| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: GPU Support Forum
Thread: Project Selection Workaround |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 18
|
|
| Author |
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
Due to some very clever thinking by the volunteers, I think that most people can achieve their desired settings with the following workarounds. Please note that these are workarounds. We are starting an effort to redesign the project selection page and device profile pages to make them easier to use and provide more flexibility.
----------------------------------------Option 1: Run ONLY GPU for World Community Grid: * For the device profile option "Allow research to run on my CPU? " select "no" * For the option "If my computer can process work on my graphics card, then please send me work to run on my graphics card for the projects that I have selected above." - Check the box. * Make sure you have opted to run work for the Help Conquer Cancer project Option 2: Run ONLY GPU for Help Conquer Cancer, Run CPU for other projects. Do not run CPU for Help Conquer Cancer * For the device profile option "Allow research to run on my CPU? " select "yes" * For the option "If my computer can process work on my graphics card, then please send me work to run on my graphics card for the projects that I have selected above." - Check the box. * Make sure that you are NOT signed up for the Help Conquer Cancer project * Make sure that you HAVE checked the box for "If there is no work available for my computer for the projects I have selected above, please send me work from another project." If someone could verify success with these options, that would be appreciated. Also, please note other desired scenarios below and I will update this post with workarounds for that as we find them. [Edit 1 times, last edit by knreed at Oct 11, 2012 2:54:24 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
That works, but will of course send a random set, not a specific, and could include HCC, similar to the conundrum of CEP2 being occasionally send when though not wanting them.
----------------------------------------Was thinking before you somehow managed to do the trick of e.g. FAAH+SN2S+GPUHCC and feeder priority [low] to coax the system in only letting HCC be send to the GPU and in this sample FAAH+SN2S to the CPU. Thanks for drafting the *workaround* FAQ :D [Edit 1 times, last edit by Former Member at Oct 11, 2012 3:04:58 PM] |
||
|
|
Bearcat
Master Cruncher USA Joined: Jan 6, 2007 Post Count: 2803 Status: Offline Project Badges: 
|
Will give it a shot to see what happens. Thanks.
----------------------------------------
Crunching for humanity since 2007!
 |
||
|
|
wplachy
Senior Cruncher Joined: Sep 4, 2007 Post Count: 423 Status: Offline |
OS: Win 7 64
----------------------------------------Client: 6.10.58 I set up the profile to: ..select GFAM and DDDT2, ..Allow research to run on my CPU to yes, .."If my computer can process work on my graphics card, then please send me work to run on my graphics card for the projects that I have selected above." is Checked .."If there is no work available for my computer for the projects I have selected above, please send me work from another project." is Checked ..Additional work buffer to 0.0 days This works fine for me as long as I have my cache set to 0.0 days. The device I tried it on is fetching HCC GPU WU and honoring the cache setting for the selected subprojects. However, if I increase the cache to anything greater than 0.0 days the client continues to ask for and receive the selected CPU tasks. After letting it run for about 5 minutes with a 0.2 cache it has pulled down over 3 days of WU. My understanding is that this is a "bug" in the 6.10.58 client. So from my experience be careful about your cache setting when using this method.
Bill P
 |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
We have temporarily limited the max number of workunits in progress for cpu to 10 due to this issue. We are examining how to address now.
|
||
|
|
hnapel
Advanced Cruncher Netherlands Joined: Nov 17, 2004 Post Count: 82 Status: Offline Project Badges:
|
hi,
----------------------------------------This is not a a success entirely, what I want is devote 75% of CPU to CEP2 (this can be done if I allow only 6 wu's since I have 8 cpu threads) and HCC may use the GPU (If I play a game I stop the entire boinc manager, including the wu's). Incidently the reason I only run CEP2 is this ability to limit the number of WU's downloaded at any time, this should be a configuration for all projects, since I simply want to keep 25% of CPU power for me myself and I :) With this device profile I get a number of other WU's for Go fight against malaria, fightaids@home, say no to schistosoma, human proteome folding, drug search for leishmaniasis AND (indeed) Help Conquer cancer 6.56 (ati_hcc1) . But the ATI project is not even running currently since all the slots are taken, so it's a big gamble and my CPU is utilized 100% which I don't like. So I cannot use this option, the only choice would be to abandon CEP2 and stop using the CPU and only opt-in for the GPU project. There's a possibility the following may work: change the device profile each time, then manually update the WCG project, I assume the active wu's will just keep running. - switch to : 'no cpu', hcc project exclusively , sync project check message log and wait until gpu project runs - switch profile back to: 'may use cpu', cep2 exclusively (of course other volunteers can choose the projects they support) I will test this and update once current (unwanted) WU's are gone from my system, greetings, Harm [Edit 3 times, last edit by hnapel at Oct 11, 2012 6:13:26 PM] |
||
|
|
CandymanWCG
Senior Cruncher Romania Joined: Dec 20, 2010 Post Count: 421 Status: Offline Project Badges: 
|
I really hope someone "up there" hears you, hnapel! I am in about the same situation, just using 3 of my 4 cores (still 75% though). I guess even though many have waited for the GPU HCC, few have anticipated all the ruckus this will cause, but if knreed or one of the techs that have a say in it are reading this, I really do hope they take heave and implement not just the option to run each project only on GPU or CPU, but also the number of concurrent tasks that can be run by project...
----------------------------------------Awww...wouldn't that be a dream come true?  Edit2: finally got the "awww" right...I think... Knowledge is limited. Imagination encircles the world! - Albert Einstein   [Edit 2 times, last edit by CandymanWCG at Oct 11, 2012 6:43:35 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks! It works
 ...but now I have to find rope to patch my Beta-CPU net  |
||
|
|
kateiacy
Veteran Cruncher USA Joined: Jan 23, 2010 Post Count: 1027 Status: Offline Project Badges:
|
Option 2: Run ONLY GPU for Help Conquer Cancer, Run CPU for other projects. Do not run CPU for Help Conquer Cancer * For the device profile option "Allow research to run on my CPU? " select "yes" * For the option "If my computer can process work on my graphics card, then please send me work to run on my graphics card for the projects that I have selected above." - Check the box. * Make sure that you are NOT signed up for the Help Conquer Cancer project * Make sure that you HAVE checked the box for "If there is no work available for my computer for the projects I have selected above, please send me work from another project." If someone could verify success with these options, that would be appreciated. This is working almost perfectly for me. My only Windows machine has received lots of HCC GPU WUs and no HCC CPU WUs. It has received CPU WUs from the sciences that I have specifically selected. However, it also received one CPU WU from a science that is not checked. (Maybe it happened to request tasks at a brief moment when no GFAM or SN2S were available.)  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
knreed, thanks for this workaround. I set buffer to 0 (just in case) and looks like it's working well. I get HCC units for gpu and HPF2 for cpu.
|
||
|
|
|