|
|
I am not telling anyone not to use nitro, even though you probably wouldn't listen to me if I did. You use nitro because you like it, and that's fine. The same goes for people who use other patchsets besides the "great and powerful" nitro-sources.
But perhaps it might be worth considering something: do you really need the patchset you use? I mean, do you use all of its features? I never did. If I was left to myself, I could use plain old ck-sources.
fbsplash? vesafb-tng?
Sure, they might look good, but what's the point since (if you're like me) you probably only see your console when you start up. Are those fifteen seconds really going to change your life if there's a picture of something behind the text as it scrolls? Moreover, if you're at all like me, you probably don't even watch it boot up. You turn it on and then go to the john or something, returning to fire up X.
ricer4(reiser4)?
Chances are that you could use reiser3 with less difficulty and an almost unnoticible amount of slowdown. Remember that the physical specifications of a drive will limit its speed anyway. If you want fast disk access, go buy yourself a couple SATA drives.
software suspend/win4lin?
This is quite possibly the only feature of patchsets that I could see as being useful. If you have a laptop, swsusp2 could save you battery power. Win4lin might be necessary for something that wine can't cover at the moment.
mm/lkml patches?
Do you actually know what they do? If not, you don't need them. Same goes for almost all lkml patches other than security fixes. Some of the patches in mm would make your hair stand on end if you actually saw the changes in them!
Plugsched?
The only real use I could see for this on normal computers would be if you alternated it between being a dedicated server and being a desktop. Somehow, I don't think there are many people like that in the world though. Pick a scheduler and stick with it. If you want a desktop machine, go with staircase with schediso/schedbatch. If you want a server, stick with the default scheduler or use staircase with interactive mode disabled (ck-server patchset). That simple.
Genetic *?
I think the real issue with this is that Moilanen wrote it up too well when he first released it. I'm not making any sort of personal attack on him, I just can't see genetic scheduling as being useful in any application except for perhaps a dedicated server.
Jack Moilanen wrote:
The patch is very useful for specific workloads that are difficult to tune.
I'd imagine that he didn't mean desktops when he said "specific workloads" on his web site.
inotify?
This is perhaps the easiest patch in the universe to apply. Just download the appropriate one and give it a little nudge. Smooth as butter.
Realtime scheduling?
Another one that just sounds good. Any performance gain that you notice while using this is likely to be merely a figment of your imagination. Hell, I do professional-grade audio work on my box and couldn't even tell the difference when I used a realtime-enabled kernel!
Again, I am not saying that you shouldn't use your current patchset, whatever that may be. But consider whether or not you need all the features in it. If the answer is no, you could probably go with ck-sources, or gentoo-sources and not notice the difference. They actually work. No, really!
This is not an attack on patchset maintainers here. I was one for a while. Remember? It's tough work, and they put in a lot of time to get things up and running. I applaud you for your work.
If the reason why you use a particular patchset is because you want to support someone's hobby (because that's all this is), then good for you. Any problems you notify them about will serve as a good way for the maintainer to further their knowledge of patching kernels.
I really only intended to say goodbye and announce my resignation in this post, but it seems to have turned into something else. I'll stop now and finish up. |
|
IP卡
狗仔卡
发表于 2005-4-13 10:30:40
提升卡
置顶卡
沉默卡
喧嚣卡
变色卡
显身卡
发表于 2005-4-14 19:54:12
楼主