-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Finder uses (struct statfs).f_iosize as a hint for it's chunk size. A lot of other utils use this too (like ls), so make sure they work with the larger size. What I want to know, is how/why the Finder special cases HFS volumes, as the HFS driver sets f_iosize to it's block size (usually 4K), but somehow the Finder uses much larger chunks. Are you porting XFS to Mac OS X? That would be kind of cool. I've been thinking about trying to port Resier once I get Ext3 journaling support done (or should I say IF I get it done). HTH. On Tuesday, December 2, 2003, at 10:02 PM, Sam Vaughan wrote:
When doing a copy to/from our file system, the Finder issues the
writes/reads in tiny 12k chunks. Since it sets the VNOCACHE_DATA
flag, we do direct I/O to the disk to get the best performance.
The problem is that we could provide vastly better performance if the
chunk sizes were much larger. Chunks of 4MB or more would be ideal,
but even increasing to 128k would improve performance by up to a
factor of 8.
Does the Finder have any logic to decide what size transfers to issue?
Does it call any VFS layer hooks that we could implement?
Thanks,
Sam
Brian Bergstrand <http://www.classicalguitar.net/brian/> PGP Key: <http://www.classicalguitar.net/brian/misc/public_key.txt> I may not have gone where I intended to go, but I think I have ended up where I intended to be. - Douglas Adams -----BEGIN PGP SIGNATURE----- Version: PGP 8.0.2 iQA/AwUBP82DDXnR2Fu2x7aiEQJs6ACgzeeKZPvYBJrYSe/SADeplnbqNDQAn0IV nW0LNmmsbYsJHFlZt5gHLt36 =g5+r -----END PGP SIGNATURE----- _______________________________________________ darwin-kernel mailing list | darwin-kernel@lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.