[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference milkwy::23class_semiconductor

Title:23class Semiconductors
Moderator:ASIC::BURBICK
Created:Wed Mar 15 1989
Last Modified:Thu Apr 17 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:190
Total number of notes:591

190.0. "FPGA BENCHMARKING" by ASIC::BURBICK () Fri Jun 14 1996 14:31

T.RTitleUserPersonal
Name
DateLines
190.1Xilinx 3100A benchmarkASIC::BURBICKFri Jun 14 1996 15:539
190.2Xilinx 4000 benchmarkASIC::BURBICKFri Jun 14 1996 16:0210
190.3Altera 8000 BenchmarkASIC::BURBICKThu Sep 12 1996 10:4820
190.4Motorola 1036ASIC::BURBICKThu Jan 23 1997 14:0318
190.5Designs to dateASIC::BURBICKMon Jan 27 1997 15:4928
Summary of FPGA designs to date:

Xilinx was a very straight forward tool to design with. I was already familiar
with the Xilinx tools, as well  as viewlogic schematic capture and simulation. 

Altera's tool was fairly easy to use, but it had some differences from what I
was used to. One, Altera's  schematic capture is entirely hierarchical, where
Viewlogic and Valid both use some sort of  paging  system. This created
confusion and slowed the design process. Secondly, Altera does not back
annotate  simulation results to the schematic, which makes it difficult to
quickly debug a mistake. Next, the tool  didn't warn me that I accidentally
shorted outputs to inputs with improper naming on some flops, and I had  to
find it in simulation. Finally, I was surprised at the performance variation
possible by changing the  routing constraints. In general, the design went
fairly well, but I would suggest using the tool you are  comfortable with to
design with Altera, and not going with their schematic capture and simulation
to avoid  the learning curve.

As for the Motorola, the actual design entry and simulation went fairly well.
Since it was viewlogic based,  I had the same ease of use and comfort that I
had with Xilinx. One minor hiccup is that you must use the  Viewlogic built-in
libraries In and Out signal on all i/o's, which is not clear in any of the
documentation.  The place and route tools were very easy to use, and fairly
well documented with the on line help. My only  real nit with the Motorola
tools is that their report files do not contain all the pertinent information 
(Fmax, set up, tco) and that they do not have any static timing capability.
This is probably due to their  immaturity in comparison to Xilinx and Altera,
and hopefully they will improve this shortly.