Date:	Sun, 25 Dec 1994 23:11:53 +0900
From:	Stephen Turnbull <turnbull AT shako DOT sk DOT tsukuba DOT ac DOT jp>
To:	djgpp AT sun DOT soe DOT clarkson DOT edu
Subject:	Math Emulator

A couple of days ago I published some balderdash concerning the
performance loss due to notebooks w/o FPU.  My algorithm (subtracting
earliest object time from executable time) simply was invalid as *all*
the builds were interrupted at least once (before the link).
    It turns out that my office system (AMI EISA 486DX/50MHz, 16MB
RAM, 4MB hardware disk cache) gives quite good performance, building
Ghostscript in well under 10 minutes for a rate of about 400
lines/sec.  (Don't forget that this is a multi-tasked environment,
with DESQview/X.)  The notebook in question (IBM ThinkPad 330cs,
486SLC2/50MHz, 12MB RAM, 1MB software cache) still did terribly,
taking 2.5--3.1 hours, for a rate of 15--20 lines/sec.
    I still don't understand why the performance is so bad, but I
don't attribute it to the lack of fp processor.  One possibility is
that there was actually only 4MB RAM in the machine at the time (it
took a while for the upgrade to arrive).  I think IBM just made a bad
machine.
    Details are available if you're interested, but I don't think
they're of great interest to the list in general.
    --Steve

- Raw text -