*BSD News Article 18257


Return to BSD News archive

Newsgroups: comp.os.386bsd.questions
Path: sserve!newshost.anu.edu.au!munnari.oz.au!network.ucsd.edu!swrinde!cs.utexas.edu!uunet!olivea!apple.com!netcomsv!netcom.com!hasty
From: hasty@netcom.com (Amancio Hasty Jr)
Subject: Re: S3 video cards supported by XS3?
Message-ID: <hastyCA1Bx3.DxH@netcom.com>
Organization: Netcom Online Communications Services (408-241-9760 login: guest)
References: <CA119w.CIC@csn.org>
Date: Mon, 12 Jul 1993 04:46:15 GMT
Lines: 141

In article <CA119w.CIC@csn.org> millert@teal.csn.org (Todd Miller) writes:
>Hi all, I'm about to put together a 486 EISA VESA local bus machine and I'm
>wondering if there is some one that can point me to a list of S3-based local bus
>video cards that XS3 supports.  I dug around but can't seem to find the info
>I need.
>

Here is an old posting and hopefully soon I will be getting a premetheus
S3 928 (ISA card).

The purpose of this posting is to explain the different kinds of S3 chipsets


In chronological order of availability, the current S3 chipsets:

1	8C911	1MB VRAM based card  	ISA/EISA/Local Bus
2	8C924	1MB VRAM based card	ISA/EISA/Local Bus
3	8C801	2MB DRAM based card	ISA
4	8C805	2MB DRAM based card	EISA/Local Bus
5	8C928   4MB VRAM based card


The Actix Graphic Ultra (S3 928) support can supports upto 2MB of VRAM.


The GraphicsEngine32 Series 8C801 with 1MB of memory specifications are:

Resolution		Number of colors	Refresh rate in MHz

640x480			16.7 Million		60
640x480			65,536			60
640x480			32,768			60
640x480			256			72
800x600			65,536			60
800x600			32,768			60
800x600			256			72
1024x768		256			72

With 2 MB of memory:

1024x768		65,536			43.5(i)
1280x1024		256			60


The fastest chipset is an S3 928, then an 801 and last the 911 or 924.

Due to a bug in 386bsd the linear mode access functionality for the
S3 805 does not work. Bill Jolitz promised to fix mmap for his
release of 386bsd 0.2.

The 805 card's bus speed runs at 16Mhz not at 33Mhz. If you try to run them
at full local bus speed the graphic engine will lock up.
(Assuming that you have a 486/33 or 486/50).



The version number of the S3 928 chip is 93133 (yy/week/day)
if you buy an S3 928 based card which has an earlier version and
try to use XS3 you are on your OWN!!!!

For 2MB of operation for resolutions like 1152x900 or 1280x1024 the
S3 928 card is highly recomended due to its higher performance.
Currently, we don't support it but with a bit more luck we will
provided it.

The S3 801 is recommended for resolutions of 1024x768 and at clock rates
no higher than 65Mhz. If you try to use higher clock values your performance
will suffer. Yes, the card will work but it will not be as fast as lets
say a sparc 2 workstation.

You can expect around 102k xstones on an ISA S3 801 with a fast
486DX2 66Mhz and XS3.


XS3 is known to work with just about all S3 801, 911/924 ISA/EISA/VLB.
Notable exception is of course Diamond.

XS3 is known to work with the following S3 928 cards:
o	 ELSA Winner 1000 VLB 
o	 Actix Ultra 
o	 #9 GXE VLB and ISA

The 386bsd release for the S3 928 is XS3.mmio available
at sunvis.rtpnc.epa.gove:/pub/386bsd/incoming

***DO NOT RUN this server with any other chipset for it will not work ***


So far the optimal cpu configuration, is a 486DX2 66Mhz VLB 

Somehow we have to sort out the fastest 486DX2 66Mhz, and this
is outside my scope.

Here is what David Greeman <davidg@implode.rain.com> has to say about
his 486DX 66Mhz VLB, XS3 with his #9 S3 928 VLB  running under 386bsd :-)

-------------------------------------------------------------------

   I spent the evening looking into the problem of not being able to compile
the ibm8514 code with the optimizer. Following are the diff's to fix this
problem. With -O -finline-functions -fstrength-reduce -fomit-frame-pointer
and -m486, performance in some areas is dramatically improved. The pixel test
in x11perf, for instance, improves by 40% to 540000 pixels/second. The xbench
results with optimization look like this:

AHJ> resolution 1024x768 at 65Mhz

TOTAL    512796 lineStones
TOTAL     74884 fillStones
TOTAL     53293 blitStones
TOTAL   7701944 arcStones
TOTAL    194218 textStones
TOTAL    147973 complexStones
TOTAL    118630 xStones

In this benchmark, it looks like arcStones benefited the most.

<stuff deleted>

-DG
-------------------------------------------------------------------

I have measured a sparc 10 with a gx at around 100k xstones running
openwindows 3.0 .


Would love to hear what are the numbers for an S3 X server running
on a  fast pentium.


Do expect higher numbers in the future...


Amancio Hasty


-- 
This message brought to you by the letters X and S and the number 3
Amancio Hasty           |  
Home: (415) 495-3046    |  ftp-site depository of all my work:
e-mail hasty@netcom.com	|  sunvis.rtpnc.epa.gov:/pub/386bsd/incoming