Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 1 Feb 1996 14:29:25 -0700 (MST)
From:      Terry Lambert <terry@lambert.org>
To:        dfr@render.com (Doug Rabson)
Cc:        hasty@rah.star-gate.com, lehey.pad@sni.de, tinguely@plains.nodak.edu, hackers@FreeBSD.ORG
Subject:   Re: Good news -- pipe stuff
Message-ID:  <199602012129.OAA13787@phaeton.artisoft.com>
In-Reply-To: <Pine.BSF.3.91.960201125238.17901E-100000@minnow.render.com> from "Doug Rabson" at Feb 1, 96 12:56:53 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> Personally, I would prefer that Intel made byte and halfword memory 
> operations go as fast as 32bit operations.  As far as I am concerned, the 
> two most useful pixel formats for 3D rendering are 8bit and 16bit.  The 
> image quality gain for 32bit pixels is never enough to justify doubling 
> the memory bandwidth requirements.
> 
> The first set of 3d hardware is a bit disappointing but its not too bad.  
> The second wave should be excellent.  Intel's MMX instruction set 
> extensions make a P5 into a pretty good 3D accelerator too...

I'd like to see unaligned acces of data cache items done with address space
shift replication.

That way I could do 32 bit I/O to the cache (re: Pentium bcopy) and not
take a hit for unaligned source/target addresses.


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199602012129.OAA13787>