Subject: Re: [boost] Proposal: Monotonic Containers - Comparison with boost::pool, boost::fast_pool and TBB
From: Simonson, Lucanus J (lucanus.j.simonson_at_[hidden])
Date: 2009-06-21 17:07:35
Christian Schladetsch wrote:
>> So, if Christian could run some tests where the allocations are
>> reserved (quite literally for vector, but one would have to
>> prepopulate other containers when using the standard allocator) and
>> only *accesses* to those elements are measured.
> Updated results for
> are here http://tinyurl.com/m83vll for GCC and here
> http://tinyurl.com/n9g8jv for MSVC.
> TBB is faster than monotonic in one test on Win32, sort_list, which is
Looking at your MSVC data it seems that the performance advantage of monotonic over other allocators diminishes as the benchmark size increases. I see this trend is most of the MSVC benchmarks. I'd like to see the benchmarks extended to see if monotonic becomes slower than the other allocators, as the trends suggest it might. Have you considered that extending the buffer instead of reusing the buffer would lead to loss of cache performance because reuse of in cache memory will be faster than bringing in new memory all the time with cache misses? If you are on a system with limited cache size this would be a more serious problem.