Driver 344.11

Message boards : Number crunching : Driver 344.11
Message board moderation

To post messages, you must log in.

AuthorMessage
Zule

Send message
Joined: 1 Jul 06
Posts: 52
Credit: 84,436,096
RAC: 0
United States
Message 1588337 - Posted: 17 Oct 2014, 16:12:48 UTC
Last modified: 17 Oct 2014, 16:13:24 UTC

Checking out my AP 7 tasks I noticed I'm finishing most in the 2000-3400 second range which seems pretty good. I'm some how beating out faster video cards with my 750TI and driver 337.88 but my CPU usage is much higher.

http://setiathome.berkeley.edu/workunit.php?wuid=1617072555

Driver 344.11 seems common on the machines I'm beating.. Anyone actually tested that driver on AP v7 vs other drivers? Might be nothing, I am ready bed and might be looking at stuff wrong:)

Any thoughts on my CPU usage? In my app_config I'm running .50 GPU and 1 CPU. The GPU's seem happy, I run quad 750TI and all stay in the 95-99% usage range.. Often hanging steady at 99..
ID: 1588337 · Report as offensive
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1588343 - Posted: 17 Oct 2014, 16:21:33 UTC - in response to Message 1588337.  
Last modified: 17 Oct 2014, 16:25:46 UTC

You could look at using a commandline to decrease your CPU usage. Might increase the time to complete by 1-2 minutes. I use the following for my 750s

-use_sleep -unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -tune 1 64 4 1 -hp

You know where to install this at? Just asking, some people are don't use them but I've seen it help with the CPU utilization.


Zalster

Edit.. LOL That's my machine you partnered with. It only seems like yours is faster. I'm running 3 APs at a time on each of my cards so that is why the time looks longer. If I only ran 1 or 2 AP per card then the times would be closer.
ID: 1588343 · Report as offensive
Zule

Send message
Joined: 1 Jul 06
Posts: 52
Credit: 84,436,096
RAC: 0
United States
Message 1588345 - Posted: 17 Oct 2014, 16:27:19 UTC

Nice machine:)

Planning to play with some settings next time I'm off.. Been using a conservative -unroll 4 -ffa_block 2048 -ffa_block_fetch 1024 which seems to work fine. Want to try and do some testing with -cpu_lock and learn more about -use_sleep and -tune.. Someone really needs to write a nice seti for dummies guide:)

Also want to test against different vid drivers.. Long ago I ran a GTX260 and found a nice boost using the old 266 drivers vs anything else.. Maybe I can find another sweet spot for the 750's..
ID: 1588345 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1588346 - Posted: 17 Oct 2014, 16:29:21 UTC

The file is called ap_cmdline_win_x86_SSE2_OpenCL_NV.txt.

add this


-use_sleep -unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -tune 1 64 4 1 -tune 2 64 4 1.

Tuning also kernel 2 makes more sense with this app.


With each crime and every kindness we birth our future.
ID: 1588346 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1588347 - Posted: 17 Oct 2014, 16:30:08 UTC - in response to Message 1588345.  

Nice machine:)

Planning to play with some settings next time I'm off.. Been using a conservative -unroll 4 -ffa_block 2048 -ffa_block_fetch 1024 which seems to work fine. Want to try and do some testing with -cpu_lock and learn more about -use_sleep and -tune.. Someone really needs to write a nice seti for dummies guide:)

Also want to test against different vid drivers.. Long ago I ran a GTX260 and found a nice boost using the old 266 drivers vs anything else.. Maybe I can find another sweet spot for the 750's..


I did that already.
Just read the readme.


With each crime and every kindness we birth our future.
ID: 1588347 · Report as offensive
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1588350 - Posted: 17 Oct 2014, 16:34:44 UTC - in response to Message 1588346.  

The file is called ap_cmdline_win_x86_SSE2_OpenCL_NV.txt.

add this


-use_sleep -unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -tune 1 64 4 1 -tune 2 64 4 1.

Tuning also kernel 2 makes more sense with this app.



So Mike, I should add this section then to my current Crunchers? That's new.

Zule, Mike is the Guru to end all Guru on this. Trust what he says
ID: 1588350 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1588354 - Posted: 17 Oct 2014, 16:36:48 UTC - in response to Message 1588350.  

The file is called ap_cmdline_win_x86_SSE2_OpenCL_NV.txt.

add this


-use_sleep -unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -tune 1 64 4 1 -tune 2 64 4 1.

Tuning also kernel 2 makes more sense with this app.



So Mike, I should add this section then to my current Crunchers? That's new.

Zule, Mike is the Guru to end all Guru on this. Trust what he says


Yes.

If you really want to speed up processing use -oclfft_plan as well.


With each crime and every kindness we birth our future.
ID: 1588354 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1588365 - Posted: 17 Oct 2014, 17:07:24 UTC

Dont use any other values than i suggested in the read me for your card.

I would prefer to continue this in the AP V7 thread.
So everybody can follow this.

If any question arises please ask.


With each crime and every kindness we birth our future.
ID: 1588365 · Report as offensive
David S
Volunteer tester
Avatar

Send message
Joined: 4 Oct 99
Posts: 18352
Credit: 27,761,924
RAC: 12
United States
Message 1588399 - Posted: 17 Oct 2014, 18:47:43 UTC - in response to Message 1588347.  

Nice machine:)

Planning to play with some settings next time I'm off.. Been using a conservative -unroll 4 -ffa_block 2048 -ffa_block_fetch 1024 which seems to work fine. Want to try and do some testing with -cpu_lock and learn more about -use_sleep and -tune.. Someone really needs to write a nice seti for dummies guide:)

Also want to test against different vid drivers.. Long ago I ran a GTX260 and found a nice boost using the old 266 drivers vs anything else.. Maybe I can find another sweet spot for the 750's..


I did that already.
Just read the readme.

Maybe I'm missing something, but where's the readme?
David
Sitting on my butt while others boldly go,
Waiting for a message from a small furry creature from Alpha Centauri.

ID: 1588399 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1588400 - Posted: 17 Oct 2014, 18:49:56 UTC - in response to Message 1588399.  
Last modified: 17 Oct 2014, 18:51:08 UTC

Nice machine:)

Planning to play with some settings next time I'm off.. Been using a conservative -unroll 4 -ffa_block 2048 -ffa_block_fetch 1024 which seems to work fine. Want to try and do some testing with -cpu_lock and learn more about -use_sleep and -tune.. Someone really needs to write a nice seti for dummies guide:)

Also want to test against different vid drivers.. Long ago I ran a GTX260 and found a nice boost using the old 266 drivers vs anything else.. Maybe I can find another sweet spot for the 750's..


I did that already.
Just read the readme.

Maybe I'm missing something, but where's the readme?

In the project directory, in the folder called 'docs'

Claggy
ID: 1588400 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1588629 - Posted: 18 Oct 2014, 8:27:31 UTC - in response to Message 1588346.  

add this

-use_sleep -unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -tune 1 64 4 1 -tune 2 64 4 1.

I wonder what will happen if the user copy the line as it is - with the dot at the end.
Will this be accepted by the app or ... all tasks will fail?
 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1588629 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1588631 - Posted: 18 Oct 2014, 8:41:46 UTC - in response to Message 1588629.  

add this

-use_sleep -unroll 12 -ffa_block 8192 -ffa_block_fetch 4096 -tune 1 64 4 1 -tune 2 64 4 1.

I wonder what will happen if the user copy the line as it is - with the dot at the end.
Will this be accepted by the app or ... all tasks will fail?


It will be accepted.


With each crime and every kindness we birth our future.
ID: 1588631 · Report as offensive

Message boards : Number crunching : Driver 344.11


 
©2024 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.