On 19/01/13 15:16, Reinhard Tartler wrote:
> On Sat, Jan 19, 2013 at 2:29 PM, Luca Barbato <[email protected]> wrote:
>> Between yesterday and today I started playing and integrating some known
>> memory allocators that provide a mean to have a realigned malloc.
>>
>> Here some numbers from an h264 + ac3 in mkv file
>>
>> ./avconv_normal
>> bench: utime=140.790s maxrss=30508kB
>> ./avconv_tlsf
>> bench: utime=140.026s maxrss=46428kB
>> ./avconv_jemalloc
>> bench: utime=137.821s maxrss=56532kB
>>
>> jemalloc is surprisingly fast, tlsf is slightly faster than the normal
>> dlmalloc used in glibc.
> 
> What machine did you use here? From what I read about it, there

Usual Core i7

> shouldn't be any benefits on a single processor. I would be interested
> to see a benchmark over n, n=1,2,4,8,16 cores here.

The patch for jemalloc had been sent, I'm not so keen in merging it
since I'm using the experimental api to support aligned realloc.

lu

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to