Posts by I7I7I7I7
log in
1) Message boards : News : Collatz v3.16 CUDA Testers for Windows Needed (Message 15648)
Posted 1856 days ago by I7I7I7I7
My GT 640 if finally with version 3.17 ( cuda 4.2) faster then 2.09 ( that version was fastest for me) but not for aprox 20% , then fro few % at most ( from 728 seconds to 698 seconds)
2) Message boards : Number crunching : Renamed brook64.dll in stock 2.09 (Message 14638)
Posted 1975 days ago by I7I7I7I7
When the 2.09 app is run, BOINC copies the files into the slot folder. When it does this, it is supposed to rename the dll.

Why, you ask, wouldn't it have the correct file name in the first place? Well, BOINC generates an MD5 hash and also checks the file size of each file. When it starts to run a WU, it first checks the hash and file size and if they don't match, BOINC assumes the incorrect file is on the system and is supposed to download the correct version.






But it doesnot do :( In my case no dll was even downloaded, so application fail to run and make few errors before i noticed that. After I rename dll, all is perfect. So rename doesnot work at least on Windows 7 x64.
And also, what is stock application, why you have few version ?
If you dont believe what I say: try yourself: restart project and look what heapen...
3) Message boards : Number crunching : Renamed brook64.dll in stock 2.09 (Message 14599)
Posted 1985 days ago by I7I7I7I7
When stock 2.09 app is downloaded there is renamed brook64.dll and with that dll app drops every WU as error. If renamed to brook64.dll all is ok
4) Message boards : News : CUDA 4.2 Version for Windows Now Available (Message 14436)
Posted 2015 days ago by I7I7I7I7
For me this new 4.0 or 4.2 app is far slower then with cuda 3.1
Drivers for Nvidia is latest, I have factory OC version of 560Ti
With 3.1 cuda mini collatz WU is finished with 240 seconds
with 4.0 or 4.2 the best I got is 353 seconds so it is way slower

Settings for cuda 3.1 is L13 I8 S0
settings for cuda 4.0 is K1024 I524288 S0
I also try K2048 I1048576 S1 ( 99% GPU) but still it is around 335 seconds
For support there is 2 CPU cores on 4 Ghz ( like in case of cuda 3.1)

Is someone have similar result like me?
5) Message boards : Windows : Wrong GPU freqency show in result (Message 13688)
Posted 2121 days ago by I7I7I7I7
Slicker ,thanks for reply, I just ask it because I sow wrong frequency.
6) Message boards : Windows : Wrong GPU freqency show in result (Message 13685)
Posted 2122 days ago by I7I7I7I7
Beside that info, my GPU works ok, so it is just cosmetic change, nothing else...
7) Message boards : Windows : Wrong GPU freqency show in result (Message 13681)
Posted 2122 days ago by I7I7I7I7
I have two ATI cards, one is on board , and one is 6850. When I disable on board Ati 4250 then frequency of my 6850 is shown correctly.But when is enabled, then it show same frequency as on board GPU...

CAL Runtime: 1.4.1703
Found 2 CAL devices
Device 0 is 6850 at 985 MHz ( not 560 MHz as shows)
Device 0: unknown ATI card 1024 MB local RAM (remote 1788 MB cached + 1788 MB uncached)
GPU core clock: 560 MHz, memory clock: 800 MHz
960 shader units organized in 12 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
not supporting double precision

Device 1: ATI Radeon HD2350/2400/3200/3400 (RV610/620) 128 MB local RAM (remote 1788 MB cached + 1788 MB uncached)
GPU core clock: 560 MHz, memory clock: 800 MHz
40 shader units organized in 1 SIMDs with 8 VLIW units (5-issue), wavefront size 32 threads
not supporting double precision

8) Message boards : Number crunching : Opt app for ati (Message 12540)
Posted 2390 days ago by I7I7I7I7
Last time I use opt app for ATI was before few months. And then I got nice performance boost on Radeon 5770.
Also I can remember that with setting f2 w0.2 screen response was very bad.
Now I use same settings on Radeon 5670. And have no screen rendering lag, and got in second same times as standard app.
So what I doing wrong? Any help?
Thanks

my app info

<app_info>
<app>
<name>mini_collatz</name>
</app>
<file_info>
<name>collatz64_cal_2.09_amd.exe</name>
<executable/>
</file_info>
<file_info>
<name>brook64.dll</name>
<executable/>
</file_info>

<app_version>
<app_name>mini_collatz</app_name>
<version_num>209</version_num>
<plan_class>ati</plan_class>
<avg_ncpus>0.02</avg_ncpus>
<max_ncpus>1</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline>w0.2 f2 p3</cmdline>
<file_ref>
<file_name>collatz64_cal_2.09_amd.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>brook64.dll</file_name>
</file_ref>
</app_version>

<app_version>
<app_name>mini_collatz</app_name>
<version_num>208</version_num>
<plan_class>ati</plan_class>
<avg_ncpus>0.02</avg_ncpus>
<max_ncpus>1</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline>w0.2 f2 p3</cmdline>
<file_ref>
<file_name>collatz64_cal_2.09_amd.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>brook64.dll</file_name>
</file_ref>
</app_version>
[/code]
9) Message boards : Number crunching : Difference from collatz and mini_colatz app? (Message 12539)
Posted 2390 days ago by I7I7I7I7
As I can see there is both , collatz and mini collatz in this app info. So I will get both wu version to crunch with this app info?
It that is not correct, then can some one write that app_info?
Thanks!
10) Message boards : Number crunching : Cuda 3.2 ili Cuda 4 app for Nvidia? (Message 12538)
Posted 2390 days ago by I7I7I7I7
Any new news about this topic?
OPEN CL maybe?
11) Message boards : Number crunching : Difference from collatz and mini_colatz app? (Message 12472)
Posted 2412 days ago by I7I7I7I7

The version number doesn't matter, i could have put in Multiple <app_version> sections to catch every possible combination of version numbers you might have been running, but that makes a big and complicated app_info,
with the app_info i posted, if you had used it unmodified you might have lost a few Wu's, but on the next scheduler request where you asked for work the server would have resent that work,
and updated your task page showing the work is being done with Anonymous platform now,

Claggy


Yes that was the case, I lost few WU but they was resent but only when I change version from 205 to 203.
When it was 205 server say there is no work unit for that application.
12) Message boards : Number crunching : Difference from collatz and mini_colatz app? (Message 12470)
Posted 2412 days ago by I7I7I7I7
From average 350 sec go to 250 sec: nice increase in speed
13) Message boards : Number crunching : Difference from collatz and mini_colatz app? (Message 12469)
Posted 2412 days ago by I7I7I7I7
I posted a suitable app_info in this post. There is no difference in the apps for collatz and mini_collatz.

Claggy


Only error in your app info was version: it is 203 for mini coolatz not 205 as you set :)
This is my app info, since I use only GPU for crunching ( if anybody need it)

SET TO max GPU speed


<app>
<name>collatz</name>
<user_friendly_name>collatz</user_friendly_name>
</app>
<file_info>
<name>collatz_2.05_windows_intelx86__cuda31.exe</name>
<executable/>
</file_info>
<file_info>
<name>cudart32_31_9.dll</name>
<executable/>
</file_info>
<app_version>
<app_name>collatz</app_name>
<version_num>203</version_num>
<plan_class>cuda</plan_class>
<avg_ncpus>0.013</avg_ncpus>
<max_ncpus>0.13</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<cmdline>L13 I8 S0</cmdline>
<file_ref>
<file_name>collatz_2.05_windows_intelx86__cuda31.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart32_31_9.dll</file_name>
</file_ref>
</app_version>
<app>
<name>mini_collatz</name>
<user_friendly_name>mini_collatz</user_friendly_name>
</app>
<file_info>
<name>collatz_2.05_windows_intelx86__cuda31.exe</name>
<executable/>
</file_info>
<file_info>
<name>cudart32_31_9.dll</name>
<executable/>
</file_info>
<app_version>
<app_name>mini_collatz</app_name>
<version_num>203</version_num>
<plan_class>cuda</plan_class>
<avg_ncpus>0.013</avg_ncpus>
<max_ncpus>0.13</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>
<cmdline>L13 I8 S0</cmdline>
<file_ref>
<file_name>collatz_2.05_windows_intelx86__cuda31.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart32_31_9.dll</file_name>
</file_ref>
</app_version>
</app_info>
14) Message boards : Number crunching : Difference from collatz and mini_colatz app? (Message 12465)
Posted 2413 days ago by I7I7I7I7
Can I use ( with app.info) app for collatz to do mini_colatz WU? Since coolatz with cuda 3.1 is faster then stock ( cuda 2.3) I would like to use it.
Any special app info?
Thanks.

P.S I can always rename app but in that case I cannot use switches, and with switches I got 15% more speed :(
15) Message boards : Number crunching : Cuda 3.2 ili Cuda 4 app for Nvidia? (Message 12369)
Posted 2431 days ago by I7I7I7I7
Is any information about topic title is available?
Will those app be released and when?
16) Message boards : Number crunching : Linux ( ubuntu 10.10 64 bit) twice slower then Win (Message 12224)
Posted 2460 days ago by I7I7I7I7
I will try it when you released this app: 20% slower is much better then 200% :)
Tnx alot :)
17) Message boards : Number crunching : Linux ( ubuntu 10.10 64 bit) twice slower then Win (Message 12005)
Posted 2475 days ago by I7I7I7I7
You will be reborn, when you do that. And get 2x RAC also :) So dont waste energy, and time!
18) Message boards : Number crunching : Linux ( ubuntu 10.10 64 bit) twice slower then Win (Message 11962)
Posted 2477 days ago by I7I7I7I7
For now using 64 bit Linux app is waste of energy and time :(
19) Message boards : Number crunching : Linux ( ubuntu 10.10 64 bit) twice slower then Win (Message 11938)
Posted 2479 days ago by I7I7I7I7
I look at others and found that on Linux 64 bit crunch time is similar as mine, about 2 times slower on Windows. I also install 32 bit Linux, but as others I got error after 2 seconds ( and found that error is documented, but no fix for that)
20) Message boards : Number crunching : Linux ( ubuntu 10.10 64 bit) twice slower then Win (Message 11930)
Posted 2480 days ago by I7I7I7I7
collatz_2.02_i686-pc-linux-gnu__cuda22.tar.bz2 (Linux 32-bit CUDA) 2.02 nVidia/CUDA 2.2 11/29/2009 12:12 CST
collatz_2.04_x86_64-pc-linux-gnu__cuda31.tar.bz2 (Linux 64-bit CUDA) 2.04 nVidia/CUDA 3.1 07/29/2010 13:11 CDT
collatz_2.02_x86_64-pc-linux-gnu__cuda22.tar.bz2 (Linux 64-bit CUDA) 2.02 nVidia/CUDA 2.2 11/29/2009 12:12 CST
collatz_2.01_x86_64-pc-linux-gnu__ati14_r1.5_mgpu.tar.bz2 (Linux 64-bit ATI) 2.01 (release 1.5) ATI CAL 1.4 02/26/2010 23:50 UTC
collatz_2.09_i686-pc-linux-gnu__ati14.tar.bz2 (Linux 32-bit ATI - EXPERIMENTAL!!!) 2.09 ATI CAL 1.4 01/17/2010 21:32 CST

There is no 2.2 for ATI GPU on linux 64 bit :(

Since you wrote Yes, the linux version is a little slower, but not twice as slow. you must have some kind of prof?


Next 20

Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.