Synthesis-Place-Route benchmark for i386-32bit

[fcdup8k]

This is a benchmark test cd, to get performance data for a specific
computer setup. Please contribute by posting a followup with the
results from the 'real' column.

It's a complete 227 MB zipped .iso file that needs to be written to a
cd and booted. Login as toor, type bench.sh, type the data in a
followup to this posting.
A complete collection of testdata can be found in the '/tmp/
sHHMMSS.tbz' file. And it can be copied by running 'mnt_usb.pl && cp -
p /tmp/s??????.tbz /mountpount/usb0 && umount /mountpoint/usb0' or you
can simple type down the numbers for 'real' and type them here.
If you have any security concerns, please feel free to browse the cd
volume.


Link to the bootcd:
http://www.eatlime.com/download.lc?sid=7F88C923-0B51-1E23-1EEC-53ED80...

http://tinyurl.com/2j8bor


I hope this will shed some better light as to which computer
configurations that will perform better or worse for fpga synhtesis-
place-route process.
A Pentium-III 800 MHz takes approx 25m on the fpga test. Intel core2
~300s, AMD64 ~500s (free from memory). Please use exact numbers.

 

[scottcarl]

Testbenching of designs on different systems is highly overrated. Writing more efficient code with well defined constraints is a much better way of reducing Place & Route time. Adding multi-path delays to non-critical pathways and loosening restrictions as much as possible will evenutally help out more. Also putting your design into a bigger or faster part will naturally help speed things up as well. To test One design on multiple computer systems and claim that system is the best is highly questionable. Just get a computer with the most RAM possible and the fastest CPU you can. But if you are a true computer geek and have time to waste working with multiple systems go for it.

Scott C