Mr. Trend
Why did you choose CFQ as IO Schduler ?
Anticipatory would be better for game servers ?
2.6.31.11 and 2.6.32.3 are built and syncing to kernel.hardwareforums.com. These are both bugfix builds containing many fixes, and as such, it is recommended to upgrade if you're running a HWF kernel.
If this is your first HWF kernel, I would recommend trying 2.6.32.3, which has some hefty performance improvements over previous kernels.
- Get HWF kernels here
- 2.6.31.11 Changelog
- 2.6.32.3 Changelog
Once again, a new kernel has been posted. 2.6.31.12 and 2.6.32.4 are built and syncing to kernel.hardwareforums.com. These are both bugfix builds containing many fixes, and as such, it is recommended to upgrade if you're running a HWF kernel.
If this is your first HWF kernel, I would recommend trying 2.6.32.4, which has some hefty performance improvements over previous builds.
- Get HWF kernels here
- 2.6.32.4 Changelog
- 2.6.31.12 Changelog
Immediately on the heels of 2.6.32.4, we have 2.6.32.5. This is a bugfix build with many scheduler-related fixes and more. As usual, it's recommended to upgrade if you're running a HWF kernel.
- Get HWF kernels here
- 2.6.32.5 Changelog
Believe it or not, we already have yet another bugfix release, 2.6.32.6, which is currently syncing with the server. As usual, it's recommended to upgrade if you're running a HWF kernel.
- Get HWF kernels here
- 2.6.32.6 Changelog
Holy crap, another bugfix release! 2.6.32.7 is currently syncing with the server. As usual, it's recommended to upgrade if you're running a HWF kernel.
- Get HWF kernels here
- 2.6.32.7 Changelog
2.6.32.8 is built and currently syncing with the server. As usual, it's recommended to upgrade if you're running a HWF kernel.
- Get HWF kernels here
- 2.6.32.8 Changelog
I'm actually building 64-bit kernels for a different project, specifically, patched with GRSec/PAX and hardware-optimized. But I've been thinking seriously about building x86_64 desktop kernels also, since it makes a lot of sense. 64-bit is much more desirable on a desktop presently than it was when I started rolling these. We also have more storage capacity on the server, which was a limitation at the time. The biggest bottleneck at this point is my free time; the initial kernel configuration for each CPU profile is what takes the largest investment.
So in summary, it is something I've been considering, and I'll probably do it. When exactly I can't say, but getting rolling is the hard part. After that, it's just another script running in the background.
What's that I hear you ask? No new bugfix builds against 2.6.32? Au contraire! 2.6.32.9 is built and currently syncing with the server. If you're running a HWF kernel, upgrade or youwill die in a horrible waymight run into bugs that were already fixed.
- Click here to
save your precious souldownload the kernel- 2.6.32.9 Changelog
This is not a PC friendly user!!!![]()
"...when you find someone you really love, all the rest is just... other stuff." - Leanne Anika Battersby
"..and no one will know my name until it's on a stone..." - e
Help with digital video?
Help My Mini City to grow
Hit or Miss
I'm not applying any patches at this point, just compiling from vanilla sources.
BFS is a very simplistic scheduler, and it probably works pretty well in many desktop conditions. However, thanks to the help of the x264 creator, CFS has actually improved substantially since BFS was written. These days, doing something like encoding a movie or compiling kernels on my quad core fully utilizes all four cores with nearly zero idle slack. Also, some apps like WINE do not work well with BFS. So even if I was applying third party patches to the HWF kernels, it would not be an easy choice to just ditch the default Linux scheduler for one that's far less well-tested.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Bookmarks