Path: utzoo!utgpu!water!watmath!clyde!att!rutgers!cs.utexas.edu!sm.unisys.com!mlogic!marcel
From: marcel@mlogic.UUCP (Marcel Samek)
Newsgroups: comp.sys.sgi
Subject: Re: 4Sight, X or NeWS?
Summary: SGI 4D pixel performence sucks
Message-ID: <128@mlogic.UUCP>
Date: 24 Oct 88 00:24:13 GMT
References: <8810202336.AA03068@ucbvax.Berkeley.EDU> <20894@sgi.SGI.COM>
Reply-To: marcel@mlogic.UUCP (Marcel Samek)
Organization: Media Logic, Inc., Santa Monica, CA
Lines: 31

In article <20894@sgi.SGI.COM> miq@chromavac.SGI.COM (Miq Millman) writes:

>graphics hardware is incredibly fast, so performace is not hurt.  On the 4D
>series, pixel performance is about 8 million pixels per second.
>

In the context of the discussion, which was about pixel based operations,
this statement is utter nonsense.

The architecture of the 4D series does not provide direct  acess  into  the
frame  buffer  and  thus any pixel data generated by an application program
must be passed down the geometry pipeline.  While the 3D  geometry  engines
may  be  able  to get pixel performance of 8 million pixels per second when
painting 3d data, application programs can achieve  pixel  performances  of
significantly  less  than  100,000  pixels  per second (roughly 2 orders of
magnitude less) when painting pixel data.

If you wish to read pixels  from  the  frame  buffer,  the  performance  is
significantly slower than if you wish to write pixels.

It is disappointing to see Silicon Graphics employees post such deceptive
garbage to the net.

Marcel


-- 
Marcel A. Samek                      | Media Logic Incorporated
				     | 2501 Colorado Blvd. Suite 350
ARPA: mlogic!marcel@unisys.sm.com    | Santa Monica, CA 90404
UUCP: ...sdcrdcf!mlogic!marcel       | (213) 453-7744