compiler option for builing x265

[applemax82]

I find one report about optimizing hevc via icc. Below is the link:

http://software.intel.com/en-us/android/articles/real-time-end-to-end-h265hevc-solution-for-intel-architecture-based-platforms

the report claim, 'In the x.265 project, Intel® SSE instructions were utilized for vectorization tuning, which contributes to over 70% performance speedup. With further Intel® C Compiler compiling optimization, we get 2x speedup¹ on the IA platform.'

In my side, I don't get the result of '2x speedup'.

what's compiler's option? I think the option is key to speed up program, which involve lots of implicit optimization technique imposed by icc.

Thanks!

[TimP]

Perhaps someone on the companion android forum could address this..

[Kittur_G_Intel]

Hi,

I'll try to contact the author and let you know what options were used in his paper for x265 project and will update accordingly....

Regards, Kittur

[Kittur_G_Intel]

Hi, ]

Pardon the delay. I received a response from the author of the paper who ratified that in figure 4 (referenced in the paper) you can see that icc contributed to 20% performance speedup (212/173) and got the 2x speedup by using SIMD+icc (367/173).  Please let me know if you need any further clarification, thanks.

Regards, 
_Kittur