v5.22 Experimental optimized app

Message boards : Number crunching : v5.22 Experimental optimized app

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 13 · Next

AuthorMessage
Profile Eremita

Send message
Joined: 20 Sep 07
Posts: 2
Credit: 1,513
RAC: 0
Message 326 - Posted: 30 Nov 2007, 22:49:44 UTC

Hi to all at Enigma.

I tried the win 64 bit executable on a AMD 64x2 running boinc 6.1

I obtained 5 errors

30/11/2007 23.00.09|Enigma@Home|Reason: Unrecoverable error for result hceyz72_0_477516_r0_0 ( - exit code -148 (0xffffff6c))

on 5 WU

http://www.enigmaathome.net/result.php?resultid=433598 is one

<core_client_version>6.1.0</core_client_version>
<![CDATA[
<message>
- exit code -148 (0xffffff6c)
</message>
<stderr_txt>
running enigma_1.00_windows_x86_64.exe
can't run app: -148

</stderr_txt>
]]>

Is a boinc core client problem or an app problem?

Thanx

Andrea



ID: 326 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
SAK

Send message
Joined: 10 Nov 07
Posts: 2
Credit: 639,323
RAC: 0
Message 327 - Posted: 1 Dec 2007, 1:41:29 UTC

CPU temp is fine, no dust accumulation (I run a Thermalright Ultra 120 Extreme cooler with filtered, ducted air).

I'm pretty certain it's the experimental client, and not the computer in any way. I put the standard client back on, and I will tell you if I have any problems (I don't think I will).

I've run every torture test I have on my system, and it's perfectly stable. I'm not sure what else I could try...perhaps Prime95 or something?

The only error code I could offer you is the error code reported AFTER reboot (one of those "System has recovered from a serious error..." ones).

0000: 53 79 73 74 65 6d 20 45 System E
0008: 72 72 6f 72 20 20 45 72 rror Er
0010: 72 6f 72 20 63 6f 64 65 ror code
0018: 20 30 30 30 30 30 30 39 0000009
0020: 63 20 20 50 61 72 61 6d c Param
0028: 65 74 65 72 73 20 30 30 eters 00
0030: 30 30 30 30 30 30 2c 20 000000,
0038: 62 61 62 34 34 30 35 30 bab44050
0040: 2c 20 62 32 30 30 30 30 , b20000
0048: 34 30 2c 20 30 30 30 30 40, 0000
0050: 30 38 30 30 0800

Doubt that will be helpful, but there ya go. I would be happy to help you in any way figure out what the problem is. Just tell me what to do :) You can reach me at email address.

ID: 327 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 328 - Posted: 1 Dec 2007, 2:28:40 UTC - in response to Message 327.  

First off, best not ever post your email address in a readable form on forums. These are read by spam-bots as well.

Error code 0000009c

I get these at times. In my case it's because my motherboard capacitors are bulged and need changed (or the mobo needs changed, but I don't have the money for that).
In any case, this error always indicates there's a (intermittent) hardware error.

This could be CPU, RAM, motherboard or any add-on cards. See this article in the MSKB.

Jord.

BOINC FAQ Service.
ID: 328 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Anonymous
Avatar

Send message
Joined: 11 Sep 07
Posts: 19
Credit: 474
RAC: 0
Message 329 - Posted: 1 Dec 2007, 16:32:42 UTC - in response to Message 328.  

First off, best not ever post your email address in a readable form on forums. These are read by spam-bots as well.

If you hold your mouse over the email link, you'll see that it's a web page link, not a mailto: link; I tried clicking on it, and it turns out that it takes you to a page where you enter a CAPTCHA and it presumably gives you the email address then. (I say presumably because I didn't actually try it.) That should keep the bots out. :-)
ID: 329 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 330 - Posted: 1 Dec 2007, 17:41:28 UTC - in response to Message 329.  

[If you hold your mouse over the email link, etc.

When I answered that yesterday it was in a legible form, so my guess is that some angel around here changed that for him, as there's no modification notice on the post and I don't have an edit button.... but TJM may have one. :-)
Jord.

BOINC FAQ Service.
ID: 330 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,762,956
RAC: 385,630
Message 331 - Posted: 1 Dec 2007, 20:45:09 UTC - in response to Message 313.  


Judging by the fact that my box has been 100% stable before installing the experimental app, I'm going to have to say there is some stability problem with it. I will keep running it, and see how it goes. If I keep experiencing the problem, I will switch back to the previous app, and let you know.


IMO there's no way that such a simple app can crash the entire system, unless you run it on a Win98 %-) It's build from the same source as the default one, only with different compiler flags. I did a lot of experiments with apps and I newer saw anything worse than a simple app error.


M4 Project homepage
M4 Project wiki
ID: 331 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,762,956
RAC: 385,630
Message 332 - Posted: 1 Dec 2007, 20:48:52 UTC - in response to Message 330.  
Last modified: 1 Dec 2007, 20:54:15 UTC

I don't have an edit button.... but TJM may have one. :-)


No, I don't have edit button, I edited it in the db. Posting email adresses here is not a very good idea, because BOINC projects are regularly visited by spammers :-(


M4 Project homepage
M4 Project wiki
ID: 332 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Palo M.
Avatar

Send message
Joined: 17 Oct 07
Posts: 3
Credit: 1,310,805
RAC: 0
Message 333 - Posted: 2 Dec 2007, 5:35:31 UTC

Recently I tried to build enigma application myself on Linux (P4 Northwood, 2.4G, HTT on, Debian Sarge).
I also created 3 test work units - I noticed, that every time I run enigma (even the stock app), the results quite differ (probably feature of hill climbing?). So I better run 3 different workunits, to be able to exclude some non-typical times.

So, the times were:

1. For stock application (and similar if I compile enigma myself without any optimization options):
Test 1 run duration: 5:58.44 (User 341.03s, System 5.72s, 96% CPU)
Test 2 run duration: 13:17.85 (User 782.27s, System 3.87s, 98% CPU)
Test 3 run duration: 13:36.29 (User 794.91s, System 3.97s, 97% CPU)

2. For optimized application I built myself with gcc 3.3.5 (times are similar for the optimized P4 app downloaded here):
Test 1 run duration: 5:12.51 (User 302.31s, System 5.75s, 98% CPU)
Test 2 run duration: 12:07.52 (User 715.78s, System 3.51s, 98% CPU)
Test 3 run duration: 12:24.19 (User 728.66s, System 3.50s, 98% CPU)

3. Just for interest (inspired by Seti@Home), I downloaded Intel C++ compiler (which is free for personal use on Linux - thanks, Intel), the current version is 10.1.008. I turned some optimization flags for P4 (took the flags from Seti@Home optimized makefile) and it was...
Test 1 run duration: 4:51.36 (User 279.21s, System 6.00s, 97% CPU)
Test 2 run duration: 11:30.01 (User 677.09s, System 4.86s, 98% CPU)
Test 3 run duration: 11:29.19 (User 666.90s, System 8.03s, 97% CPU)

Just finished with test-units yesterday late night, but Intel-compiler results were really impressive. So I put this version into real usage by BOINC - so far 11 workunits (types 0, 1 and 2), all successful.


But, I have noticed one thing, which surprises me: the bench_cipher published in this thread is in fact the ciphertext2 with some extra stuff in it:
ciphertext2:
hceyztcsopuppzdiuqrdlwxxfacttjmbhdvcjjmmzrpyikhzawglyxwtmjpquefszbctvrla
bench_cipher:
hceyztcsopuppzdiuqrdlwxxfacttjmbhdvcjjmmzrpyikhzawglyxwtmjpquefszbctvrlalzxwvxtslfffaudqfbwrryapsbow

Is there some reason for that? Is the bench_cipher the full encrypted message, but it's not needed for BOINC and part of message (ciphertext2) is enough? Or bench_cipher has some stuffing for benchmarking purposes only?
ID: 333 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,762,956
RAC: 385,630
Message 338 - Posted: 2 Dec 2007, 15:15:44 UTC - in response to Message 333.  
Last modified: 2 Dec 2007, 15:24:42 UTC

Recently I tried to build enigma application myself on Linux (P4 Northwood, 2.4G, HTT on, Debian Sarge).
I also created 3 test work units - I noticed, that every time I run enigma (even the stock app), the results quite differ (probably feature of hill climbing?). So I better run 3 different workunits, to be able to exclude some non-typical times.


Also checkpointing and updating result file both have a minor effect on the runtime.


3. Just for interest (inspired by Seti@Home), I downloaded Intel C++ compiler (which is free for personal use on Linux - thanks, Intel), the current version is 10.1.008. I turned some optimization flags for P4 (took the flags from Seti@Home optimized makefile) and it was...
Test 1 run duration: 4:51.36 (User 279.21s, System 6.00s, 97% CPU)
Test 2 run duration: 11:30.01 (User 677.09s, System 4.86s, 98% CPU)
Test 3 run duration: 11:29.19 (User 666.90s, System 8.03s, 97% CPU)

Just finished with test-units yesterday late night, but Intel-compiler results were really impressive. So I put this version into real usage by BOINC - so far 11 workunits (types 0, 1 and 2), all successful.


Hm I've tried app built with Intel compiler on my PIII's some time ago, it was slightly slower than one build with a gcc 3.2/march=pentium3. I assumed that it won't be faster on any other CPUs, now I know that I was wrong :-)



But, I have noticed one thing, which surprises me: the bench_cipher published in this thread is in fact the ciphertext2 with some extra stuff in it:
[...]
Is there some reason for that? Is the bench_cipher the full encrypted message, but it's not needed for BOINC and part of message (ciphertext2) is enough? Or bench_cipher has some stuffing for benchmarking purposes only?


It's not the full ciphertext, but first half of it.

"Full ciphertext" wrote:

hceyztcsopuppzdiuqrdlwxxfacttjmbhdvcjjmmzrpyikhzawglyxwtmj
pquefszbctvrlalzxwvxtslfffaudqfbwrryapsbowjmklduyupfuqdowv
hahcdwauarswtxcfvoyfpufhvzfdggpoovgrmbpxxzcankmonfhxpckhjz
bumxjwxkauodyzucvcxpft


for benchmarking purposes, the content doesn't really matter, but the length of text is important.
The project is currently working on a part of full ciphertext - first 72 letters. This is a method of dealing with missing letters, I wrote about it in one of the other threads.


M4 Project homepage
M4 Project wiki
ID: 338 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 339 - Posted: 2 Dec 2007, 22:12:20 UTC - in response to Message 326.  
Last modified: 2 Dec 2007, 22:40:13 UTC

Is a boinc core client problem or an app problem?

It's a BOINC core client problem. 6.1.x is not to be used. Not when downloaded from Berkeley, not when compiled from source code. It just isn't ready yet.

Just do not use 6.1 clients, use the 5.10 branch of clients instead.

This is also a request from the BOINC developers. If you do run into trouble, we can't help you.
Jord.

BOINC FAQ Service.
ID: 339 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Eremita

Send message
Joined: 20 Sep 07
Posts: 2
Credit: 1,513
RAC: 0
Message 343 - Posted: 5 Dec 2007, 21:59:05 UTC

Ok, i understand.

Thanks a lot for your answer!!

Andrea
ID: 343 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
MJ

Send message
Joined: 17 Nov 07
Posts: 16
Credit: 95,844
RAC: 0
Message 347 - Posted: 10 Dec 2007, 7:09:11 UTC

I switched to using the Core 2 exe a couple of weeks ago. I don't really know how to know if it helped or not. But I haven't seen any errors, and since I have a core 2 quad I figure it can't have hurt right? I don't remember that I used the files from that first archive. Are those necessary?
I am running vista x64. Is that x64 build likely to be faster? I don't think my boinc supports x64. I tried to get that build to run a couple of days ago and I didn't have any luck.
Thanks!
ID: 347 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,762,956
RAC: 385,630
Message 348 - Posted: 10 Dec 2007, 7:32:58 UTC - in response to Message 347.  

I switched to using the Core 2 exe a couple of weeks ago. I don't really know how to know if it helped or not. But I haven't seen any errors, and since I have a core 2 quad I figure it can't have hurt right? I don't remember that I used the files from that first archive. Are those necessary?


Without app_info.xml BOINC will redownload any changed files, so probably you are still using standard app.


I am running vista x64. Is that x64 build likely to be faster? I don't think my boinc supports x64. I tried to get that build to run a couple of days ago and I didn't have any luck.
Thanks!


Probably it's a bit faster, but I don't have a Windows x64 machine to check.

M4 Project homepage
M4 Project wiki
ID: 348 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
MJ

Send message
Joined: 17 Nov 07
Posts: 16
Credit: 95,844
RAC: 0
Message 351 - Posted: 11 Dec 2007, 0:06:25 UTC

http://jjoshua2.googlepages.com/enigmaathome.net.md5

I posted md5 sums of the files in my project folder besides the ones I thought were just results and stuff and pics. I don't know if you can tell anything from that.
ID: 351 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
JAMC

Send message
Joined: 15 Dec 07
Posts: 3
Credit: 181,657
RAC: 0
Message 369 - Posted: 17 Dec 2007, 17:11:54 UTC

The optimized app is cutting ~22% off the stock times on my Q6600 quad ( running 50/50 with SETI)... good job and thanks!
ID: 369 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile tommy2x

Send message
Joined: 20 Dec 07
Posts: 3
Credit: 26,860
RAC: 0
Message 383 - Posted: 24 Dec 2007, 22:47:02 UTC

Works great.... and no problems.
ID: 383 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
ps69

Send message
Joined: 5 Jan 08
Posts: 1
Credit: 1,409
RAC: 0
Message 396 - Posted: 5 Jan 2008, 17:35:28 UTC

Which optimized app should I use for Celeron M 1,4GHz? Celeron D or FCPGA?
ID: 396 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,762,956
RAC: 385,630
Message 399 - Posted: 6 Jan 2008, 17:09:18 UTC - in response to Message 396.  

FCPGA I think, if it's a linux machine then exec build with Intel C++ compiler will be the fastest one.


M4 Project homepage
M4 Project wiki
ID: 399 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Annika

Send message
Joined: 16 Dec 07
Posts: 6
Credit: 42,355
RAC: 0
Message 400 - Posted: 6 Jan 2008, 17:28:37 UTC

What would you recommend for a Core Duo (Yonah)? Default or P3SSE2?
ID: 400 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Nicolas
Avatar

Send message
Joined: 10 Sep 07
Posts: 3
Credit: 5,176
RAC: 0
Message 411 - Posted: 10 Jan 2008, 18:27:43 UTC

AMD Athlon 64 X2 Dual Core 4200+, but running 32-bit Windows. What app should I run? Is the Athlon64 a 64-bit binary?

Also, please use zip instead of rar.
ID: 411 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 13 · Next

Message boards : Number crunching : v5.22 Experimental optimized app




Copyright © 2017 TJM