Folding@Home new ATI core available

theAnimal

Diamond Member
Mar 18, 2003
3,828
23
76
Linky

This is not yet an auto-update, so stop your client(s) and rename (or delete) FahCore_11.exe, then restart and new core will be downloaded.

I'm getting a nice PPD increase on my 3870.
 

Drsignguy

Platinum Member
Mar 24, 2002
2,264
0
76
Installed and seems to be running a bit better....Thanks For the link "theAnimal"


I was running the (6.20) client on that rig and had to uninstall it. It ran into errors with the new core 1.22 beta version. After the removal, installed the latest 6.23 client and all is a'okay.
 

Drsignguy

Platinum Member
Mar 24, 2002
2,264
0
76
Originally posted by: SwampKracker
I am now running the new core with FLUSH_INTERVAL set to 256 for my HD4850. Looks good so far.


Explain this in a bit more detail if you could.....;)

 

GLeeM

Elite Member
Apr 2, 2004
7,199
128
106
Originally posted by: Drsignguy
Originally posted by: SwampKracker
I am now running the new core with FLUSH_INTERVAL set to 256 for my HD4850. Looks good so far.


Explain this in a bit more detail if you could.....;)

From OPs linky:

There are some flags people can play with until we can set these automatically. In your environment, you can set FLUSH_INTERVAL to a value between 2 and 1024. This flag sets the maximum command packet size sent to the GPU. The default in the core is 16 which seems to solve the lag problems for most people, but at some added CPU overhead. People with higher-end boards can manually adjust this value until we deprecate this flag in later cores. The higher the value this flag is, the less driver overhead and lower CPU usage you will have, but there is still CPU optimization work going into later CAL releases and F@H cores. Setting this value to high will cause UI lag and potentially VPU recover and is designed more for advanced users. The lower the value, the smaller the command packet which translates into other graphics apps being able to play nice.
 

SwampKracker

Junior Member
Dec 16, 2008
5
0
0
Another useful snippet from OPs linky:

PPD in general should have come up for all WUs. The previous core + CAL 1.2.1 dlls (6.23 client) had a *much* higher setting for FLUSH_INTERVAL than the current default, but that was causing issues with lower end boards and even some with high-end boards and graphics apps/video playback. You can try resetting to a higher value to see if things get happier. FLUSH_INTERVAL=256 seems to be where 48XX boards seem happiest in terms of performance and not much more is seen with higher values. However, this may cause lag on Vista
 

Assimilator1

Elite Member
Nov 4, 1999
24,137
510
126
good news as I've just bought a HD 4830! :)

Originally posted by: SwampKracker
I am now running the new core with FLUSH_INTERVAL set to 256 for my HD4850. Looks good so far.
What sorta score are you getting?
 

SwampKracker

Junior Member
Dec 16, 2008
5
0
0
I don't have enough info yet to know how much improvement with the new core. I am very new to FaH, thus experimenting with client options and different clients. Wish I had a better answer for you.

EDIT: Something I have been meaning to do is figure out what I need to do to get max points per day on my E8500 w/ HD4850 and still allow for normal daily use. Both cores running at 100% does not seem to be ideal.
 

Cheex

Diamond Member
Jul 18, 2006
3,123
0
0
I followed the steps and deleted the FahCore_11.exe but when it downloaded the core....it re-downloaded the same one.....no FahCore_12.exe :(
 

SwampKracker

Junior Member
Dec 16, 2008
5
0
0
I do now. It calculates 2567ppd from a 511 credit WU. FYI - I have an overclocked HD4850. GPU @ 680MHz and Mem @ 1050MHz.

Thanks for the tip.
 

Cheex

Diamond Member
Jul 18, 2006
3,123
0
0
Originally posted by: SwampKracker
Cheex - It's a new version of FahCore_11.exe. Version 1.22

Thanks for that tip. How can I check for sure what version i have running??
 

Cheex

Diamond Member
Jul 18, 2006
3,123
0
0
Thanks.

I copied this from my fahlog.txt file.

[18:35:15] Folding@Home GPU Core - Beta
[18:35:15] Version 1.22 (Mon Dec 8 12:57:56 PST 2008)


It seems as if I've been running version 1.22 since as early as December 8.

:confused: