myhdl-list

[myhdl-list] Update on MyHDL benchmarks with pypy

[Jan D. <ja...@ja...>]

For your info.

Since pypy 1.6, the benchmarks for MyHDL have stagnated
or become slightly worse (within 15%).

Still, the results are much better than the average
speedup over the pypy set of benchmarks. I guess it is
to be expected that improvements depend heavily on
a particular application.

http://myhdl.org/doku.php/performance?&#historical_data


-- 
Jan Decaluwe - Resources bvba - http://www.jandecaluwe.com
     Python as a HDL: http://www.myhdl.org
     VHDL development, the modern way: http://www.sigasi.com
     World-class digital design: http://www.easics.com

Re: [myhdl-list] Update on MyHDL benchmarks with pypy

[Norbo <Nor...@gm...>]

Am 15.06.2012, 10:53 Uhr, schrieb Jan Decaluwe <ja...@ja...>:

> For your info.
>
> Since pypy 1.6, the benchmarks for MyHDL have stagnated
> or become slightly worse (within 15%).
>
> Still, the results are much better than the average
> speedup over the pypy set of benchmarks. I guess it is
> to be expected that improvements depend heavily on
> a particular application.
>
> http://myhdl.org/doku.php/performance?&#historical_data
>
>

hey, i just was able to run the lfsr24 example with my implementation of a  
myhdl simulator (yes sounds crazy but is true,i also was able to run my  
RS232 implementation with it, but it has still a lot of shortcommings and  
i bassically avoided every design rule i ever heard of, so it is nothing  
for production!!)
here are my benchmark results:


lfsr24:
-------------
Myhdl orginal with pypy:  user    1m29.174s
my Myhdl with pypy:       user    0m45.687s


RS232:
-------------
Myhdl orginal with pypy: user    0m2.936s
my Myhdl with pypy:      user    0m1.492s

(what i bassically did is i appended the function of an e.g always block  
directly to the signal on which
the block reacts on) So the yield is only possible in a instance block.