Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!dali.cs.montana.edu!uakari.primate.wisc.edu!caen!sol.ctr.columbia.edu!ira.uka.de!fauern!NewsServ!!roell
From: roell@informatik.tu-muenchen.de (Thomas Roell)
Newsgroups: comp.unix.sysv386
Subject: Re: Roell's X11 and ET4000 performance
Message-ID: <1991May31.081607.4435@Informatik.TU-Muenchen.DE>
Date: 31 May 91 08:16:07 GMT
References: <217@gem.stack.urc.tue.nl>
Sender: news@Informatik.TU-Muenchen.DE
Organization: Inst. fuer Informatik, Technische Univ. Muenchen, Germany
Lines: 18
In-Reply-To: marc@gem.stack.urc.tue.nl's message of 30 May 91 14: 56:27 GMT

>62 MHz vs. 65 MHz dot clock, the SCLK's of the ET4000 were 36 MHz and 50 MHz
>respectively):(a 386 25MHz system without cache was used):

I think the SCLK is the important thing. The SCLK is the base for all DRAM IO.
The faster the SCLK the faster is the DRAM access (if you baord has slower
DRAMs, the manufactorer would have used a slower, while inexpensiver quartz).
Thus a SCLK with 50MHz provides much more speed. But the imense improvement of
speed results from another fact, too: If you SCLK is driver at 36MHz, all
almost available memory bandwidth is spent up for display refresh. If you have
a SCLK of 50MHz about half the bandwidth is free for CPU accesses. Thus the
386 can driver the Et4000 at full io-speed.

- Thomas
--
_______________________________________________________________________________
E-Mail (domain):	 roell@lan.informatik.tu-muenchen.de
UUCP (if above fails):   roell@tumult.{uucp | informatik.tu-muenchen.de}
famous last words: "diskspace - the final frontier..."