[Mesa-dev] [PATCH v2 3/3] util: use faster zlib's CRC32 implementaion
Marek Olšák
maraeo at gmail.com
Thu Jan 11 01:17:41 UTC 2018
Reviewed-by: Marek Olšák <marek.olsak at amd.com>
Marek
On Tue, Jan 9, 2018 at 11:58 PM, Grazvydas Ignotas <notasas at gmail.com> wrote:
> zlib provides a faster slice-by-4 CRC32 implementation than the
> traditional single byte lookup one used by mesa. As most supported
> platforms now link zlib unconditionally, we can easily use it.
>
> Improvement for a 1MB buffer (avg MB/s, n=100, zlib 1.2.8):
>
> i5-6600K C2D E4500
> mesa zlib mesa zlib
> 443 1443 225% +/- 2.1% 403 1175 191% +/- 0.9%
>
> It has been verified the calculation results stay the same after this
> change.
>
> Signed-off-by: Grazvydas Ignotas <notasas at gmail.com>
> ---
> v2: drop the size threshold check because size is unlikely to be that
> low of things mesa is typically hashing
>
> src/util/crc32.c | 12 ++++++++++++
> 1 file changed, 12 insertions(+)
>
> diff --git a/src/util/crc32.c b/src/util/crc32.c
> index f2e01c6..9edd3e1 100644
> --- a/src/util/crc32.c
> +++ b/src/util/crc32.c
> @@ -31,10 +31,13 @@
> *
> * @author Jose Fonseca
> */
>
>
> +#ifdef HAVE_ZLIB
> +#include <zlib.h>
> +#endif
> #include "crc32.h"
>
>
> static const uint32_t
> util_crc32_table[256] = {
> @@ -112,10 +115,19 @@ uint32_t
> util_hash_crc32(const void *data, size_t size)
> {
> const uint8_t *p = data;
> uint32_t crc = 0xffffffff;
>
> +#ifdef HAVE_ZLIB
> + /* zlib's uInt is always "unsigned int" while size_t can be 64bit.
> + * Since 1.2.9 there's crc32_z that takes size_t, but use the more
> + * available function to avoid build system complications.
> + */
> + if ((uInt)size == size)
> + return ~crc32(0, data, size);
> +#endif
> +
> while (size--)
> crc = util_crc32_table[(crc ^ *p++) & 0xff] ^ (crc >> 8);
>
> return crc;
> }
> --
> 2.7.4
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
More information about the mesa-dev
mailing list