Originally posted by tabicat
View Post
/*
* This is not the fastest implementation, but it is pretty compact.
* The fastest versions of xz_crc32() on modern CPUs without hardware
* accelerated CRC instruction are 3-5 times as fast as this version,
* but they are bigger and use more memory for the lookup table.
*/
* This is not the fastest implementation, but it is pretty compact.
* The fastest versions of xz_crc32() on modern CPUs without hardware
* accelerated CRC instruction are 3-5 times as fast as this version,
* but they are bigger and use more memory for the lookup table.
*/
Yes a lot of using of public domain code that has no valid reason to be there other than attempting to avoid license taint of closed source code or not spending the time to correctly support the platforms.
Yes as you go around the unix directory of the Nvidia open source module you find more and more problems like this one. Yes these problems would make the Nvidia provided open source driver not merge-able into upstream Linux kernel.
tabicat once you take the time to look behind the curtain as Nvida now allows the Nvidia closed source module is not in that good of shape really. Has very poor integration with Unix/Linux platforms with a lot of NIH problems caused by developing to support having a closed source blob. Then you have that Nvidia has never spent the time to develop a stable ABI for their driver from kernel space to user space on Unix/Linux/bsd platforms.
Interesting point right that lot of the fastest implementations of stuff turn out not to be under public domain licenses.
Comment