Subject: Re: [boost] [Review] Boost.Endian mini-review
From: Peter Dimov (lists_at_[hidden])
Date: 2015-01-24 11:09:39
> This approach - as shown - does have a drawback. If you have an array of
> 802511 32 bit lsb integers in the file, and the native int is 32 bit lsb,
> one read of 802511*4 bytes is vastly superior in performance to a loop
> that would read 4 bytes and call read_32_lsb on the byte[4]. Which is why
> the above is generally combined with
>
> void read_32_lsb_n( int * first, int * last, FILE * fp );
It occurs to me that the library could benefit from similar additions to
little_to_native et al, which are currently scalar-only and would,
therefore, depend on a Sufficiently Smart(tm) compiler to optimize out the
empty calls.
Batch endianness fixups are very common and very performance sensitive.