Re: [boost] Proposal: Monotonic Containers: Performance test

Subject: Re: [boost] Proposal: Monotonic Containers: Performance test
From: Vicente Botet Escriba (vicente.botet_at_[hidden])
Date: 2009-06-11 06:19:00

Scott McMurray-2 wrote:
> 
> 2009/6/10 Christian Schladetsch <christian.schladetsch_at_[hidden]>:
>>
>> monotonic: 0.005
>> std: 0.09
>>
> 
> How about intrusive?  I had to up the count to 0x101010 to get a
> measurable value, which gave the following timings (three runs,
> reported in order):
> 
> mono: 1.37/1.41/1.38
> std: 1.51/1.51/1.51
> intrusive: 1.34/1.33/1.33
> 
> Upping it to 0x1010101 gave this:
> 
> mono: 35.82/35.71/35.82
> std: 37.94/37.83/37.81
> intrusive: 36.17/35.78/35.62
> 
> GCC 4.3.3 with -O3, 32-bit linux on Core2.
> 
> 

Hi,

As you code shows there is essential difference between the intrusive and
the monotonic code. In the case of monotonic the destruction and not the
deallocation is called implicitly while in the case of intrusive, it is up
to the user to call the destruction of the removed element explicitly.

I'm curious to know why the measures are so different concerning the std
part.

Best,
Vicente
-- 
View this message in context: http://www.nabble.com/Re%3A-Proposal%3A-Monotonic-Containers%3A-Performance-test-tp23974719p23978706.html
Sent from the Boost - Dev mailing list archive at Nabble.com.