On Mon, 2007-08-13 at 12:40 +0100, Ross Burton wrote: > On Mon, 2007-08-13 at 04:28 -0700, Alex Malinovich wrote: > > Try as I might, I just can't find any reason why this should be like > > this. My CPU usage never goes above 1-2% during the copy, so it > > shouldn't be CPU-bound. > > It's an inherent issue in the design of gnome-vfs. The replacement, > gvfs, is being written now and is already an order of magnitude faster > with SMB despite having no performance tuning yet. :) I've actually been reading up on gvfs for a while, and I agree that it's going to be great once it's done. Between the stateful connections and the direct filesystem access for legacy apps I think it will be great. But the key term is that it WILL be great. I don't see gvfs replacing gnome-vfs for at least another 1-2 releases of Gnome, so we're looking at 6-12+ months. But in the meantime, gnome-vfs is so bad that I find it hard to believe that it's just an architectural thing and not a problem with how something is configured. If it was just sloppy code I would expect to either see my CPU usage go through the roof, or my bandwidth go through the roof with retransmits while the actual USEFUL data coming across is very low. But neither of these is the case. It's almost as though someone decided to put a 10 MB/sec cap on all gnome-vfs data transfers, which seems pretty crazy. -- Alex Malinovich Support Free Software, delete your Windows partition TODAY! Encrypted mail preferred. You can get my public key from any of the pgp.net keyservers. Key ID: A6D24837
Attachment:
signature.asc
Description: This is a digitally signed message part