nanogui: Nano-X running on BF531 uClinux board: extermely slow updates
Subject:
Re: Nano-X running on BF531 uClinux board: extermely slow updates
From:
"Aaron J. Grier" ####@####.####
Date:
8 Jun 2004 18:07:34 +0100
Message-Id: <20040608170700.GO17401@mordor.unix.fryenet>
On Mon, Jun 07, 2004 at 02:43:15PM +0530, vishal wrote:
> Does anybody have any benchmarks for running Nano-X smoothly....i mean
> the kind of preocessing power required..or is 75MHz sufficient?
I run with a 25MHz 68331. the platform started out with an Epson
S1D13505 (used to be SED1355). those was the nano-X 0.89-pre7 days, and
things were pretty slow until we started optimizing things for our
platform. a few of them have since made it into the mainline source.
> I would loike to know if there is something that could be done inside
> Nano-X to optimise it futher, or whether it could be a limitation of
> our hardware itself.
you need to do some profiling and determine what calls are taking the
most time.
also helpful would be some raw video bandwidth tests, to give you an
idea of the absolute fastest speed you can go. test video to video,
video to main memory, and main memory to video copy speeds. use
zero-overhead looping if your CPU supports it.
> I have also been asked to look at alternative options to Nano-X,
> specifically directfb with a directfb vnc client. How does nano-X
> compete with programs of a similar nature? I would really appreciate
> it if anyone would suggest any ideas.
I belive VNC makes heavy use of blitting; if you can get the blit
routines sped up, I suspect your performance will increase.
--
Aaron J. Grier | Frye Electronics, Tigard, OR | ####@####.####
