Re: Re: CRC slicing-by-8

[p . z . l]

Thanks for the suggestion, but I have already tried faster ways of

calculating the CRC and I do not think that the (small) increase in speed is

worth the increase of portability problems it causes.

I understand.

Have you got any idea why this:

    c = crc32[(c ^ buffer[i]) & 0xFF] ^ ( c >> 8 ); /// lzip

is faster than this:

    c = (c >> 8) ^ crc32[(c & 0xFF) ^ buffer[i]]; /// sbrumme crc

?

I noticed the difference in lzip (mentioned mod) and tested it here: (https://pastebin.com/tkcMbTHR). Although in this setup difference is smaller in lzip is more noticeable.

PS. What do you think about Time/Speed of de/compression summary (https://lists.nongnu.org/archive/html/lzip-bug/2022-02/msg00001.html)?