Any helpfull advice on building a fast kernel?
I hate to ask this question but how do I build a fast kernel? I'm not talking about compiling a kernel in a short amount of time, I mean how do I get my custom kernels to process data fast. I'm having a hard time with this, I know that a smaller kernel is supposed to be faster than a larger kernel, but I find this information misleading I've compiled plenty of kernels that were smaller than the default installed kernel but that performed noticeably slower in comparison. I know about the "preemtible kernel" option under processor types and features and I found the preempt-kernel-rml-2.4.20-3.patch which has helped a lot when performing several tasks at once. So far I've got a kernel to perform a little faster than the default slack kernel which is a major step forward but still performs a little slower than WinXP gaming wise. And my goal is to have it perform at least as fast as WinXP on games if thats possible. I'm under the impression that that is possible, so any tips, info or links would be appreciated.