From: Yaroslav R. <ya...@re...> - 2003-12-13 08:47:17
|
On Saturday 13 December 2003 04:27, William Lee Irwin III wrote: > On Wed, Dec 10, 2003 at 09:29:29PM -0800, William Lee Irwin III wrote: > > Successfully tested on a Thinkpad T21 and 32GB NUMA-Q. Any feedback > > regarding performance would be very helpful. Desktop users should > > notice top(1) is faster, kernel hackers that kernel compiles are faster, > > and highmem users should see much less per-process lowmem overhead Performance is, indeed, better. My Thinkpad T21 feels slightly on steroids with -wli-2 :-). Some problems, though, were encountered: 1. fs/dcache.c/d_validate() function was removed in your patch, but it is used in at least one place ( fs/smbfs/cache.c/smb_dget_fpos() ). 2. With 2.6.0-test11 vanilla, suspend-to-ram/suspend-to-disk was working nicely. With -wli-2 suspend fails on 'stopping tasks' phase, and aborts with Dec 13 01:08:47 localhost kernel: Stopping tasks: ====================== Dec 13 01:08:47 localhost kernel: stopping tasks failed (1 tasks remaining) Dec 13 01:08:47 localhost kernel: Restarting tasks...<6> Strange, swapper not stopped Hope this helps. -- With all the best, yarick at relex dot ru. |