IPC benchmarking Project Experiments and benchmarks difference

Hi Team Ti,

                  why is this huge cycles/msg transfer on IPC at this link. Also i see there is huge cycles being reported on IPC communication. Even for shared or qmss, these resuls contradicts IPC benchmark results on wiki page

http://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/p/199157/711580.aspx#711580

Thanks

RC Reddy

Hi Team Ti,

          I am awaiting your response on this topic. Kindly reply

Thanks

RC Reddy

Hi Reddy,

I am not familiarized with this specific IPC project. But, I am looking into it and will reply as soon as possible.

Thanks and regards,

Javier

Hi Javier,

                 Any further understanding/debugging/exploration on the issue and any conclusions on the same.

Thanks

RC Reddy

Hi Reddy,

The thing is that the results of this benchmark (as any other benchmark) will be highly dependant on the configuration used when they were run. Cache utilization, memory endpoints, compiler optimization etc. Even if it is the same code, any change in this configuration can yield different results. I am trying to find out what was the original configuration used in these benchmarks so that I can help you understand it better.

Regards,

Javier

Hi Reddy,

One question. Did you run with "Debug" build profile or with "Release" build profile? "Release" build profile will always yield the most optimized build.

Regards,

Javier

Hi Javier,

               Thanks for reply.

1. I didn't change anything in the example proejct given.I ran the project "as is". 

2. I didn't change example project to "release build profile" as i re-iterate the point 1 [i.e. i ran the project "as is"].

Thanks

RC Reddy

Hi Reddy,

I suggest then to start by changing the build profile to "Release" instead of "Debug". Menawhile, I am getting in touch with the designers of this benchmark to get more information.

Regards,

Javier

Hi Javier,

              Thanks a lot for clarification. It answered my question.

Thanks

RC Reddy