825/500 as opposed to 790/500. Speeded up as you would expect, still around 25% slower. CPU hit is still high, the PRPNet app took a 30% + hit. Screen started to shows initial signs of lag and faint juddering.

Overall it looks stable, run real real cool, CPU is still too high, and its about 25% slower than 0.23 even running at 825/500. Feels stable enough to start working on parts of it one by one to improve performance now its stable, still long way to go to get back to old performance levels.

Regards
Zy
ID: 47545 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Ed.T

Send message
Joined: 1 Feb 11
Posts: 17
Credit: 16,245,184
RAC: 0
Message 47551 - Posted: 11 Apr 2011, 4:24:11 UTC - in response to Message 47543.  

This is embarassing. I think the project would be better off if they just restored last week's version and our administrators backed off, took a few deep breaths and did some real software engineering and project planning.



ID: 47551 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Sunny129
Avatar

Send message
Joined: 25 Jan 11
Posts: 271
Credit: 346,072,284
RAC: 0
Message 47553 - Posted: 11 Apr 2011, 4:42:56 UTC
Last modified: 11 Apr 2011, 4:51:32 UTC

everything is going fine for me w/ MW@H v0.59. of course i wasn't having any problems with v0.57 either. task run times are no slower with v0.59 than they were with v0.57. likewise, task run times were no slower with MW@H v0.57 than they were with v0.23, for me anyways. in fact, the only difference i noticed is that with MW@H versions 0.23 and 0.57, my GPU load would randomly walk between 85% and 100% during any given task, with an average of about 95% load. with the new MW@H v0.59, my GPU usage is down lightly, now randomly walking between 70% and 100%, with an average of about 80% load. but this is with just some tasks...other tasks casue the GPU load to walk like it did with MW@H v0.23 and v0.57. overall, i take this as a good thing considering my run times are no slower than before, and theoretically a lower GPU load on some tasks should mean generally lower GPU temps (though i haven't confirmed it with MSI Afterburner yet). i got one invalid result, but this was due to the task erroring out too many times on other clients.

- i'm crunching on an HD 5870 2GB GPU
- i'm using the Catalyst 11.3 drivers
- i'm running BOINC v6.12.18
- my OS is WinXP Pro SP3 32-bit

*EDIT* - i should clarify what i mean when i say that my run times haven't increased. what i mean to say is that my maximum run times for the various types of tasks (those that earn ~160, ~213, and ~267 points each) have not increased. however, i will say that, for the range it crunch times i would observe for each type of task in the past, it does seem like the run times of an increasing number of tasks running on the new MW@H v0.59 are near the higher end of the crunch-time ranges i've historically observed...so perhaps in the big picture, tasks are running slightly slower for me with MW@H v0.59. but it doesn't seem too significant.
ID: 47553 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
HalfEmpty

Send message
Joined: 16 Nov 10
Posts: 17
Credit: 17,999,210
RAC: 0
Message 47554 - Posted: 11 Apr 2011, 4:47:06 UTC - in response to Message 47551.  

This is embarassing. I think the project would be better off if they just restored last week's version and our administrators backed off, took a few deep breaths and did some real software engineering and project planning.

If this were a release from a for-profit software company I would agree with you. Considering this is a volunteer science project run on a very tight budget I have to disagree.

There was an initial unplanned oops which took the project down and was going to necessitate a few days to recover. What better time to throw in the upcoming changes that might mess things up and take the project down for a few days? It also saves a lot of time and money avoiding all that pesky beta testing, scheduled roll outs and downtime. People tend to be more productive when they're under the gun. ;-)
ID: 47554 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Sunny129
Avatar

Send message
Joined: 25 Jan 11
Posts: 271
Credit: 346,072,284
RAC: 0
Message 47557 - Posted: 11 Apr 2011, 6:22:00 UTC - in response to Message 47553.  

everything is going fine for me w/ MW@H v0.59. of course i wasn't having any problems with v0.57 either. task run times are no slower with v0.59 than they were with v0.57. likewise, task run times were no slower with MW@H v0.57 than they were with v0.23, for me anyways. in fact, the only difference i noticed is that with MW@H versions 0.23 and 0.57, my GPU load would randomly walk between 85% and 100% during any given task, with an average of about 95% load. with the new MW@H v0.59, my GPU usage is down lightly, now randomly walking between 70% and 100%, with an average of about 80% load. but this is with just some tasks...other tasks casue the GPU load to walk like it did with MW@H v0.23 and v0.57. overall, i take this as a good thing considering my run times are no slower than before, and theoretically a lower GPU load on some tasks should mean generally lower GPU temps (though i haven't confirmed it with MSI Afterburner yet). i got one invalid result, but this was due to the task erroring out too many times on other clients.

- i'm crunching on an HD 5870 2GB GPU
- i'm using the Catalyst 11.3 drivers
- i'm running BOINC v6.12.18
- my OS is WinXP Pro SP3 32-bit

*EDIT* - i should clarify what i mean when i say that my run times haven't increased. what i mean to say is that my maximum run times for the various types of tasks (those that earn ~160, ~213, and ~267 points each) have not increased. however, i will say that, for the range it crunch times i would observe for each type of task in the past, it does seem like the run times of an increasing number of tasks running on the new MW@H v0.59 are near the higher end of the crunch-time ranges i've historically observed...so perhaps in the big picture, tasks are running slightly slower for me with MW@H v0.59. but it doesn't seem too significant.

i now believe this somehow has to do with my running E@H on all 6 CPU cores. when i set BOINC to use only 85% of the processors (forcing BOINC to use only 5 of my 6 CPU cores), or when i suspend E@H altogether (freeing up my CPU entirely), GPU load jumps back up to a constant and steady 98-99%. accordingly, run times approach the lower end of the crunch-time ranges i've historically observed. so my revised "big picture" is this: when i run E@H on all 6 CPU cores, MW@H GPU crunching efficiency suffers a bit in the form of slightly increased run times. when i leave a single CPU core free (or if i suspend E@H altogether and leave all 6 cores free), MW@H GPU crunching efficiency doesn't suffer at all, and run times are slightly decreased.
ID: 47557 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Matt Arsenault
Volunteer moderator
Project developer
Project tester
Project scientist

Send message
Joined: 8 May 10
Posts: 576
Credit: 15,979,383
RAC: 0
Message 47558 - Posted: 11 Apr 2011, 6:33:16 UTC - in response to Message 47557.  

i now believe this somehow has to do with my running E@H on all 6 CPU cores. when i set BOINC to use only 85% of the processors (forcing BOINC to use only 5 of my 6 CPU cores), or when i suspend E@H altogether (freeing up my CPU entirely), GPU load jumps back up to a constant and steady 98-99%. accordingly, run times approach the lower end of the crunch-time ranges i've historically observed. so my revised "big picture" is this: when i run E@H on all 6 CPU cores, MW@H GPU crunching efficiency suffers a bit in the form of slightly increased run times. when i leave a single CPU core free (or if i suspend E@H altogether and leave all 6 cores free), MW@H GPU crunching efficiency doesn't suffer at all, and run times are slightly decreased.
This seems to confirm what I've been finding. It seems that the slow happens with a high CPU load (at least in Windows. It seems to not happen for me in Linux). The time is about what it should be with a low CPU load, and quite a bit higher if the CPU load is light. This is still a problem to fix though.
ID: 47558 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TJ

Send message
Joined: 12 Aug 09
Posts: 262
Credit: 92,631,041
RAC: 0
Message 47559 - Posted: 11 Apr 2011, 6:55:16 UTC - in response to Message 47532.  

I've updated the new ATI application to 0.59. This fixes the 100% CPU problem, and adds a command line flag -r <some number> or --responsiveness-factor <some number>. This number just multiplies the estimate for how long things are expected to take to decide how to break things into smaller pieces to allow the screen a chance to redraw. Numbers greater than 1 should make things more responsive. 0 ignores any need for responsiveness for maximum speed.

I'm not exactly happy with how it decides to break the problem down to keep the screen responsive; I think how it's done now can still cause slowdowns on some GPUs (and is kind of fragile), and you shouldn't have to do anything to keep things responsive. I'm not really sure the best way to do it. If you feel you actually need to use this, complaints here with what GPU you have might be useful.


Hi Matt,

Well with this update (0.59) my pc (i7, Win7-64, 12GbRAM, 2HD5870 driver 11.3) is still using a full cpu load. I let run out all the cpu tasks (einstein@home) and then only MW. 8 Core 100% divided by 8 is 12.5 (12-13% use of one core) and that is what is happening. I see it also in teh results page, Runtime 130.09/CPUtime 128.22

I read other posts in this thread that more have this experience. So you have some more thinking.
Good luck.

Greetings from,
TJ
ID: 47559 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · 3 · 4 . . . 5 · Next

Message boards : News : ATI application updated again

©2024 Astroinformatics Group