| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Help Conquer Cancer
Thread: Updated Science Application for Help Conquer Cancer v7.05 |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 53
|
|
| Author |
|
|
branjo
Master Cruncher Slovakia Joined: Jun 29, 2012 Post Count: 1892 Status: Offline Project Badges: 
|
Why do I have to change things called "App Files". When stuff had be updated before I have never had to change anything........ Confused of Winsford!  Scribe, the app_info file that people are changing is for GPU. They are telling BOINC that they can process more than one GPU unit at a time on a graphics card. Since the GPU app takes a few minutes of CPU time before and after, they are keeping thier GPU processor busy instead of idle during the CPU portion of the GPU application. If you did not modify your app_info file, then you have nothing to worry about. Normal operations should occur. I say should because if it's not, we would like to know about it. Thanks, -Uplinger It is not only about keeping our GPU's busy during CPU part of GPU application. It is more about maximizing GPU during GPU computation. If I run only 1 GPU task, my GPU is utilized to, let say, 70%. If I run 8 of them via app_info.xml, it it utilized to 99%. Also in this case I am shrubbing 40 HCC1 tasks+ per hour (ca. 12 mins per task GPU X 8 CPU threads) . In case of running it w/o app_info file, I am crunching 26 HCC1 tasks per hour (0.9 CPU task per thread x 7 threads + ca. 3 mins per task GPU + 1 CPU). And my CPU is not that loaded (and hot) as in case of 7x CPU + 1 GPU shrubbing  Cheers and NI!  Crunching@Home since January 13 2000. Shrubbing@Home since January 5 2006  [Edit 4 times, last edit by branjo at Nov 10, 2012 5:07:46 PM] |
||
|
|
David Autumns
Ace Cruncher UK Joined: Nov 16, 2004 Post Count: 11062 Status: Offline Project Badges:
|
Still stuttering at my end but whatever you have incorporated into 7.05 it's a step in the right direction
---------------------------------------- Web Browsing is almost acceptable but anything involving Video is still a no go  What's this trick with avoiding the GPU pause at the end of each WU ? at least that way I can get more WU's completed when I am not watching a Movie. (I have to set "Use GPU Never"  )Dave  |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
What's this trick with avoiding the GPU pause at the end of each WU ? That's the way it's suppose to work. The "pause" is when the GPU passes the last portion of the task back to the CPU to finish it.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
  |
||
|
|
dskagcommunity
Senior Cruncher Austria Joined: May 10, 2011 Post Count: 219 Status: Offline Project Badges: 
|
you can only "avoid" it with multiple running WUs. when one tasks runs on the cpupart the other task(s) run on GPU for example and vice versa.
---------------------------------------- |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
What's this trick with avoiding the GPU pause at the end of each WU ? That's the way it's suppose to work. The "pause" is when the GPU passes the last portion of the task back to the CPU to finish it. David, the trick is messing with the app_info file. You'll need to look into what others are doing in the GPU forum about that. WCG has no official support for it. If you leave your machine with just the standard BOINC installed, then the GPU "pause" is designed that way. It's how the application works. Thanks, -Uplinger |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If you are already unhappy with stutter, I would not suggest you run multiple concurrent WUs as that will exacerbate the situation.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
[OT]My next comp actually will have either a 7770 or 7950 included in the off the shelf kit [Just waiting on the wave of W8 hardware to deflate prices for the W7 stuff]. Looking forward to the joy of running 12 quasi parallel, self staggering, while letting the 3770 "on CPU" GPU do the video stuff. W7/W8 and CCC have excellent tools to automate which resource a program/stream uses. This will be the point of retiring the Q6600 box. Results per watt will not be a contest, at all.
P.S. The Q6600 I got down from running 170 Watt/Hour headless under W7 to 162 Watt/Hour under W8. That part of the promises is true. The credit per hour went up, in the margin, but never the less, a slightly better throughput.[OT] |
||
|
|
depriens
Senior Cruncher The Netherlands Joined: Jul 29, 2005 Post Count: 350 Status: Offline Project Badges:
|
@Sek, the 7770 I have is doing a great job and it's quite economical. My i7-920 on 8 threads went from about 150W/hr with an (unused) HD4550 to 190W/hr with the HD7770.
----------------------------------------I haven't been able to get the app_info to work yet, so it's only doing 1 unit at a time. Will give it another try when the double-image units are there.   |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
What is you OS and a bit more about your system specs :) so that more specific help can be given.
|
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
@Sek, the 7770 I have is doing a great job and it's quite economical. My i7-920 on 8 threads went from about 150W/hr with an (unused) HD4550 to 190W/hr with the HD7770. I like the HD 7770 just fine also, but have not had good luck with the app_info. It works OK to get two work units running at once, but that is where the good news ends. On one machine, with an E8400 dual-core at 3.0 GHz (and a PCIe 2.0 bus), they all result in "errors" in the WCG Results Status, even though they do not show errors on my PC and run for the full time, both on the CPU and the GPU. On another PC with a Core i7-3770 at 3.7 GHz (and a PCIe 3.0 bus), it is even worse, since it crashes the whole PC (with a BSOD no less) after a few hours of work. With only single work units, both PCs run fine and have no errors.I haven't been able to get the app_info to work yet, so it's only doing 1 unit at a time. Will give it another try when the double-image units are there. Nothing is overclocked on either PC, and the temps are quite low. It could probably be straightened out with the right drivers and eliminating other distributed computing projects (I run both WCG CPU projects and Folding@home on the second PC), but I don't think it is worth the trouble. And I have Nvidia cards also on each PC, which might contribute to the problems, but I don't want to get rid of them either. However, the good news is that the HD 7770 runs quite well with only a single CPU core on each machine, and in my experience you do not get double the points just because you double the work units (since they overlap with CPU usage), and so you don't lose that much just by running one at a time anyway. If you are willing to devote a whole PC just to feeding the card, it could work, but I have other uses for my PCs as noted above. |
||
|
|
|