Posts by Furex - [Ometti Verdi]

1) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 518224)
Posted 16 Feb 2007 by Furex - [Ometti Verdi]
Post:
I guess that according Intel even the release of the original Athlon would be illegal.

Of course crippling a compiler to harm your competitor is absolutely legal. Wonder why they didn't start bombing AMD fabs, that must be legal for them, too ;)
2) Message boards : Number crunching : Memory Bandwidth is uuurhting me (Message 516753)
Posted 12 Feb 2007 by Furex - [Ometti Verdi]
Post:
:) you call that a rant Jan? very good post, it was funny to read.

However, I wonder what's the reliability (when it comes specifically to SETI performance) of the memory benchmark which is built in to the KWSN apps.

This is what I get on an Opteron 2.7 with DDR420 2-2-2-5 1T, GenSSE3:

speed: 2666 MHz -- read megs/sec: L1=14617, L2=7248, RAM=3709

Memtest86+ reported similar numbers IIRC.
3) Message boards : Number crunching : Memory Bandwidth is uuurhting me (Message 516689)
Posted 12 Feb 2007 by Furex - [Ometti Verdi]
Post:
Dual-channel AMD systems using DDR2 can achieve over 2GBytes/s memory bandwidth.


Only 2 GB/s ? :D I should be glad I'm still running DDR ;D seriously, what's the preferred benchmark to rate the effect of memory bandwidth on SETI performance?
4) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 516687)
Posted 12 Feb 2007 by Furex - [Ometti Verdi]
Post:
to hiamps, lurkers and all that can crunch without errors :)

please post your experience comparing GenSSE2, IntelSSE2 patched and IntelSSE3, my teammates (several SETI+AMD crunchers) would find it very useful. Thank you!
5) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 516351)
Posted 11 Feb 2007 by Furex - [Ometti Verdi]
Post:
You're right. Must be something internal to the Opteron vs. the Athalon X2.


The Dual Core Opteron 1xx has 1 MB L2 cache per core, while your 4400+ should have only 512 kB.

But I don't think it should make such a big difference, my Athlon64 3200+ Winchester (single core with 512 kB L2) did the 62.40 in 8500-8900 seconds at 2300 MHz with KWSN 2.0 SSE2-Generic.
6) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 516197)
Posted 11 Feb 2007 by Furex - [Ometti Verdi]
Post:
Which patch did you use?

The one Simon provided seems to work, I've got a couple of valid results:

http://setiathome.berkeley.edu/result.php?resultid=471542706
http://setiathome.berkeley.edu/result.php?resultid=471542682

These two long (62.40 cr) units seemed a little bit faster (maybe a 10%), yet it's too early to say anything for sure.
7) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 516036)
Posted 11 Feb 2007 by Furex - [Ometti Verdi]
Post:
I did this before I tried Seti w/the patched files and I stopped Boinc before any WU's were reported as Computation errors


Excuse me, it must be the late hour but something's not pretty clear to me now. You're saying that your newly patched SSE3-Generic has problems ? Or is this just a generic (welcome nonetheless) advice?
8) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 515983)
Posted 10 Feb 2007 by Furex - [Ometti Verdi]
Post:
Thank you Simon ;) (been reading that "Opteron and SSE3" thread on lunatics.at)
9) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 515981)
Posted 10 Feb 2007 by Furex - [Ometti Verdi]
Post:
Hmmmm it might be a perl interpreter issue. I just tried the precompiled version (iccpatch.exe) and THAT worked.

Now it's time to try it on the field :)

edit:

Confirmed! It works with ActivePerl 5.8.8 MSWin32-x86-multi-thread, but not with cygwin's perl 5.8.7 cygwin-thread-multi-64int.
10) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 515976)
Posted 10 Feb 2007 by Furex - [Ometti Verdi]
Post:
This is with ICCpatch:

$ ./iccpatch.pl -v SaH_5.15_KWSN_SSE3_Ben-Joe_2.0_B.exe
Use of uninitialized value in concatenation (.) or string at ./iccpatch.pl line
51.
Patching file ...
WARNING: 10 lines were to be patched in 'SaH_5.15_KWSN_SSE3_Ben-Joe_2.0_B.exe': should be a multiple of 3! YMMV, something may go wrong.
Patching SaH_5.15_KWSN_SSE3_Ben-Joe_2.0_B.exe in 10 places...Patch operation for 'SaH_5.15_KWSN_SSE3_Ben-Joe_2.0_B.exe' successful, original file at 'SaH_5.15_K
WSN_SSE3_Ben-Joe_2.0_B.exe~'


When I run the patched executable, I receive this error:

X:\\cygwin\\home\\noyb>SaH_5.15_KWSN_SSE3_Ben-Joe_2.0_B.exe
Access is denied.
11) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 515969)
Posted 10 Feb 2007 by Furex - [Ometti Verdi]
Post:
Just tried the supplied script with ActivePerl and cygwin, no matter what it spits out this error:

.\\intel_check_executable_patch -v SaH_5.15_KWSN_SSE3_Ben[..]

Use of uninitialized value in concatenation (.) or string at x:\\cygwin\\home\\noyb\\intel_check_executable_patch.pl line 51.
Patching file ...
WARNING: 10 lines were to be patched in 'SaH_5.15_KWSN_SSE3_Ben-Joe_2.0_B.exe':
should be a multiple of 3! Skipping this file.

edit: thanks for the fast reply Simon :) my Opteron is a JH9-E6 so maybe it might benefit a little.
12) Message boards : Number crunching : Open Message to "KWSN - Chicken of Angnor" (Message 515966)
Posted 10 Feb 2007 by Furex - [Ometti Verdi]
Post:
Hello Simon,

I thought that SSE3 didn't offer any speed improvement on the A64. Can you confirm this?
13) Message boards : Number crunching : KWSN 2.0 ...B? (Message 459360)
Posted 16 Nov 2006 by Furex - [Ometti Verdi]
Post:
The stderr output had some leftovers from development, which were removed with the "B" version.


Eager to see the new output, I might go back to "A" if it's more geeky :D

Thank to all of you who have worked on this optimized app, it really rocks!
14) Message boards : Number crunching : KWSN 2.0 ...B? (Message 459281)
Posted 16 Nov 2006 by Furex - [Ometti Verdi]
Post:
Just found out when downloading clients for my team mates.

Simon, mind if I ask what's the difference between A and B?
15) Message boards : Number crunching : Inconsistent Team Statistics (Message 457687)
Posted 14 Nov 2006 by Furex - [Ometti Verdi]
Post:
This screenshot clearly shows there are five members with credit > 0 thus, according to my pretty limited language skills, the "members (with credit)" figure should sum up to 5.



I hope the problem is clear enough, now.
16) Message boards : Number crunching : Inconsistent Team Statistics (Message 457342)
Posted 14 Nov 2006 by Furex - [Ometti Verdi]
Post:
The BOINCStats site is a 3rd party site.


I know, in fact I was talking about the official http://setiathome.berkeley.edu, and only quoted boincstats to explain that I probably jumped to conclusions too fast about the second link I've reported.

However, for our team, the official site keeps stating #members (with credits) < #active members.

This can't be regarded as boincstats' fault ;)
17) Message boards : Number crunching : Inconsistent Team Statistics (Message 456792)
Posted 13 Nov 2006 by Furex - [Ometti Verdi]
Post:
Our "members with credit" figure has been stuck at 3 for days, we're not talking about yesterday's update: our fourth member have reported results (and had credit accounted for) every day in the last few days, and have always been in the active members list: nonetheless the team has only 3 members :)

As for the other team I've linked, looking at the last months on boincstats, probably the team has been founded some months after the founder joined the SAH project. So it shouldn't be held in account.
18) Message boards : Number crunching : Inconsistent Team Statistics (Message 456742)
Posted 13 Nov 2006 by Furex - [Ometti Verdi]
Post:
Hello all,

I've noticed some inconsistencies in the team statistics, here are two examples:


  • In my team there are 5 active members, but team info says there are 3. We have all reported results (and had credit accounted for) today or yesterday.
  • In this other team, total credit says 16,504.20 but the only active member is over 22,300.



Can anyone explain me what's going on?

19) Questions and Answers : Web site : Problem joining team (Message 454228)
Posted 9 Nov 2006 by Furex - [Ometti Verdi]
Post:
How would I force the page to reload to get the page to display the join link?


Just press Ctrl-F5 :)
20) Questions and Answers : Web site : Problem joining team (Message 453546)
Posted 8 Nov 2006 by Furex - [Ometti Verdi]
Post:
We're having the same problem...

The "Join Team" link is missing from the team page, and there are other teams without the join link.

What's going on? please help...

edit: nevermind, it is now fixed; it needed a forced page reload to show up, though.





 
©2020 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.