On Dec 31, 2:15 am, Patrick F <
patrick.fl...@gmail.com> wrote:
> Hi Dmitriy,> What exactly queue do you mean?at first i took a look at your mpmc block queue i got from
1024cores.net.Im not sure if I understand the code correctly, but it only works for 32 Bits pointer types, correct?
>
> In turn i took your code from the bounded queue also found at
>
1024cores.net, which was much easier to understand.You were right, the
> porting to g++ compatible code was very straight forward.
> Unfortunately the performance was not as hoped. The following shows
> the results in comparison to the TBB concurrent queueand a simple
> wrapped fifo queue using ordinary mutexes. The Queues are previously
> filled with 1,000,000 (1 million) elements. The execution time (in
> seconds) was measured to completely empty the queue via the dequeue
> operation. So only dequeue operations are measured.
> Threads Dmitriy MPMC Wrapped FIFO TBB queue1 0.055961
> 0.051541 0.0572042 0.294067 0.635973
> 0.1890954 0.313675 0.632279 0.2216298
> 0.410893 0.36722 0.323356
> Are these results plausible? I hoped the bounded queue would be alot
> faster ...
> Thanks in advance! ------------- Patrick F. P.S. Sorry for any
> spelling or grammar errors - I am not a native speaker of english.