From: Rodolphe Ortalo <Rodolphe.Ortalo@cert.fr>
  To  : ggi-develop@eskimo.com
  Date: Tue, 03 Aug 1999 11:34:30 +0200

Re: Matrox GGI accellerator (Correction)

Oops, sorry, I did a big mistake when speaking
of pingpong buffers.

Rodolphe Ortalo wrote:

> But the
> idea is that the latency of issuing ONE command
> will be greatly reduced, and this is very desirable,
> especially for 3D.

It's not the latency that is improved. It's even
the opposite, the latency is worse (because
you may need to fill a whole page of command
before it is sent to the card driver).

However, you may expect a better THROUGHPUT
as commands can be handled in batch (several of
them at once in sequence) by the card driver.
Especially with 3D display lists, the card driver can
check several commands at once, chain them and
set up the acceleration engine for processing all
the display lists. This within a single switch
to kernel mode.

Sorry for the confusion (between throughput
and latency)...

Rodolphe