Re: [lwip-devel] SMEMCPY()

[address@hidden]

Leon Woestenberg wrote:
> Who did performance measurements on this, before converting our code over?
What's the point in detailed performance measurements here? If I 
measured this on my platform, that doesn't say it's optimized equally on 
your platform, too.

As to the theoretical background:
In our port, we are using a hand-written memcpy function (instead of the 
C library), which is optimized for copying big blocks and is much faster 
when copying unaligned data (the C library uses byte-copy if src or dst 
is not 32-bit aligned). However, this function has a large overhead:
- function call
- check src and dst
- unrolled loop checking

In this case, I don't have to do performance measurements to know that 
defining SMEMCPY() to an inlined byte-copy routine is faster than 
calling our memcpy function optimized for large blocks. The optimized 
function ends up with doing the same byte-copy as the inline version 
does, so it adds the overhead of a call and argument checking.

Anyway, this change has been mimal-invasive by defining both MEMCPY() 
and SMEMCPY() to memcpy(), so I cannot see the downside of it.

Simon