[phpBB Debug] PHP Warning: in file [ROOT]/phpbb/session.php on line 580: sizeof(): Parameter must be an array or an object that implements Countable
[phpBB Debug] PHP Warning: in file [ROOT]/phpbb/session.php on line 636: sizeof(): Parameter must be an array or an object that implements Countable
3.14.by forum • New brook beta
Page 1 of 1

New brook beta

Posted: Fri Apr 03, 2009 11:25 am
by BarsMonster
http://3.14.by/files/brook_test_32.zip

Should be able to restore state. gpu_time works. SHould decide better on task size = higher speed sometimes, should not hang on slow cards.
Sorry no x64 version, I am quite busy.

Re: New brook beta

Posted: Fri Apr 03, 2009 11:25 am
by BarsMonster
This is not yet a SLI one.

Re: New brook beta

Posted: Fri Apr 03, 2009 12:04 pm
by shifter1
Sweet! Ill test as soon as I get home. :mrgreen: :mrgreen:

Re: New brook beta

Posted: Sat Apr 04, 2009 5:37 am
by asl
It's slower than the old version

Image

Re: New brook beta

Posted: Sun Apr 05, 2009 7:26 pm
by BarsMonster
could you send me log file?
could you try -gpu_time 1500?

Re: New brook beta

Posted: Mon Apr 06, 2009 9:09 pm
by asl
Now it's faster. :D
x.PNG
x.PNG (13.73 KiB) Viewed 18952 times
log with -gpu_time 1500:
http://rafb.net/p/lh8dzf37.html


log:
http://rafb.net/p/Hd49m290.html

Re: New brook beta

Posted: Fri May 29, 2009 9:21 pm
by c0sm4ky
hi, I tried this new beta but crashes :(
I'm using latest ati drivers 9.5. and brook+/cal 1.4 (i had also had to rename atical*.dll to amdcal*.dll)

here is a screenshot
Image


Also, if i start it via commandline (without arguments) (cmd.exe), it crashes here:
Image



but if i start it by double click on it, it does NOT crashes! very strange behaviour.

btw best compliments for your work! ;)

Re: New brook beta

Posted: Sat May 30, 2009 5:57 pm
by ash
...maybe users DNA error.... :lol: :lol: :lol:
use 8.12 driver version.




1. it crashes when it runs from command prompt and when it's double clicked-the same-it got no precheck on input arguments(i guess);
2.why it crashes:
first of all olly 1.10 doesn't support sse2 only sse anx mmx but command prefix:66-it means it's wrong disassm code:it operates on xmm registers
so command- movdqa [ecx+10],xmm0;movdqa unlike movdqu needs aligned memory access arguments but in ecx is 622d98 in general memory source gets
622dd8 and its no aligned .....so we all got... and the winner is GP(0) :joy: :joy: :joy:
... so dear don't play with that, leave it for ... 8-) 8-) 8-) others
michail, algo: in yours IL: it got a lot of different constants :? , not sin constants, not initialization constants...???...maybe stupid brook, but it's interesting,
moreover i'd tried yours stuff on R600(2600 pro)- speed:~50 mh/s; so it works :joy: :joy: :joy:

Re: New brook beta

Posted: Sun May 31, 2009 6:00 am
by BarsMonster
ash wrote:michail, algo: in yours IL: it got a lot of different constants :? , not sin constants, not initialization constants...???...maybe stupid brook, but it's interesting,
moreover i'd tried yours stuff on R600(2600 pro)- speed:~50 mh/s; so it works :joy: :joy: :joy:
Great, first positive results from HD2xxx :crazy:
These constants are already modified a bit to make it faster.

Re: New brook beta

Posted: Tue Feb 15, 2011 10:24 pm
by aRcTiC
Sorry for bumping old thread but since i couldnt get oclhashcat to work...
...I did however got my 3850 agp to work with barswf brook by doing what c0sm4ky (renaming atical* to amdcal*) did and also with the newest drivers not 8.12 :D

Image

Re: New brook beta

Posted: Fri May 20, 2011 3:49 pm
by Retired_raver
i can not get my 2 4850 cards to work with 0.9b version work.....it does not find passes...i tried simple like 1234 and it goes to more digits and continious working...any idea?

Re: New brook beta

Posted: Thu May 26, 2011 9:00 pm
by BarsMonster
Retired_raver wrote:i can not get my 2 4850 cards to work with 0.9b version work.....it does not find passes...i tried simple like 1234 and it goes to more digits and continious working...any idea?
Must be driver issues, and as brook is abandoned for quite a while this problem will rest in piece...