Posts by Nuadormrac

41) Message boards : RALPH@home bug list : RALPH Version News! - Version 4.99 (Windows) released! (Message 1080)
Posted 12 Apr 2006 by Nuadormrac
Post:
It just gave me a bunch of HBLR_* WUs... Lets keep our fingers crossed that 4.99 fairs better on these...
42) Message boards : RALPH@home bug list : RALPH Version News! - Version 4.97 (Win/Lin/Mac) released! (Message 1073)
Posted 11 Apr 2006 by Nuadormrac
Post:
As for the Macs they have been running all of the Ralph and Rosetta versions with out problems. The current situation is mostly a windows problem. But it is also affecting systems that are marginal on memory as the protien being tested in the new work units is the largest I have seen to date. Systems with 256Mb of memory are just going to have trouble with the new workunits. Even system with 512Mb where some of the memory is in use for other purposes may have trouble.[/color][/b]


Umm, it isn't exactly a lack of memory situation I was running into... I do have a GB of RAM in my system, not 256 or 512 MB, and what's more my computer can quite happily crunch a CPDN combined ocean WU (this is in fact the same WU they distribute in the BBC climate model project, but they're distributing them on CPDN). What's more, these combined ocean models take a lot longer then a sulpher cycle to run, and are also more resource intensive then sulpher cycles according to the people at the CPDN project. Seasonal attribution, which recommends no less then 1 GB of RAM runs quite happily on my PC also...

Unless these models are larger/more intensive then CPDN work units, I think something else might have been going on, at least here. It also isn't that these projects were running, because earlier I had done backups on each, bringing them both to a save point (have them both scheduled in BOINC), and then suspended them. Shutdown BOINC, backed up the folders, and did not resume then. The projects hadn't loaded back up when I was running RALPH, and only now resumed CPDN with the coupled oceans model, while it's finishing up some LHC units...
43) Message boards : Current tests : WinXP 64bit/AMD64bit-Support? (Message 1057)
Posted 9 Apr 2006 by Nuadormrac
Post:
As for 64bit, well it would be a nice thing (as in run natively).
But you would think solaris and MAC-intel would come first as they are official BOINC platforms ;-)

... I think there is already this suggestion in Feedback as it's not a current test....


Things might begin to change next year, as it's already been mentioned that Conroe (Intel's to be replacement for the Pentium 4), not only will drop the whole net-burst arch, but will be a 64-bit x86 processor as well. Conroe will also be dual proc... Given the A64s/x2s, as well as Conroe supplying x86-64 procs from both Intel as well as AMD, I imagine a look at 64-bit software support will begin to increase. Afterall, when both PC proc manufacturers will be supporting it...

The only aside that leaves is the OS, and on the Windows sphere, and unlike winXP, Vista will be compiled with both 32-bit and 64-bit variants (which would run native 64-bit on both the A64s as well as the up-comming Conroe), right out the gate. Won't be like XP-64 which was made sometime latter...

I do suspect that after Conroe, as well as Vista make there appearance, we might begin to see a change here. In part, because people will probably start getting 64-bit systems straight from Dell and other OEMs...
44) Message boards : RALPH@home bug list : RALPH Version News! - Version 4.97 (Win/Lin/Mac) released! (Message 1056)
Posted 9 Apr 2006 by Nuadormrac
Post:
On checking Rosetta's forum, version 4.98 is really a rollback to version 4.83, not a newer version then 4.97. Apperently 4.97 was released onto Rosseta and caused many a problem/discussion there. This prompted a rollback, which was given a latter version number so as to convince BOINC to download the older version I gather...

http://boinc.bakerlab.org/rosetta/forum_thread.php?id=1106

I just got this message from David Kim who is currently addressing this problem.

"I just reverted back to the previous app. You should notice a version
4.98 now, which is really version 4.83 for windows and mac, and 4.82
for linux."

You all should see some relief very soon. If you force an update it should load the new version once the server is set up.


Some of the users error reports and what not can be seen in that thread. I imagine they still need us to test 4.97, and that a newer version to test will be comming out soon.

Which brings up a bit of an aside. What happens when we hit 4.99 and possibly need another version number. Do we go to 5.00 or? I only ask, because usually the number before the decimal place, with many software packages is a major version number, commonly associated with major changes/feature updates. Think of BOINC 4.x vs. 5.x for instance... Not that it entirely matters in the grand scheme of things if we do need to move to 5.xx on the science app number...
45) Message boards : RALPH@home bug list : Report - Previously Unclassified Work Unit Errors (Message 1055)
Posted 9 Apr 2006 by Nuadormrac
Post:
OK, this was a WU that was stuck at about 19%, so does not classify as a 1% hang. It was also the newer type (not the HB...) 7449_largescale* Now, all other units of this type that I've thus far received, have completed successfully and without incident. So seems to be a lone one.

What it did, was got up to model 2, step 0, and then it just sat there/hung without any progress at all. It ran for longer then some of the longest running RALPH units I got thus far, but exhibited one other oddity. On the accepted energy graph on the left hand side of the screen, when I looked at it, the thing was a complete blur for the most part, with no line or data points in the least bit visable. I've never seen the graph become a blured out/washed out mess like that when I've looked, and didn't seem normal.

I let it run a bit longer (to have it's run time go beyond about the longest run times other units that went successfully. The thing was just searching wildly, and on the graph portion bluring up to an indistinguishable mess... I aborted then... Here's the result:

http://ralph.bakerlab.org/result.php?resultid=82276
46) Message boards : Number crunching : Interesting credit differences between Linux & Windows wu\'s (Message 1053)
Posted 9 Apr 2006 by Nuadormrac
Post:
I would have to look into it further to confirm, but thought I read somewhere that the way Linux counts CPU time vs. the way Windows counts it differs a bit. I'd have to check up on it again before I'd feel comfortable being quoted on what I'm only half remembering now...

If they're close in actual run times then this is a good sign, as the Linux variant has many times had lower performance... Einstein@home was facing that for a bit, with continual tweaks to the linux science app being made. Now that akosf has his optimized Albert app, and after some of his improvements get introduced into the official science app; I imagine this discrepency will increase on that project again. Well, until further work, allows some of his improvements to be ported over to the Linux variant of their science app also...
47) Message boards : Number crunching : Welcome back... (Message 1052)
Posted 9 Apr 2006 by Nuadormrac
Post:
Yeah, it's a test project... As such, they only need us here when testing apps. Otherwise Rosetta is where they need crunchers on a continual basis.

For the past day LHC had gotten a burst of available work, so my BOINC client has been running RALPH for testing and LHC...
48) Message boards : RALPH@home bug list : RALPH Version News! - Version 4.97 (Win/Lin/Mac) released! (Message 1051)
Posted 8 Apr 2006 by Nuadormrac
Post:
OK, a pattern seems to have emerged. The WUs that are having problems (I had another 7 go bad) are all of type HBLR_1.0_*, the other parts of the WUs having changed. I'm gathering this is a new WU type. The succession of successes, were of type barcode*, and were WU types I had seen in the past, so I guess that WU type had been nailed down already.

4 went late morning, so I rebooted. Never had that problem on Rossetta itself, but had seen on the boards that some mentioned it fixed things, so thought after 4 failures I'd try it here anyhow... I then got 2 successful completions with 2 failures, tried rebooting again just to see what would happen. Shutting down BOINC, rebooting, and then restarting comp then caused this one to fail... (Definitely a problem associated with removing from memory in my case.)

http://ralph.bakerlab.org/result.php?resultid=81520

However, one other person also got an access violation on that unit, so not sure what was going on around the time of failure in their case. Definitely a repeated failure...

The other failed WUs (and these do not have the same removal from memory problem associated with them at least in my case)

http://ralph.bakerlab.org/workunit.php?wuid=75104

This one also has a second failure beyond the one on my comp
---------------------
http://ralph.bakerlab.org/workunit.php?wuid=75045

3 failure results
--------------------
http://ralph.bakerlab.org/workunit.php?wuid=75894

2 failures but one success
---------------------
http://ralph.bakerlab.org/workunit.php?wuid=75780
---------------------
http://ralph.bakerlab.org/workunit.php?wuid=75778

all 3 results show failure
---------------------
http://ralph.bakerlab.org/workunit.php?wuid=75777
---------------------

All failures are for the same error that both I and others have been reporting. The access violation...
49) Message boards : RALPH@home bug list : RALPH Version News! - Version 4.97 (Win/Lin/Mac) released! (Message 1049)
Posted 8 Apr 2006 by Nuadormrac
Post:
OK, make that a second

http://ralph.bakerlab.org/result.php?resultid=80149
50) Message boards : RALPH@home bug list : RALPH Version News! - Version 4.97 (Win/Lin/Mac) released! (Message 1048)
Posted 8 Apr 2006 by Nuadormrac
Post:
Perhaps I spoke a bit too soon, but then again, perhaps not. Percentage wise, had far more successes then before, but one did just fail...

http://ralph.bakerlab.org/result.php?resultid=80038

stderr out

<core_client_version>5.2.13</core_client_version>
<message> - exit code -1073741819 (0xc0000005)
</message>
<stderr_txt>
# random seed: 3893788
# cpu_run_time_pref: 3600

***UNHANDLED EXCEPTION****
Reason: Access Violation (0xc0000005) at address 0x007022EA read attempt to address 0x0EC5FC1C


Dump of the Worker(offending) thread:
1: 04/08/06 04:25:45


Dump of the Timer thread:
2: 04/08/06 04:25:45


Dump of the Graphics thread:
3: 04/08/06 04:25:45


Exiting...

</stderr_txt>
51) Message boards : RALPH@home bug list : RALPH Version News! - Version 4.97 (Win/Lin/Mac) released! (Message 1042)
Posted 8 Apr 2006 by Nuadormrac
Post:
Don't know if this is worth mentioning or not. But thus far 4.97 has been working fine on my winXP SP2 A64 box... All WUs I've started crunching have been returning successfully. If I do run into a problem, I'll be sure to report it, but all seems well for on my comp for now.

BTW, this is a bit of an aside from previous versions where 1 of the WUs I had gotten out of 3 or so, tended to exit with a problem. I've gone through 6 or so now on this version...
52) Message boards : RALPH@home bug list : 4.92 crash w/ MS popup (Message 876)
Posted 14 Mar 2006 by Nuadormrac
Post:
Got another one of these

The culprit here

tderr out

<core_client_version>5.2.13</core_client_version>
<message> - exit code -1073741811 (0xc000000d)
</message>
<stderr_txt>

</stderr_txt


<?xml version="1.0" encoding="UTF-16"?>
<DATABASE>
<EXE NAME="rosetta_beta_4.92_windows_intelx86.exe" FILTER="GRABMI_FILTER_PRIVACY">
<MATCHING_FILE NAME="rosetta_beta_4.92_windows_intelx86.exe" SIZE="6541312" CHECKSUM="0xE32C363" MODULE_TYPE="WIN32" PE_CHECKSUM="0x645CD7" LINKER_VERSION="0x0" LINK_DATE="03/03/2006 00:05:42" UPTO_LINK_DATE="03/03/2006 00:05:42" />
</EXE>
<EXE NAME="kernel32.dll" FILTER="GRABMI_FILTER_THISFILEONLY">
<MATCHING_FILE NAME="kernel32.dll" SIZE="983552" CHECKSUM="0x4CE79457" BIN_FILE_VERSION="5.1.2600.2180" BIN_PRODUCT_VERSION="5.1.2600.2180" PRODUCT_VERSION="5.1.2600.2180" FILE_DESCRIPTION="Windows NT BASE API Client DLL" COMPANY_NAME="Microsoft Corporation" PRODUCT_NAME="Microsoft® Windows® Operating System" FILE_VERSION="5.1.2600.2180 (xpsp_sp2_rtm.040803-2158)" ORIGINAL_FILENAME="kernel32" INTERNAL_NAME="kernel32" LEGAL_COPYRIGHT="© Microsoft Corporation. All rights reserved." VERFILEDATEHI="0x0" VERFILEDATELO="0x0" VERFILEOS="0x40004" VERFILETYPE="0x2" MODULE_TYPE="WIN32" PE_CHECKSUM="0xFF848" LINKER_VERSION="0x50001" UPTO_BIN_FILE_VERSION="5.1.2600.2180" UPTO_BIN_PRODUCT_VERSION="5.1.2600.2180" LINK_DATE="08/04/2004 07:56:36" UPTO_LINK_DATE="08/04/2004 07:56:36" VER_LANGUAGE="English (United States) [0x409]" />
</EXE>
</DATABASE>
53) Message boards : Number crunching : Team stats not adding up... (Message 872)
Posted 14 Mar 2006 by Nuadormrac
Post:
Actually it is, if one wasn't previously a member of a team. However, once credit has been assigned, it can't be re-assigned...

Looking at other projects, it was accumulated as such, from the apperance of it. Note however, if one was a member of a team on something (or signed up for the wrong listing), and then moved it, one doesn't show up in the team ranks with that initial credit. It's more a handling on un-assigned credits... Course 2 teams couldn't get the same credit, but in this case doesn't apply.

For instance, in QMC, the totals were most definitely added... These of course are the easiest to calculate/compare, because both being new/alpha projects, the numbers are small enough, with few enough team members that the math can be easily compared in one's head.
54) Message boards : Number crunching : Team stats not adding up... (Message 870)
Posted 13 Mar 2006 by Nuadormrac
Post:
Not sure this is something effecting Rosseta or not, though thought I would mention... Joined a team recently, and well the math doesn't seem consistent

http://ralph.bakerlab.org/team_display.php?teamid=61

Total credit 132.72
Recent average credit 4.74

Name Total credit Recent average credit Country
1) Son Goku 527.02 32.25 United States
2) Wabbit98 79.27 4.89 Channel Islands
3) Tazman 32.84 1.63 Australia
4) Clark 20.60 1.47 United Kingdom


Note, that both my total and my RAC are higher then the team total and RAC, shown below, but not computed above. Thought I'd pass this lil apperent computation issue along...
55) Message boards : Feedback : Difference F@H and R@H??? (Message 869)
Posted 13 Mar 2006 by Nuadormrac
Post:
There are some differences, which were also posted at the ProteinPredictor board, comparing it to both folding@home, as well as to rosseta@home... BTW, as my credits might indicate, I was on Predictor@home longer, but also joined up with Rosseta, not long after it was released... The view there when people asked, as that the work done here (well on the parent project, not the alpha of course) is complementary, and it's of value to crunch for both...

As to folding, I have crunched for them in the past; however I'm now using BOINC. If they want to get me to consider crunching there again, then they'll need to get the BOINC client out. Though there is a way to split CPU time between both apps (by throttling the f@h CPU time back, so BOINC gets some CPU time also); it's easier to just allow the BOINC CC to task switch between projects, especially as more and more projects are setup to run in BOINC.

/me just not as interested in a stand alone client anymore...
56) Message boards : Number crunching : Nothing happening (Message 867)
Posted 13 Mar 2006 by Nuadormrac
Post:
My BOINC client keeps saying communication deferred for RALPH and then in the messages section it says 'NO WORK FROM CLIENT'

Any ideas? I seem to be sliding down in the results league so i presume others have work to process


Sign up for more projects ;) Rosetta itself could use crunching...

BTW, I do have a bit of a question wrt this work shortage myself, but it is in no way a complaint, or a gripe as was referenced above... Do we have a stable version that is working on Rosetta now? Or is there more testing that is comming up the pike? I ask in part because some settings for Rosseta vs. Ralph (aka the remove from memory) weren't entirely compatible, and in part because I'm signed up for so many projects now, I disable/rotate some to keep my Athlon 64 from getting too bogged down... Eh OK, if it takes a couple days to complete some WUs it would still be in deadline, and projects probably wouldn't care. But somewhat out of respect for each, I've tried to keep crunch times down. Meh, I could use an X2 (aka dual proc) sometime down the pike :D

Just wondering if I should start a Rosseta WU now, or if Ralph WUs are going to be need crunching soon, so we can work out some more of the app bugs tis all ;)
57) Message boards : RALPH@home bug list : 4.92 WU errors with \"incorrect function (0x1)\" (Message 846)
Posted 11 Mar 2006 by Nuadormrac
Post:
Just got one of these WUs

http://ralph.bakerlab.org/result.php?resultid=15707

<core_client_version>5.2.13</core_client_version>
<message>Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
ERROR:: Unable to obtain sequence information.
fasta file must be provided.


Further looking at the WU shows the only other person who has tried to return anything on this WU, has also gotten a client error.

http://ralph.bakerlab.org/workunit.php?wuid=12658

15655 1679 11 Mar 2006 0:30:05 UTC 11 Mar 2006 1:36:46 UTC Over Client error Computing 0.00 0.00 ---
15707 1024 11 Mar 2006 1:37:15 UTC 11 Mar 2006 9:45:04 UTC Over Client error Computing 0.00 0.00 ---
16012 1454 11 Mar 2006 9:45:11 UTC 25 Mar 2006 9:45:11 UTC In Progress Unknown New --- --- ---
58) Message boards : Number crunching : BOINC Wiki Will Continue! (Message 801)
Posted 3 Mar 2006 by Nuadormrac
Post:
Hmm, Paul is leaving?

I guess it was a lot for one guy to have put the level of commitment he had put into this. Having gone to (well most of the message boards I have been on wrt BOINC itself), having spoken with the various devs, gathered the info, and put the wiki together and all...

Well, for whatever it's worth, he did one hell of a good job up to present :thumbsup: Sorta surprised at this announcement however...
59) Message boards : Number crunching : WU housekeeping (Message 800)
Posted 3 Mar 2006 by Nuadormrac
Post:
Actually there is 146 WUs according to the front page now, and there was about 343... I just downloaded a RALPH WU, which is about 14.9% done now. So should I manually delete, wait to reset?

It would be good if this got cleared out. Also, don't know if it was the suspending of those units before (now abortion), or something else, but my client's getting less work from project when WUs are available now, vs. before...
60) Message boards : Number crunching : WU housekeeping (Message 795)
Posted 3 Mar 2006 by Nuadormrac
Post:
To avoid the penalty wrt canceling several WUs, I did so slowly wrt the 4.86 WUs we were asked to cancel. In the interum I suspended them, so my comp wouldn't spend time processing WUs the project admins already asked us to cancel...

However, having aborted another one of the 4.86 today, upon a reconnect, BOINC would neither upload the aborted unit, or clear it from the list of WUs. I then tried resuming the result, and getting it to report back, and then abort a second time...

I'd rather not the old WU stay on my list? How do I clear this out of BOINC manager? What would I manually delete, if I need to do this? Somehow BOINC needs to get rid of these 4.86 WUs and stop listing them...

? ? ?

Anyhow, my core client is 5.2.13...


Previous 20 · Next 20



©2024 University of Washington
http://www.bakerlab.org