Path: utzoo!attcan!uunet!cs.utexas.edu!tut.cis.ohio-state.edu!cica!gatech!bloom-beacon!EXPO.LCS.MIT.EDU!keith
From: keith@EXPO.LCS.MIT.EDU
Newsgroups: comp.windows.x
Subject: Re: Speedups for sample server color code?
Message-ID: <8907211536.AA20204@xenon.lcs.mit.edu>
Date: 21 Jul 89 15:36:51 GMT
References: <1697@brwa.inmos.co.uk>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: The Internet
Lines: 14


> I am aware of the Purdue speedups for ddx/mfb.  Are there any generic
> improvements to cfb?  Will the R4 server have improvements in this area?

As I am currently using a color DECStation 3100, I think you can expect to
see some cfb speedups in R4.  Just last night I was hacking around with
TEGlyphBlt and sped that up by a factor of 7 (and, no, you can't get that
code yet, it doesn't really work yet).

But, your best bet for raw speed will probably always be the commercial
product servers; most companies are spending quite a bit of effort in
hand-tuned assembly code to get that last bit of speed out of whatever
hardware they're using.  Our code will continue to be as portable as we
can conveniently make it.