Re: [boost] [Block Pointer] benchmark

Subject: Re: [boost] [Block Pointer] benchmark
From: Mathias Gaunard (mathias.gaunard_at_[hidden])
Date: 2011-05-25 08:15:33

On 25/05/2011 00:40, Phil Bouchard wrote:
> Hi,
>
> Just for the records I have benchmarked block_ptr using:
> https://svn.boost.org/svn/boost/sandbox/block_ptr/libs/smart_ptr/example/benchmark.cpp
>
>
> And I get the following:
> auto_ptr: 9208870 ns
> shared_ptr: 22958516 ns
> block_ptr: 75860365 ns
>
> Which is pretty good given the complexity it goes thru.

If you want a really useful benchmark, I'd suggest you give the max and 
average cycle count of each primitive associated with those types 
(construction, copy constructor, assignment, dereference, destructor)

> I was wondering if there is any hope in having a pool::ordered_malloc()
> call of O(1) instead of O(n)

I don't see how such a thing would be possible, unless you can guarantee 
that all your allocations/deallocations happen in a LIFO order (in which 
case you can use the non-ordered allocation).

> and a faster pool::is_from()? The latter
> could be just a check to see if a pointer is within reserved memory
> pages of the pool. I tried to talk to the author of pool but I got no
> response.

The pages are not contiguous, so you need to walk through each page at 
least.