Extensions for fine-grained image quality control
James Weatherall
jnw at realvnc.com
Mon Sep 7 16:17:38 BST 2009
Hi DRC,
It makes sense for tile size to affect performance in a *bandwidth* dependent way, but not for it to affect latency sensitivity. That said, I gather you're working on a project based on the old VNC 3.3.x codebase, so it may well be that there's some I/O scheduling issue in that.
HTH,
--
Wez @ RealVNC Ltd
> -----Original Message-----
> From: DRC [mailto:drcommander at gmail.com] On Behalf Of DRC
> Sent: 07 September 2009 16:06
> To: <james.weatherall at realvnc.com>
> Cc: <vnc-list at realvnc.com>
> Subject: Re: Extensions for fine-grained image quality control
>
> I'm just reporting what I observed. I don't really know the underlying
> reasons behind it, either. I see the same effect with the Tight
> protocol. I can't get peak performance out of that protocol with tile
> sizes less than 64 kpixels.
>
> On Sep 7, 2009, at 8:23 AM, "James Weatherall" <jnw at realvnc.com> wrote
> > I'm not sure I understand why the tile size affects latency-
> > sensitivity, unless you're flushing each tile to the TCP stack for
> > transmission separately, or something. Can you elaborate?
More information about the VNC-List
mailing list