| commit | 075bf1349b23914d237f4e676d71da57ecfa4b7c | [log] [download] |
|---|---|---|
| author | yann.collet.73@gmail.com <yann.collet.73@gmail.com@650e7d94-2a16-8b24-b05c-7c0b3f6821cd> | Sun Jun 05 21:23:42 2011 +0000 |
| committer | yann.collet.73@gmail.com <yann.collet.73@gmail.com@650e7d94-2a16-8b24-b05c-7c0b3f6821cd> | Sun Jun 05 21:23:42 2011 +0000 |
| tree | de4436c5c700d01f3e9892a67fd92f9bed2e5b1d | |
| parent | 6b798d5e40e4c0788f7c51f6e9b0e9c4e966d068 [diff] |
Greatly improved compression and decompression speed, at the expense of some compression ratio. Most of the change is due to a modification in the performance parameter (HASH_LOG) now set to 12, to match Intel L1 cache processors. You can change it back to 17 to get back previous compression ratio. AMD users are invited to try HASH_LOG = 13, since AMD L1 cache is twice larger. git-svn-id: https://lz4.googlecode.com/svn/trunk@10 650e7d94-2a16-8b24-b05c-7c0b3f6821cd