Re: RE: Space Wars

        Long-time vectorlisters will note that Clay's been dying to find an
excuse to use this part in a project for a while now =)

                        Drew

   X-Authentication-Warning: beacon.synthcom.com: majordom set sender to owner-vectorlist@synthcom.com using -f
   X-Originating-IP: [24.6.217.124]
   From: "Clay Cowgill" <vector_clay@hotmail.com>
   Date: Wed, 02 Feb 2000 19:44:30 PST
   Content-Type: text/plain; format=flowed
   Sender: owner-vectorlist@synthcom.com
   Precedence: bulk
   Reply-To: vectorlist@synthcom.com
   X-UIDL: 1a27ba091ae3aaf1ffa0377703c4b441

>Actually, this is exactly what the Mathbox is doing.

   Yep. It's just a two-chip mathbox. It's nothing super-tricky-- just using
   the app-note math routines for the Scenix/PIC. The Scenix for the most part
   is a true one-cycle per instruction processor (so it's about 2-3 times as
   fast as a 6502, cycle-wise) and then it's running at a healthy 50MHz. So
   basically the performance is really close to 100-150 times that of a 1.5MHz
   6502 running equivalent code on average. It can be a little better or worse
   depending on some instruction-set differences.

   (So for a sorta crude example: let's say that a 16-bit multiply in the 6502
   takes about 650 cycles. At 1.5MHz that's about 433us. The Scenix would
   blast through the same problem in about 96 Scenix-cycles. However, those
   cycles are at 50MHz, so the total time is about 1.92us, or a bit over 220
   *times* faster. Even if it takes another 20 cycles to get stuff written
   back out to the CPLD registers that's only around 2.32us... That's about
   half the amount of time it'd take the 1.5MHz 6502 to do the "JSR"
   instruction to go start the problem on its own. Most operations will take
   longer for the 6502 to read the results than for the Scenix to calculate
   them!)

   -Clay
   ______________________________________________________
   Get Your Private, Free Email at http://www.hotmail.com

   ---------------------------------------------------------------------------
   ** To UNSUBSCRIBE from vectorlist, send a message with "UNSUBSCRIBE" in the
   ** message body to vectorlist-request@synthcom.com. Please direct other
   ** questions, comments, or problems to neil@synthcom.com.

---------------------------------------------------------------------------
** To UNSUBSCRIBE from vectorlist, send a message with "UNSUBSCRIBE" in the
** message body to vectorlist-request@synthcom.com. Please direct other
** questions, comments, or problems to neil@synthcom.com.
Received on Wed Feb 2 23:19:19 2000