64 bit BOINC client & BOINC MANAGER for Windows XP64

Message boards : Number crunching : 64 bit BOINC client & BOINC MANAGER for Windows XP64
Message board moderation

To post messages, you must log in.

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

AuthorMessage
Profile Benher
Volunteer developer
Volunteer tester

Send message
Joined: 25 Jul 99
Posts: 517
Credit: 465,152
RAC: 0
United States
Message 440803 - Posted: 20 Oct 2006, 21:19:46 UTC

Crunch3r,

Regarding the benchmark speeds...I think you are correct that even when using 64 bit direct non WOW code the results are too high for the test you showed.

Back in the day the Whetstone and Dhrystone benchmarks were being really over-optimized by the compiler...such that whole loops were never being converted to code...they were just thrown out by the compiler. Benchmarks by definition just do busy work for a while and then never use the computed values from their testing in later code...just the calculated start and end times. As such the compilers saw this and never created code for many loops.

I changed Whetstone (the FP benchmark) to try to fool the compiler into producing all the code specified in the source (using extern variables, and saving results there). Eric incorporated these changes into boinc. But with whole program optimization and such, perhaps the compiler needs to be fooled again into doing what we tell it to do.

Contrast the assembly output of compile of Whetstone,c, and Dhrystone.c source and figure out what is happening. Note: I never modified Integer Bench (Dhrystone.c).
ID: 440803 · Report as offensive
Profile enzed
Avatar

Send message
Joined: 27 Mar 05
Posts: 347
Credit: 1,681,694
RAC: 0
New Zealand
Message 440833 - Posted: 20 Oct 2006, 21:42:36 UTC - in response to Message 430067.  

as the message says... that client will have some basic features like return_results_immediately...

Note:
It'll be a stock client without any optimizations turned on. (like cpu affinity, double_num_cpus, reset_debs etc etc etc...)

you can decide here:
POLL

feel free to leave a comment.


Good to see you are back Crunch3r, this is the sort of thing that science needs.... people who are willing to tinker/explore
ID: 440833 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 440837 - Posted: 20 Oct 2006, 21:46:15 UTC - in response to Message 440803.  

Crunch3r,

Regarding the benchmark speeds...I think you are correct that even when using 64 bit direct non WOW code the results are too high for the test you showed.

Back in the day the Whetstone and Dhrystone benchmarks were being really over-optimized by the compiler...such that whole loops were never being converted to code...they were just thrown out by the compiler. Benchmarks by definition just do busy work for a while and then never use the computed values from their testing in later code...just the calculated start and end times. As such the compilers saw this and never created code for many loops.

I changed Whetstone (the FP benchmark) to try to fool the compiler into producing all the code specified in the source (using extern variables, and saving results there). Eric incorporated these changes into boinc. But with whole program optimization and such, perhaps the compiler needs to be fooled again into doing what we tell it to do.

Contrast the assembly output of compile of Whetstone,c, and Dhrystone.c source and figure out what is happening. Note: I never modified Integer Bench (Dhrystone.c).



Hi Ben ,
well i did not use Whole Program Optimisation and none optimization at all.. just set it to x64 and enabled _fastcall (done automatically if not set at all cuz win64 only uses fastall) and set structure allignment to 16... i didn't even use the INTEL compiler at all cuz it would make it even worse...

Simply compiledit using VS2005 that's it ... i myself have some strange results regarding benchmarks on my P4D, so that why i asked of the code edit to the benchmark files...

IMHO... while using flop counting now on most projects ( even if it couts wrong)

The benchmarks could be set to a fixed value ... i tried that one usinf 1000 for int an 2000 for FP... or even disabled running after the first time of starting boinc ....

IMHO this is useles as well....







Join BOINC United now!
ID: 440837 · Report as offensive
Pepperammi

Send message
Joined: 3 Apr 99
Posts: 200
Credit: 737,775
RAC: 0
United Kingdom
Message 440883 - Posted: 20 Oct 2006, 22:48:38 UTC

Though you might be interested to know the the memory usage has changed now after its been going for a while.

boinc 10,400
boincmgr 6,870
ID: 440883 · Report as offensive

Send message
Joined: 3 Sep 02
Posts: 396
Credit: 5,293
RAC: 0
Message 441862 - Posted: 22 Oct 2006, 16:23:35 UTC

talk about trigger words, that nanoseconds just reminded me of my earlly years in denmark, i used to laugh at some guy called dork; "dork calling olson come here olson, nano nano" well i was about 4 years old then, so maybe i got the nano nano wrong? but thats what it reminded me off :)
I AM NOT FAT! I AM BIG-BONED!
ID: 441862 · Report as offensive
Profile hiamps
Volunteer tester
Avatar

Send message
Joined: 23 May 99
Posts: 4292
Credit: 72,971,319
RAC: 0
United States
Message 441907 - Posted: 22 Oct 2006, 17:10:32 UTC - in response to Message 441862.  

talk about trigger words, that nanoseconds just reminded me of my earlly years in denmark, i used to laugh at some guy called dork; "dork calling olson come here olson, nano nano" well i was about 4 years old then, so maybe i got the nano nano wrong? but thats what it reminded me off :)


I bet is was "Mork" as in Mork and Mindy...
Official Abuser of Boinc Buttons...
And no good credit hound!
ID: 441907 · Report as offensive
Profile Andy Lee Robinson
Avatar

Send message
Joined: 8 Dec 05
Posts: 630
Credit: 59,973,836
RAC: 0
Hungary
Message 442316 - Posted: 23 Oct 2006, 1:33:47 UTC - in response to Message 441985.  

It was Mork.


calling Orson... :)
ID: 442316 · Report as offensive
Profile Keck_Komputers
Volunteer tester
Avatar

Send message
Joined: 4 Jul 99
Posts: 1575
Credit: 4,152,111
RAC: 1
United States
Message 442343 - Posted: 23 Oct 2006, 2:41:25 UTC - in response to Message 442316.  

It was Mork.


calling Orson... :)

Nanu nanu
BOINC WIKI

BOINCing since 2002/12/8
ID: 442343 · Report as offensive
Profile Gecko
Volunteer tester
Avatar

Send message
Joined: 17 Nov 99
Posts: 454
Credit: 6,946,910
RAC: 47
United States
Message 442367 - Posted: 23 Oct 2006, 4:46:38 UTC

Shaazbot!
ID: 442367 · Report as offensive
Profile Michael
Avatar

Send message
Joined: 12 Apr 02
Posts: 10
Credit: 4,217
RAC: 0
China
Message 442487 - Posted: 23 Oct 2006, 14:41:47 UTC

maybe...but at first, i must get enough money to buy a 64bit computer and wait for Microsoft Windows Vista.
okay,okay, maybe that is just a dream. but i will make it to be my target and try my best to do it.
that is a long way to go...
at that time, i think i need a 64bit BOINC client and BOINC MANAGER.
haha...
Sir Michael Enista
Michael is here, and Michael is everywhere...
ID: 442487 · Report as offensive
Teamriot (TEAM USA)

Send message
Joined: 13 Oct 06
Posts: 1
Credit: 1,808,518
RAC: 0
United States
Message 442622 - Posted: 23 Oct 2006, 20:24:47 UTC

Yes, I would like to see it move into the future where 64bit technology is heading, 32bit is going to be the past in the next few years.
ID: 442622 · Report as offensive
ebahapo
Volunteer tester
Avatar

Send message
Joined: 2 Jun 99
Posts: 31
Credit: 452,292
RAC: 0
Message 443104 - Posted: 24 Oct 2006, 20:29:33 UTC

This client will request the 32-bit application. But given that any Windows x64 can run either 32 or 64-bit applications, it would be a good idea if the client would try to fetch a 64-bit application first (windows_amd64 platform) and if it fails then it would try to get the 32-bit one (windows_intelx86 platform).

I proposed this here before, but it fell on deaf ears...
ID: 443104 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 464212 - Posted: 23 Nov 2006, 16:19:27 UTC - in response to Message 443857.  
Last modified: 23 Nov 2006, 16:19:44 UTC


Hello Folks,

there's an update for the BOINC clients available my page.

Versions:
5.7.5.32 32 bit Windows
5.7.5.64 64 bit Windows

features:
cpu affinity & return resuls immediately





Join BOINC United now!
ID: 464212 · Report as offensive
Profile Geek@Play
Volunteer tester
Avatar

Send message
Joined: 31 Jul 01
Posts: 2467
Credit: 86,146,931
RAC: 0
United States
Message 464297 - Posted: 23 Nov 2006, 18:35:07 UTC - in response to Message 464212.  


Hello Folks,

there's an update for the BOINC clients available my page.

Versions:
5.7.5.32 32 bit Windows
5.7.5.64 64 bit Windows

features:
cpu affinity & return resuls immediately





Hi Crunch.........
I downloaded your 32 bit application. Installed ok on 1 computer but not on my other computers. When I tried to start the service again a window popped up saying something about "improper version". Did uninstall of of Boinc, installed Boinc 5.7.5, copied in your files again, same result.

Don't know why but you may need to help, I did the same thing as on the computer that was successful!

Oh by the way.........thanks a lot for your work and time!


Boinc....Boinc....Boinc....Boinc....
ID: 464297 · Report as offensive
John McLeod VII
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jul 99
Posts: 24806
Credit: 790,712
RAC: 0
United States
Message 464347 - Posted: 23 Nov 2006, 19:48:48 UTC

return_results_immediately was meant for debugging purposes only. Since it was used for other than that, it was removed from the stock client. It does add load to the servers as a large portion of the time is spent per connection to the scheduler rather than per result.


BOINC WIKI
ID: 464347 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 21921
Credit: 7,508,002
RAC: 20
United Kingdom
Message 464401 - Posted: 23 Nov 2006, 20:56:46 UTC - in response to Message 464347.  
Last modified: 23 Nov 2006, 20:58:14 UTC

return_results_immediately was meant for debugging purposes only. Since it was used for other than that, it was removed from the stock client. It does add load to the servers as a large portion of the time is spent per connection to the scheduler rather than per result.

So why would someone want to always return results immediately?

We've had it explained a few times how it hammers the servers. Rom has put up a good example explaining how and why and even by how much.

I guess if the feature gets abused enough, then there'll be the distraction of some more server-side code to restrict the few "over-eager" renegades...


So why have a permanent "return results immediately"?
What am I missing with the stock boinc manager?

Happy crunchin',
Martin
See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 464401 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 464465 - Posted: 23 Nov 2006, 22:34:47 UTC - in response to Message 464231.  


Hello Folks,

there's an update for the BOINC clients available my page.

Versions:
5.7.5.32 32 bit Windows
5.7.5.64 64 bit Windows

features:
cpu affinity & return resuls immediately





Hey Crunch3r, I'm having a problem with Your 64bit Windows XP x64 Client and DeNitro's XP x64 64bit apps, On startup of Boinc and when doing a benchmark(Not all the time though on either one) and regardless of My cpu speed(Stock speed is 2.0GHz and overclocked with Ntune is 2.3GHz, Bios provides voltage support only) I get this:

Note: I do not get any problems when I use a 32bit app of course, Just with 64bit apps for some reason. And this also happened with Your original 5.7.64 client too. DeNitro has said He can't find a problem with the app and since the Benchmark is done by the client, I'm thinking something in Your clients doesn't quite like 64bit XP apps.

11/23/2006 8:40:53 AM||Starting BOINC client version 5.7.5 for windows_intelx86
11/23/2006 8:40:53 AM||log flags: task, file_xfer, sched_ops
11/23/2006 8:40:53 AM||Libraries: libcurl/7.15.5
11/23/2006 8:40:53 AM||Data directory: C:\\Program Files (x86)\\BOINC
11/23/2006 8:40:53 AM|SETI@home|Found app_info.xml; using anonymous platform
11/23/2006 8:40:53 AM||Processor: 4 AuthenticAMD Dual Core AMD Opteron(tm) Processor 270
11/23/2006 8:40:53 AM||Memory: 2.00 GB physical, 3.87 GB virtual
11/23/2006 8:40:53 AM||Disk: 74.52 GB total, 7.44 GB free
11/23/2006 8:40:53 AM||Version change (5.7.64 -> 5.7.5); running CPU benchmarks
11/23/2006 8:40:53 AM|Einstein@Home|URL: http://einstein.phys.uwm.edu/; Computer ID: 732308; location: home; project prefs: default
11/23/2006 8:40:53 AM|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: 2483868; location: home; project prefs: default
11/23/2006 8:40:53 AM||General prefs: from SETI@home (last modified 2006-11-20 20:10:33)
11/23/2006 8:40:53 AM||Host location: home
11/23/2006 8:40:53 AM||General prefs: no separate prefs for home; using your defaults
11/23/2006 8:40:53 AM||
11/23/2006 8:40:53 AM||
11/23/2006 8:40:53 AM||BOINC 5.7.5.64 - 64 bit Edition by Crunch3r
11/23/2006 8:40:53 AM||enabled features:
11/23/2006 8:40:53 AM||-cpu_affinity
11/23/2006 8:40:53 AM||-return_results_immediately
11/23/2006 8:40:53 AM||
11/23/2006 8:40:53 AM||
11/23/2006 8:40:55 AM||Running CPU benchmarks
11/23/2006 8:41:54 AM||Benchmark results:
11/23/2006 8:41:54 AM|| Number of CPUs: 4
11/23/2006 8:41:54 AM|| 2145 floating point MIPS (Whetstone) per CPU
11/23/2006 8:41:54 AM|| 5535 integer MIPS (Dhrystone) per CPU
11/23/2006 8:41:55 AM|SETI@home|Restarting task 16jn03ab.14213.23793.761084.3.24_3 using setiathome_enhanced version 515
11/23/2006 8:41:55 AM|SETI@home|Restarting task 18jn03aa.23749.14912.659648.3.69_0 using setiathome_enhanced version 515
11/23/2006 8:41:55 AM|SETI@home|Restarting task 06jn03aa.25044.2416.547146.3.14_0 using setiathome_enhanced version 515
11/23/2006 8:41:55 AM|SETI@home|Restarting task 06jn03aa.25044.2464.209640.3.50_2 using setiathome_enhanced version 515
11/23/2006 8:42:04 AM|SETI@home|Unrecoverable error for result 18jn03aa.23749.14912.659648.3.69_0 ( - exit code -1073741811 (0xc000000d))
11/23/2006 8:42:04 AM|SETI@home|Deferring scheduler requests for 1 minutes and 0 seconds
11/23/2006 8:42:04 AM|SETI@home|Computation for task 18jn03aa.23749.14912.659648.3.69_0 finished
11/23/2006 8:42:04 AM||Starting 06jn03aa.25044.2705.636084.3.85_2
11/23/2006 8:42:04 AM|SETI@home|Starting task 06jn03aa.25044.2705.636084.3.85_2 using setiathome_enhanced version 515
11/23/2006 8:42:06 AM|SETI@home|Started upload of file 18jn03aa.23749.14912.659648.3.69_0_0
11/23/2006 8:42:08 AM|SETI@home|Finished upload of file 18jn03aa.23749.14912.659648.3.69_0_0
11/23/2006 8:42:08 AM|SETI@home|Throughput 43110 bytes/sec
11/23/2006 8:42:27 AM||Suspending computation - running CPU benchmarks
11/23/2006 8:42:29 AM||Running CPU benchmarks
11/23/2006 8:43:06 AM|SETI@home|Sending scheduler request: To report completed tasks
11/23/2006 8:43:06 AM|SETI@home|Reporting 1 tasks
11/23/2006 8:43:11 AM|SETI@home|Scheduler RPC succeeded [server version 507]
11/23/2006 8:43:11 AM|SETI@home|Deferring scheduler requests for 11 seconds
11/23/2006 8:43:28 AM||Benchmark results:
11/23/2006 8:43:28 AM|| Number of CPUs: 4
11/23/2006 8:43:28 AM|| 2144 floating point MIPS (Whetstone) per CPU
11/23/2006 8:43:28 AM|| 5520 integer MIPS (Dhrystone) per CPU
11/23/2006 8:43:29 AM||Resuming computation
11/23/2006 8:43:53 AM|SETI@home|Sending scheduler request: Requested by user
11/23/2006 8:43:53 AM|SETI@home|(not requesting new work or reporting completed tasks)
11/23/2006 8:43:58 AM|SETI@home|Scheduler RPC succeeded [server version 507]
11/23/2006 8:43:58 AM|SETI@home|Deferring scheduler requests for 11 seconds


Hi Batman or Joker ; )

the stuff or error your refering to is a ( MY GUESS !!) function call from BONC to the windows API.... it's not replated to the seti app.... it's somthing that has popped up on most projects like malaria,chess960, rosetta etc...

I guess it's a problem in theboinc souce itself that causes these issues. I myself have never experienced those problems at all ... but google yourself ...


My best bet is that it's a bad function call from boinc on the windows api that causes that apps to crash ... not my issue..ask Rom Walton ... he develops 95% of BOINC ..he's the one you should ask...





Join BOINC United now!
ID: 464465 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 464468 - Posted: 23 Nov 2006, 22:36:58 UTC - in response to Message 464347.  

return_results_immediately was meant for debugging purposes only. Since it was used for other than that, it was removed from the stock client. It does add load to the servers as a large portion of the time is spent per connection to the scheduler rather than per result.


Sorry John,

i do debugging... so it's i can use it on my own ... no one is forced to use third party clients & apps...

If you don't like it don't use it ... sorry






Join BOINC United now!
ID: 464468 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 464471 - Posted: 23 Nov 2006, 22:40:32 UTC - in response to Message 464401.  
Last modified: 23 Nov 2006, 22:41:18 UTC

return_results_immediately was meant for debugging purposes only. Since it was used for other than that, it was removed from the stock client. It does add load to the servers as a large portion of the time is spent per connection to the scheduler rather than per result.

So why would someone want to always return results immediately?

We've had it explained a few times how it hammers the servers. Rom has put up a good example explaining how and why and even by how much.

I guess if the feature gets abused enough, then there'll be the distraction of some more server-side code to restrict the few "over-eager" renegades...


So why have a permanent "return results immediately"?
What am I missing with the stock boinc manager?

Happy crunchin',



Why one what's to use return rtesults immediately....

bee discussed before .. it's for debuging ...and if your in a "unstable" region regarding power supply... well... imgagine the rest to yourself ... and saving previous client_state.xml" does still not work properly...

if you don't like it ...don't use it ...



Join BOINC United now!
ID: 464471 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 464476 - Posted: 23 Nov 2006, 22:47:20 UTC - in response to Message 464297.  


Hello Folks,

there's an update for the BOINC clients available my page.

Versions:
5.7.5.32 32 bit Windows
5.7.5.64 64 bit Windows

features:
cpu affinity & return resuls immediately





Hi Crunch.........
I downloaded your 32 bit application. Installed ok on 1 computer but not on my other computers. When I tried to start the service again a window popped up saying something about "improper version". Did uninstall of of Boinc, installed Boinc 5.7.5, copied in your files again, same result.

Don't know why but you may need to help, I did the same thing as on the computer that was successful!

Oh by the way.........thanks a lot for your work and time!



Hey Geek,

mail me you got my mail adress we'll solve that issue together ;)




Join BOINC United now!
ID: 464476 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · 6 · Next

Message boards : Number crunching : 64 bit BOINC client & BOINC MANAGER for Windows XP64


 
©2025 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.