[Shootout-list] Re: Integer overflow
Isaac Gouy
igouy2@yahoo.com
Sun, 26 Sep 2004 09:35:39 -0700 (PDT)
> >> the producer/consumer threads test ...
> >> originally designed to show the cost of forking new processes.
> > It only forks 2 threads. It switches thread context many times.
> Ahh -- yes, you are right. So this would be our concurrency test.
> Perhaps it is not complicated enough?
Discounting startup time, that benchmark shows Erlang HiPE #2 and
Erlang #3. Isn't that what we would expect from a language designed for
massive concurrency?
http://igouy.port5.com/shoot/prodcons-all50.htm
imo it would be reasonable to add a test to measure process/thread
creation time. (Create a 1000 processes/threads and tear them down,
n-times - like the object creation test.)
(Why we would want a more "complicated" test?)
> Why don't we modify the input file so that it includes sufficient
> distribution of positive and negative numbers that any combination of
> copies of the file is always clamped below 2^30?
Well done, that's a simple fix for that particular benchmark.
We need to check the other benchmarks for problems caused by increasing
n.
_______________________________
Do you Yahoo!?
Declare Yourself - Register online to vote today!
http://vote.yahoo.com