Re: [boost] [gsoc] heap review

Subject: Re: [boost] [gsoc] heap review
From: Phil Endecott (spam_from_boost_dev_at_[hidden])
Date: 2011-05-12 12:27:19

Tim Blechmann wrote:
>>>> there will be a `formal' review announcement, i want to invite people
>>>> to check out code, examples and documentation ...
>>>
>>> One quick note: your benchmarks operate on just 16 items 
>
> not quite: they operate on differently sized heaps (roughly 1 to 100000 
> elements), but each operation is performed 16 times (to get some kind of 
> averaging).

OK, but the issue is that I don't trust that you can accurately measure 
execution times of the order of microseconds.  The overhead of, for 
example, the system call to read the clock will be significant.  If you 
believe that you can accurately measure such short periods, you should 
justify your methodology.  Otherwise, repeat a few thousand times.

Also, it might be interesting to look at non-random data.  Howard 
Hinnant (IIRC) did some great work on the libc++ sort implementation to 
benchmark things like sorting already-sorted, reverse-sorted, and other 
non-random inputs.  (Sorry, I can't find a link right now.)  I could 
imagine that non-random inputs might be quite important for heaps too.

Otherwise, this looks like useful work!

Regards,  Phil.