From: Carlos Murillo <cmurillo_at_emtelsa.multi.net.co>
Date: Sat Oct 27 21:16:37 2001

At 12:17 AM 10/28/01 +0100, Iggy wrote:
>I've got an IBM-made OEM machine (mine is branded by Lap Power) with a big
>daughter-board with seven or so ISA slots and three VLB slots. The mobo
>features a Blue Lightning processor and parity RAM.
>Oh, it had an all-plastic DLC-33 FPU from IIT, too.

I liked IIT FPU's. Long before MMX, they had things like 4x4
matrix multiply instructions (for implementation of 3D homogeneous
transforms, rotation/scaling & such). And they were better at
raw computations than the corresponding ix87 part. I remember
hand-coding some robot control laws using those instructions.
Even though the scaling row is usually left unused in robot apps
(but it still is used for computations in the IIT FPU), thus
producing some inefficiency, the built-in 4x4 was still much faster than
a 3x4 matrix multiply algorithm.


Carlos E. Murillo-Sanchez carlos_murillo_at_nospammers.ieee.org
Received on Sat Oct 27 2001 - 21:16:37 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:34:21 BST