Posts by Khangollo
Posts by Khangollo |
|
| log in |
|
21)
Message boards :
Number crunching :
Panic Mode On (78) Server Problems?
(Message 1302717)
Posted 197 days ago by  Khangollo
Timeouts are back plus now I can't download tons of lost tasks I still have; when scheduler is finally successful, all I'm getting is limit reached message. Looks like limits are ridiculous like they were before (50tasks/CPU) which means even my slowest hosts are going to make scheduler requests *endlessly*, never able to fill a 5 day cache, only compounding the scheduler overload problem... |
|
22)
Message boards :
Number crunching :
Panic Mode On (78) Server Problems?
(Message 1301662)
Posted 200 days ago by Khangollo
Besides barely functional scheduler, uploads are now slowly starting to stick, too. Yellow alert mode on. |
|
23)
Message boards :
Number crunching :
Development BOINC 7.0.62
(Message 1301270)
Posted 201 days ago by Khangollo
There's no 7.0.38 in official svn (boinc.berkeley.edu/svn/tags/). Has the repository changed? Edit: Sorry, nevermind. I just noticed it moved to git. |
|
24)
Message boards :
Number crunching :
Panic Mode On (77) Server Problems?
(Message 1301151)
Posted 201 days ago by Khangollo
After several timeouts, found that I had downloaded 16 wu's. However, on looking at My computers page, found that I had 119 wu's, but 103 marked "abandoned"? You're lucky. My computer got entire 5-day cache of 692 tasks (including some of new APs) abandoned today for absolutely no reason. The worst of it is, none of the workunits were deleted locally, so if I hadn't aborted them, I'd be doing 5 days of pointless work. |
|
25)
Message boards :
Number crunching :
Panic Mode On (77) Server Problems?
(Message 1301054)
Posted 201 days ago by Khangollo
One of my hosts is getting nothing but scheduler timeouts and http errors. Uploads work fine. The only difference between this and other host is it's on a different network using different ISP. Scheduler requests DO get through to server, but no response is received. This host has now probably collected more "lost tasks" than it could ever process in time. Edit: All my tasks suddenly got marked as "Abandoned" on server (all of them!!) but I didn't touch the boinc in any way and it still has its cache and crunches! http://setiathome.berkeley.edu/results.php?hostid=5323998 Brilliant. |
|
26)
Message boards :
Number crunching :
New GPUs, both not in use now
(Message 1296361)
Posted 216 days ago by Khangollo
I can confirm that total memory detection is broken on Linux, too. 10/18/2012 12:03:59 AM Starting BOINC client version 7.0.36 for x86_64-pc-linux-gnu ... 10/18/2012 12:03:59 AM NVIDIA GPU 0: GeForce GTX 570 (driver version unknown, CUDA version 4.20, compute capability 2.0, 134214911MB, 1212MB available, 1498 GFLOPS peak) 10/18/2012 12:03:59 AM OpenCL: NVIDIA GPU 0: GeForce GTX 570 (driver version 295.71, device version OpenCL 1.1 CUDA, 1279MB, 1212MB available) It works for OpenCL, but not CUDA (and still no driver version detection; will this be fixed?). |
|
27)
Message boards :
Number crunching :
Xeon Phi (aka Knights Corner, MIC)
(Message 1252052)
Posted 329 days ago by Khangollo
If you are an ordinary user and am using the console only (no graphics or multi-user/multi-tasking environment) and input the command sleep 30, the PC goes to sleep for that 30 seconds. Actually, "sleep" doesn't make PC go to sleep. It is simply a process that executes for the given time (while not cunsuming any CPU cycles), then exits. It is commonly used in scripts (without &) to add artificial delays. But replace the command with sleep 30 & instead, making it a background process. What happens then? It simply launches process in the background (that does absolutely nothing) where it exists for the given number of seconds then exits. If you use it in a script like this (with &) it of course doesn't add delay to a script, because the script will continue executing while sleep is executing in the background, detached from the parent process. (Oh, and non-multitasking OS wouldn't support any kind of launching into background anyway, except specially programmed processes - like drivers/memory resident programs - like in DOS). What if the environment is more advanced/sophisticated? How would this command appear as a running process? It's a process like any other program, with its own PID (process ID) and name "sleep N" (N = seconds). Finally, if you are root or superuser on a system, what may happen if you try running this command either by means of the console or in a graphical multi-user/multi-tasking environment? Makes no difference. "sleep" process will simply exist under root credentials. |
|
28)
Message boards :
Number crunching :
Is SLI faster
(Message 1250718)
Posted 332 days ago by Khangollo
No, for crunching/CUDA, each card is used separately. You must leave SLI disabled - it is only for 3D. Crunching doesn't work in SLI. With 1 GB per card you can do 2-3 workuints at a time on each. |
|
29)
Message boards :
Number crunching :
Question about feeding the GPU
(Message 1250463)
Posted 333 days ago by Khangollo
Should be quite a bit more with 2600K and 580. Just for comparison, my i7 920 (oc @ 3.2GHz, HT off), GTX 570 (stock), Linux, was doing between 35-37K RAC (not right now, since I'm doing E@H). |
|
30)
Message boards :
Number crunching :
A question about RAC
(Message 1249408)
Posted 335 days ago by Khangollo
Now it takes over 2 months for RAC to stabilize on S@H for a new host. Possibly more. Other projects not that much, but definitely at least one month. It's not just how it gets calculated over past days but how your completed workunit queue gets validated (completed by your wingmen). Your rac will be more-or-less stable when your "Pending" list won't grow any more and this takes a lot at S@H with 10+10 cache ******* out there and tons of unchecked rogue hosts creating invalid results, causing workunits to linger for months in you "pending" queue. |
|
31)
Message boards :
Number crunching :
Anonumous host throwing only errors, 3223 right now
(Message 1247245)
Posted 339 days ago by Khangollo
When will this slop stop? From what I was checking a bit through my ever growing pending list, in a lot of cases it is the faulty windows nvidia driver ("sleep bug"). Unless admins update the scheduler to not issue cuda tasks to hosts with certain driver versions - like they did at einstein@home - this is probably not going to stop on its own until everyone updates to 301.x... which may take months or even never. |
|
32)
Message boards :
Number crunching :
Invalid App_Info file problems
(Message 1245591)
Posted 342 days ago by Khangollo
I use xml marker to edit all my xml files instead of notepad or notpad++ I have recomended it before it is free to use and it does a good job also a good editor is xml notepad. xml marker will parse the file and find errors and will indent the lines and other usefull stuff. One should be cautious using some XML-aware editors when editing boinc's files and not let them try to reformat the code in any way. I think notepad++'s default settings are okay; it just colors the code. Improperly named ".xml" files used in boinc are not really XML for boinc (has a super simple parser) so an editor might suggest a change that is allowed in XML but boinc will not understand it. |
|
33)
Message boards :
Number crunching :
Anonumous host throwing only errors, 3223 right now
(Message 1245500)
Posted 342 days ago by Khangollo
...That's a nasty one. I just noticed that my pending list keeps growing and I found a few pages of inconclusives, all with this host as wingman. Why is it fetching workunits for cuda/cuda23 when there's no GPU shown on its host details page? (stderr says GTX 280 and GTX 295) |
|
34)
Message boards :
Number crunching :
High Priority Tasks?
(Message 1245390)
Posted 343 days ago by Khangollo
I must say I love it how they silently enabled this option server side while recommended BOINC is still 7.0.25, which has a KNOWN, massive bug with this same option. Bravo! Of course this is beneficial to S@H itself... everyone is doing seti now, whether they like it or not. LOL |
|
35)
Message boards :
Number crunching :
Sure Wish I Could Fine More Like WU 997538979
(Message 1244968)
Posted 344 days ago by Khangollo
That's alright. The other day I got 0.97 cr for 10+ hour astropulse. Apparently wingman ran it for 5 minutes (resent lost tasks bug)! It's all balanced. |
|
36)
Message boards :
Number crunching :
Windows 8 Release Preview now available
(Message 1241725)
Posted 351 days ago by Khangollo
They should name it Millennium Edition 3.0. More appropriate. |
|
37)
Message boards :
Number crunching :
Strange result......................
(Message 1240684)
Posted 353 days ago by Khangollo
Looks like it tried to run on GPU then failed and was re-scheduled to CPU? Maybe with one of those external rescheduler programs. Edit: yes, the 5 minute difference between issued and reported suggests it is indeed from this new, stupid server bug. :@ |
|
38)
Message boards :
Number crunching :
transient HTTP error
(Message 1240614)
Posted 353 days ago by Khangollo
I'm not sure this will help, but you might try. In notepad, create a file called cc_config.xml in directory F:\ProgramData\Boinc with these contents: <cc_config> <options> <http_1_0>1</http_1_0> </options> </cc_config> then restart boinc (or reboot). This will instruct boinc to make http/1.0 requests instead of 1.1. It helped me once when there was a misconfigured transparent proxy on my network. |
|
39)
Message boards :
Number crunching :
Slow and steady: Crunching on the cheap
(Message 1239726)
Posted 354 days ago by Khangollo
I'm afraid that FX5500 predates CUDA computing, big time. 8xxx series is the first that supports CUDA. |
|
40)
Message boards :
Number crunching :
62 AP_V5 Left In The Field
(Message 1239425)
Posted 355 days ago by Khangollo
THE LAST ONE http://setiathome.berkeley.edu/workunit.php?wuid=931699188 is still out there after the latest outage. Looks like another forever stuck workunit. Even if it gets returned (unlikely?), the WU will just stay in validation. More important question is... when will the admins change astropulse status display in status page to APv6. Is it really that difficult? |
| Copyright © 2013 University of California |