Welcome to MilkyWay@home

Posts by James Nelson

1) Message boards : Number crunching : Cache (Message 43027)
Posted 20 Oct 2010 by James Nelson
Post:
The 4 valid units took 40-43000 seconds. Also has tons of errors, most seem to have timed out.

my point is how did he download so many he should only have 100 or 120 at most how did he get so many tasks if I can only get 4 per core ????
2) Message boards : Number crunching : Cache (Message 43012)
Posted 20 Oct 2010 by James Nelson
Post:
That Opteron looks like a 4 way server using 6 core Opteron CPUs. So, BM tells the project that there are 24 cores (6 cores per CPU and 4 CPUs present). That means it can download 6 Milkyway work units per core, or a total of 24 x 6 = 144 WUs.

Is that a lot?

Seems to me it is the same as for me, with a quad core limit of 24 WUs.


If you look at his task list he has over 3000 I stopped looking at over 800 all showing in process.
3) Message boards : Number crunching : Cache (Message 43008)
Posted 20 Oct 2010 by James Nelson
Post:
If I can only have 4 units cashe per core how does this computer have so many ???
4) Message boards : Number crunching : Deadline problem (Message 34674)
Posted 18 Dec 2009 by James Nelson
Post:
Some people just don't want to help themselves.

An optimised application is the only way forward with this project. CPU or GPU. If you are not running an optimised app you really are "cutting your nose off to spite your face".


if you took the time to look you would see that I do run optimized they just keep upping the size till my slowest box cant return one in time.
5) Message boards : Number crunching : Deadline problem (Message 34663)
Posted 18 Dec 2009 by James Nelson
Post:
but it looks like the older systems will need more than 72 hours of CPU-time … in which case they’ll have to drop out.


I've already dropped my 400MHz host from MW because it can't finish a WU in time. Assuming that the benefits of leaving the deadline at 3 days outweighs the loss of some crunchers, I don't have a problem with that. But the project will have to consider how many hosts they will lose, when deciding to leave the deadline at 3 days, or to increase it.



Project has considered that and decided 3 days because they need wu's back as soon as possible to generate new wu's.
If they make deadline longer that would slow down the project.

And when a GPU can do the same wu in minutes, I don't think any project would care if a few very slow hosts that take days were dropped. It is but a fact of life.


If the project only cares about GPU and not about my lowliest cruncher then Ill vote with my feet NNT for all 4 boxes and then Ill be gone
6) Message boards : Number crunching : Panic Mode On (Message 7944)
Posted 23 Dec 2008 by James Nelson
Post:
[/quote]

Well, you have to make the best out of this situation, my other 8 backup projects are still there and waiting, all of them very interesting too !!

Just keep on crunching !

Kurt

edit : 7 projects, `cause RieselSieve isn`t with us anymore !!![/quote]

Just making a point that often two projects aren't enough. Murphy has a way insuring that at some point both will be out of work.
as you can see by my sig I have more than two active projects and my rigs will stay warm crunching for the others through the drought, no panic here.
7) Message boards : Number crunching : Panic Mode On (Message 7942)
Posted 23 Dec 2008 by James Nelson
Post:
The best advice I can give is to turn panic mode off!

A lot of Boinc projects run out of work from time to time, hey it happens, and for various reasons, but there are ways around the problem. Find a secondary project that you want to crunch for, and set MW preferences to 90% and the secondary project to 10% of processor time.

When MW runs out of work the secondary project automatically cuts in. At all other times you get 90% MW crunching time. Better than no work anywhere at all. I'm sure that Travis will kick the servers as and when he is able to, until then let's give the guys here a break. :-)


but when they both are out of work it is time to panic.
8) Message boards : Number crunching : Error on connect can't attach shared memory (Message 6630)
Posted 24 Nov 2008 by James Nelson
Post:
seems to be working now reported work and got more.
thanks.
9) Message boards : Number crunching : more wus generated: nm_test1 (Message 6423)
Posted 22 Nov 2008 by James Nelson
Post:
I renamed my app_info.xml and restarted boinc on all my systems and they are still using Milksop's 1.22 and the test app without any trouble so far.



Thank you! I have renamed the app_info.xml file and Milksop's 1.22 is indeed running. Have not gotten a download of the test app or any of the test WU's yet--yes I did set my preferences to allow the test app.

Best regards,
EigenState



What are you renaming in the app_info.xml file & what are you renaming it to ???


I think renaming the file isn't neccessary. Just add the new "test app" and there you go! ;-)
____________
Lovely greetings, Cori

just change the xml to bak and restart, unless you are good at writing app_info files and know where to download the new app from, you can't just add the new test app,because the app_info.xml file needs to have entry's that include the name of the test app and you have to download the files yourself boinc won't do it for you, then each time they release a new test app you need to start over, edit the app_info file and download the new app file.
10) Message boards : Number crunching : No Work ? (Message 6040)
Posted 8 Nov 2008 by James Nelson
Post:
I tend to prefer to maintain something of a balance in my project total work rankings -- as you can see from my signature here.

HUH! not to nitpick but your milkyway rac is almost 4 time as much as the rest put together how is that balanced
11) Message boards : Number crunching : Server Problems? - currently the U/L one (Message 5818)
Posted 2 Nov 2008 by James Nelson
Post:


any that I've gotten last two days have been re sends and by looking at the other results from those computers it looks like they just suspended MW and let them time out.
why not abort them or detach ???
seems some are boycotting the project.
12) Message boards : Number crunching : errors on 1.12 (Message 899)
Posted 3 Dec 2007 by James Nelson
Post:



This is pretty weird. We're looking into the memory leaks (and the app crashing on suspend) -- however i'm not quite sure why windows ME is acting so weirdly. I'm going to try and get access to a computer with that so we can test it.

Result ID 619215
Name gs_25_1196659793_12699_0
Workunit 581426
Created 2 Dec 2007 15:04:09 UTC
Sent 2 Dec 2007 15:15:35 UTC
Received 2 Dec 2007 21:34:58 UTC
Server state Over
Outcome Client error
Client state Compute error
Exit status -2147483645 (0xffffffff80000003)
Computer ID 2541
Report deadline 7 Dec 2007 15:15:35 UTC
CPU time 3557.335197
stderr out <core_client_version>5.10.7</core_client_version>
<![CDATA[
<message>
One or more arguments are invalid (0x80000003) - exit code -2147483645 (0x80000003)
</message>

well its not just windows ME this one is from a system running win 2K, the app really doesnt like to be suspended.


I've gotten the same problem with suspend a few times, and i'm taking a look into it. I think it might have something to do with the leaky memory.


Thanks for the update.
13) Message boards : Number crunching : errors on 1.12 (Message 892)
Posted 2 Dec 2007 by James Nelson
Post:



This is pretty weird. We're looking into the memory leaks (and the app crashing on suspend) -- however i'm not quite sure why windows ME is acting so weirdly. I'm going to try and get access to a computer with that so we can test it.

Result ID 619215
Name gs_25_1196659793_12699_0
Workunit 581426
Created 2 Dec 2007 15:04:09 UTC
Sent 2 Dec 2007 15:15:35 UTC
Received 2 Dec 2007 21:34:58 UTC
Server state Over
Outcome Client error
Client state Compute error
Exit status -2147483645 (0xffffffff80000003)
Computer ID 2541
Report deadline 7 Dec 2007 15:15:35 UTC
CPU time 3557.335197
stderr out <core_client_version>5.10.7</core_client_version>
<![CDATA[
<message>
One or more arguments are invalid (0x80000003) - exit code -2147483645 (0x80000003)
</message>

well its not just windows ME this one is from a system running win 2K, the app really doesnt like to be suspended.
14) Message boards : Number crunching : errors on 1.12 (Message 883)
Posted 2 Dec 2007 by James Nelson
Post:
I think I will push it up to 90 and see if it effects things.


James, have you tried the leaving in memory option. I have that set and have had no problems with units crashing on suspend (so far). Milkyway only seems to use about 16.7 Megs of memory, not sure about seti though. Dave.


used to do that but I stoped, this rig is limited on ram (160 M) and seti uses 32 megs i think and on occasion I would end up with two or even three instances in ram at once.
15) Message boards : Number crunching : errors on 1.12 (Message 877)
Posted 2 Dec 2007 by James Nelson
Post:
isnt it default that boinc switch between the projects you run on your machine every 60 minutes? at least my machines to that. it can be changed to highr/lower intervals in your boinc config.

there have been endless discussions about that, the switch intervol setting is kind of a rule of thumb, it has more to do with Short term debt as to when the switch accures, although I haven't tried to set it higher.
I think I will push it up to 90 and see if it effects things.
16) Message boards : Number crunching : errors on 1.12 (Message 875)
Posted 2 Dec 2007 by James Nelson
Post:

This is pretty weird. We're looking into the memory leaks (and the app crashing on suspend) -- however i'm not quite sure why windows ME is acting so weirdly. I'm going to try and get access to a computer with that so we can test it.[/quote]

I'm not sure if its just a win ME problem as my other hosts run the wu so fast that they never need to get suspended.
the host in question is a 500 Mhz celeron laptop, most units run in under an hour but a few run longer, thats when the trouble starts, boinc suspends milkyway and runs seti. now with the new partical swarm units this host might not be able to complete one if this isnt resolved.

Thankyou for all that you have done and I appriciate all your hard work and dedication.

hopefully this old laptop can continue to contribute to your project, if not its not the end of the world.
17) Message boards : Number crunching : errors on 1.12 (Message 862)
Posted 1 Dec 2007 by James Nelson
Post:
Result ID 554311
Name gs_21_1196403149_1639_0
Workunit 518316
Created 29 Nov 2007 20:55:57 UTC
Sent 29 Nov 2007 21:05:11 UTC
Received 1 Dec 2007 19:51:33 UTC
Server state Over
Outcome Client error
Client state Compute error
Exit status -1 (0xffffffffffffffff)
Computer ID 2538
Report deadline 4 Dec 2007 21:05:11 UTC
CPU time 3547
stderr out <core_client_version>6.1.0</core_client_version>
<![CDATA[
<message>
- exit code -1 (0xffffffff)
</message>
<stderr_txt>
Unrecognized XML in GLOBAL_PREFS::parse_override: suspend_if_no_recent_input
Skipping: 0.000000
Skipping: /suspend_if_no_recent_input
APP: error writing checkpoint (resolving checkpoint file) -108
output_path: astronomy_checkpoint


**********
**********

Memory Leaks Detected!!!

Memory Statistics:
0 bytes in 0 Free Blocks.
5601186 bytes in 200031 Normal Blocks.
1088 bytes in 3 CRT Blocks.
0 bytes in 0 Ignore Blocks.
0 bytes in 0 Client Blocks.
Largest number used: 5610561 bytes.
Total allocations: 6121729 bytes.



app is still craching when it is suspended
18) Message boards : Number crunching : errors on 1.12 (Message 830)
Posted 30 Nov 2007 by James Nelson
Post:
Have any worked at all? I'm wondering if this is a windows ME problem. If so it's going to be interesting finding a windows ME machine to test on *lol*.


well the next one went fine too so I guess all is well.
19) Message boards : Number crunching : errors on 1.12 (Message 817)
Posted 30 Nov 2007 by James Nelson
Post:
Have any worked at all? I'm wondering if this is a windows ME problem. If so it's going to be interesting finding a windows ME machine to test on *lol*.

not yet Ill keep an eye on it to see when the next one runs.
an interesting thing was when i looked at it there was one running with no time progresing (stuck on like 5m 48sec ) with two others waiting,(each with some progress) shortly after I noticed this all three failed one after the other.


the next one went no problem, the first ones might have been a fluke, Ill keep an eye on this host and see how it goes.the strange thing was why did it start one stop, start another stop, and then start a third.
20) Message boards : Number crunching : errors on 1.12 (Message 812)
Posted 29 Nov 2007 by James Nelson
Post:
Have any worked at all? I'm wondering if this is a windows ME problem. If so it's going to be interesting finding a windows ME machine to test on *lol*.

not yet Ill keep an eye on it to see when the next one runs.
an interesting thing was when i looked at it I was one running with no time progresing (stuck on like 5m 48sec ) with two other waiting,(each with some progress) shortly after I noticed this all three failed one after the other.


Next 20

©2024 Astroinformatics Group