| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Suggestions / Feedback
Thread: Problems with the Due Time |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 49
|
|
| Author |
|
|
mirabilos
Cruncher Joined: Mar 25, 2008 Post Count: 37 Status: Offline Project Badges: 
|
Hi,
----------------------------------------I am running the BOINC/Windows client on my 2nd computer (I only have laptops), which is a Pentium 233MMX machine, because you do not allow to run the OpenBSD client or the anonymous client on my main box running MirBSD (which I am the head developer of), and because the Linux client does not work on the Linux emulation of MirBSD (get random Signal 12). Now this box is rather slow, even if otherwise idle and running the hcc process at above-normal priority I guess it won’t make the result within the Due Time. Can I request a larger due time and/or smaller work units for a specific device, and if so, where and how? Thanks! //mirabilos  ☮  ☯  |
|||
|
||||
- 
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
ATM that's not possible, but we have a selection . e.g. the DDD-T jobs are not much longer, but have a 14 day deadline.
----------------------------------------As for OpenBSD, visit this previous thread: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=16397#143221 ttyl
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It's just not practical to run very old computers on a grid. Thinking in terms of cost, you are spending a week's worth of electricity for the equivalent of half an hour crunching on a recent computer. Add in the fact that modern processors are more energy efficient, and the problem is even more pronounced.
World Community Grid recommend 500 MHz as a rough minimum (although I expect this will be raised in a year or so). The sad fact is that processor specs are always increasing, and due to the kind of computation the scientists need, they always want faster processors and more memory. Faster computers open up avenues of research that were previously thought impossible. It's an exciting time, but we have to accept that old computers will have to be retired eventually. On BSD: We have had good results on FreeBSD using Linux emulation. I'm less sure about OpenBSD. Note that you need to list Linux compatibility using the new alternate platform mechanism. Remember, you can use the native client, but it has to be able to run the Linux ELF binaries that are provided precompiled to run the actual science. Please ask if you want to discuss this more deeply. |
||
|
|
mirabilos
Cruncher Joined: Mar 25, 2008 Post Count: 37 Status: Offline Project Badges:
|
ATM that's not possible, but we have a selection . e.g. the DDD-T jobs are not much longer, but have a 14 day deadline. Okay. The bad thing is that that box has barely enough RAM for HCC. As for OpenBSD, visit this previous thread: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=16397#143221 I actually used the precompiled binary and had patched the machine ID string to i686-pc-linux-gnu ;) The issue was that the work unit process terminated with signal 12 (SIGSYS, bad system call) – they seem to use a syscall which is not implemented in the linux emulation layer. (So WCG wouldn’t work well on Linux 2.4 or 2.2 either.) It's just not practical to run very old computers on a grid. Thinking in terms of cost, Yeah, but I only have old boxen. But I can get a large number of them ;) And the cost is my problem, isn’t it? On BSD: We have had good results on FreeBSD using Linux emulation. I'm less sure about OpenBSD. Note that you need to list Linux compatibility using the new alternate platform mechanism. Remember, you can use the native client, but it has to be able to run the Linux ELF binaries that are provided precompiled to run the actual science. Please ask if you want to discuss this more deeply. See above. But I’ll retry, maybe with a different project. I think I had faah. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
The present ram/vm size needs can be found in the Matrix FAQ of the Start Here forum. From small to large memory needs with due times:
----------------------------------------1. HCC (10 days) 2. HPF2 (20 days) 3. FA@H (10 days) 4. DDDT (14 days) 5. AC@H (10days)
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
mirabilos
Cruncher Joined: Mar 25, 2008 Post Count: 37 Status: Offline Project Badges:
|
Yes, I saw them, thanks.
----------------------------------------Okay, I now built a MirBSD native BOINC client, as per the suggestions in the other thread for OpenBSD, with the configure options (among others): --disable-server --disable-dependency-tracking --enable-fast-install --disable-libtool-lock --with-boinc-alt-platform=i686-pc-linux-gnu --with-ssl --with-pic --without-x --with-wxdir=junk I got faah running without signal 12 now, but I have some weird output in “./boinc_cmd --get_state” – I think it should be done with its first job now, but it isn’t: name: faah3395_ZINC00608170_xMut_md01930_00_1 WU name: faah3395_ZINC00608170_xMut_md01930_00 project URL: http://www.worldcommunitygrid.org/ report deadline: Sat Apr 12 16:27:05 2008 ready to report: no got server ack: no final CPU time: 0.000000 state: 2 scheduler state: 2 exit_status: 0 signal: 0 suspended via GUI: no active_task_state: 1 stderr_out: app version num: 542 checkpoint CPU time: 49454.234374 current CPU time: 49703.874999 fraction done: 1.000000 swap size: 0.000000 working set size: 0.000000 estimated CPU time remaining: 0.000000 supports graphics: no I had restarted it once, and the current CPU time went back to 49454.whatever then, but the fraction done is staying at 100% no matter what, and it doesn’t seem to complete. Also, it doesn’t use up any CPU any more. Is this normal? I have an Athlon XP CPU in my primary laptop where I am running it now, and that one has K7 PowerNow, which I use to dynamically regulate the CPU speed (i.e. heat and fan usage), could that be an issue? |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Possibly, but it is more likely that you need to patch BOINC to deal with some differences in how MirBSD measures CPU time. That's my guess, anyway.
I would be able to give a better guess based on the BOINC message log and the work unit error log. It may help to contact Pav, the FreeBSD port maintainer. At the very least, take a look at the patches the Open and FreeBSD ports apply. |
||
|
|
mirabilos
Cruncher Joined: Mar 25, 2008 Post Count: 37 Status: Offline Project Badges:
|
I would be able to give a better guess based on the BOINC message log and the work unit error log. How do I get these to you (and where do I get them from)? It may help to contact Pav, the FreeBSD port maintainer. At the very least, take a look at the patches the Open and FreeBSD ports apply. Hmm. As I used the 6.1 code from trunk, the patches from FreeBSD don’t seem necessary: my CPU and memory size are correctly calculated. OpenBSD has no boinc port. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Don't build trunk! It's not stable. Get the 5.10 branch.
Message log should be in stdoutdae.txt and the error log is in the relevant slot directory, named stderr.txt |
||
|
|
mirabilos
Cruncher Joined: Mar 25, 2008 Post Count: 37 Status: Offline Project Badges:
|
Don't build trunk! It's not stable. Get the 5.10 branch. ;) 6.1.12 (just looked) is somehow even recommended, I saw that somewhere on this site that they’re going to switch soon. Message log should be in stdoutdae.txt and the error log is in the relevant slot directory, named stderr.txt No, message log isn’t there. This is what I can copy/paste from the GNU screen session where I ran boinc from:
This is stderr.txt:
I do have a wcg_ag.log in the slot directory. Maybe this helps too:
|
||
|
|
|