GNU/Linux S5R3 App 4.14 available for Beta test

th3
th3
Joined: 24 Aug 06
Posts: 208
Credit: 2208434
RAC: 0

I looked at your results

I looked at your results Tevault, thats a really impressive speedup, looks like 4.14 is even better for AMD than for Core 2. Is your 6000+ overclocked? My C2D Xeon is 20% higher clocked than a stock 6000+, and only ~30% faster than your 6000+. On average Core 2 is better clock for clock than that, so either 4.14 is better optimized for AMD K8 or this is some result of AMDs considerably faster memory controller. In any case, congratulations, insane improvement you got there.

ps, other users cant see your hostnames in your Computers list.

Donald A. Tevault
Donald A. Tevault
Joined: 17 Feb 06
Posts: 439
Credit: 73516529
RAC: 0

RE: I looked at your

Message 74664 in response to message 74663

Quote:

I looked at your results Tevault, thats a really impressive speedup, looks like 4.14 is even better for AMD than for Core 2. Is your 6000+ overclocked? My C2D Xeon is 20% higher clocked than a stock 6000+, and only ~30% faster than your 6000+. On average Core 2 is better clock for clock than that, so either 4.14 is better optimized for AMD K8 or this is some result of AMDs considerably faster memory controller. In any case, congratulations, insane improvement you got there.

ps, other users cant see your hostnames in your Computers list.

No, it's not overclocked. In fact, it runs so hot that I wouldn't even want to try. (It's still an old-style 90-nanometer design, so overclocking potential is severely limited.)

Odd, of all the times that I've looked at other folks' computer lists, I never noticed that the hostnames weren't visible. From now on, I'll reference computer ID numbers, instead. ;)

Edit:

The 6000+ machine is ID number 924295, and the P-III is number 813895.

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

Just finished a WU with 4.12

Just finished a WU with 4.12 and started a new one with 4.14. All seems OK, 190.78 credits were granted. All my downloads (Einstein, SETI and QMC) work full throttle with core client 5.10.21, so QMC@home has just gone in EDF panic mode. Only boincprog, which I use because it uses very little CPU, has not liked the transition from 5.8.17 to 5.10.21 and behaves lazily, not switching from one app to another. But this is not a great loss, I fire up the manager when I want to check what is going on, then I kill it to spare cycles of my brave PII Deschutes, running SuSE Linux 10.1.
Tullio

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

Something is going wrong with

Something is going wrong with 4.14. Here is an excerpt from stderr.txt:

2007-10-25 04:46:28.0726 [normal]: Start of BOINC application 'einstein_S5R3_4.1
4_i686-pc-linux-gnu'.
FPU status flags: COND_3 PRECISION
FPU masked exceptions now: 37f: PRECISION UNDERFLOW OVERFLOW ZERO_DIVIDE DENORM
ALIZED INVALID
FPU masked exceptions set: 37e: PRECISION UNDERFLOW OVERFLOW ZERO_DIVIDE DENORM
ALIZED
2007-10-25 04:46:29.6159 [debug]: Reading SFTs and setting up stacks ... done
2007-10-25 04:47:02.7584 [normal]: INFO: Couldn't open checkpoint h1_0237.65_S5R
2__24_S5R3a_2_0.cpt
2007-10-25 04:47:02.7587 [debug]: Total skypoints = 1181. Progress: 0,
$Revision: 1.75 $ OPT:0 SCV:2, SCTRIM:8
2007-10-25 04:52:47.3866 [CRITICAL]: ERROR: Couldn't sync h1_0237.65_S5R2__24_S5
R3a_2_0.cpt.tmp: line:811, ferr:0, errno:9: Bad file descriptor
c
1, 2007-10-25 04:58:32.0871 [CRITICAL]: ERROR: Couldn't sync h1_0237.65_S5R2__24
_S5R3a_2_0.cpt.tmp: line:811, ferr:0, errno:9: Bad file descriptor
c

Any hint?
Tullio

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4312
Credit: 250473470
RAC: 35063

RE: Any hint? You can

Message 74667 in response to message 74666

Quote:
Any hint?


You can put a file "EAH_NO_SYNC" in the BOINC directory and restart the Client in order not to immediately sync the file. However it would be interesting to know what's causing it. Do you know which filesystem you're using (ext2, ext3, xfs, ReiserF, whatever)?

BM

BM

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

RE: RE: Any hint? You can

Message 74668 in response to message 74667

Quote:
Quote:
Any hint?

You can put a file "EAH_NO_SYNC" in the BOINC directory and restart the Client in order not to immediately sync the file. However it would be interesting to know what's causing it. Do you know which filesystem you're using (ext2, ext3, xfs, ReiserF, whatever)?

BM


reiserfs, but it gives me no problem with either SETI and QMC.

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4312
Credit: 250473470
RAC: 35063

RE: reiserfs, but it gives

Message 74669 in response to message 74668

Quote:
reiserfs, but it gives me no problem with either SETI and QMC.


Well, the immediate syncing is a new feature, currently probably unique to E@h. ReiserFS is rather unusual, in fact I have not used it for years (and not tested the App on this); this might be the reason for the problems you see. Try to switch off syncing.

BM

BM

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

RE: RE: reiserfs, but it

Message 74670 in response to message 74669

Quote:
Quote:
reiserfs, but it gives me no problem with either SETI and QMC.

Well, the immediate syncing is a new feature, currently probably unique to E@h. ReiserFS is rather unusual, in fact I have not used it for years (and not tested the App on this); this might be the reason for the problems you see. Try to switch off syncing.

BM


I followed your suggestion, with no result. I've gone back to the standard app, and things seem OK. Trouble started with 4.14. Sorry.
Tullio

M. Schmitt
M. Schmitt
Joined: 27 Jun 05
Posts: 478
Credit: 15872262
RAC: 0

RE: RE: reiserfs, but it

Message 74671 in response to message 74669

Quote:
Quote:
reiserfs, but it gives me no problem with either SETI and QMC.

Well, the immediate syncing is a new feature, currently probably unique to E@h. ReiserFS is rather unusual, in fact I have not used it for years (and not tested the App on this); this might be the reason for the problems you see. Try to switch off syncing.

BM

I'm also using reiserfs and the next WU that gets downloaded will be crunched by the 4.14 app, so we might see if I get this error too.

Btw, Bernd, did you read my report about a lot of errors with the 4.09 app on a system under heavy load in the 4.09 thread? Since the code didn't change so much, this might still be relevant with 4.14.

cu,
Michael

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4312
Credit: 250473470
RAC: 35063

RE: Bernd, did you read my

Message 74672 in response to message 74671

Quote:
Bernd, did you read my report about a lot of errors with the 4.09 app on a system under heavy load in the 4.09 thread? Since the code didn't change so much, this might still be relevant with 4.14.


I read it. The signal handler used in 4.14 is supposed to give some more information about what's happening there. Actually I'm hoping this error occurs with a 4.14 App. Maybe you can help reproduce it?

BM

BM

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.