hodasemi/cgmath
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Documented the mat4 multiply optimization.

Browse source
This commit is contained in:
Colin Sherratt 2013-11-27 12:20:13 -05:00
parent e4689ce22b
commit 149c781e81
1 changed files with 8 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
src/cgmath/matrix.rs
View file

@ -458,12 +458,16 @@ for Mat3<S>
}
}
// Using self.r(0).dot(other.c(0)) like the other matrix multiplies
// causes the LLVM to miss identical loads and multiplies. This optimization
// causes the code to be auto vectorized properly increasing the performance
// around ~4 times.
macro_rules! dot_mat4(
($A:expr, $B:expr, $I:expr, $J:expr) => (
*$A.cr(0, $I) * *$B.cr($J, 0) +
*$A.cr(1, $I) * *$B.cr($J, 1) +
*$A.cr(2, $I) * *$B.cr($J, 2) +
*$A.cr(3, $I) * *$B.cr($J, 3)
(*$A.cr(0, $I)) * (*$B.cr($J, 0)) +
(*$A.cr(1, $I)) * (*$B.cr($J, 1)) +
(*$A.cr(2, $I)) * (*$B.cr($J, 2)) +
(*$A.cr(3, $I)) * (*$B.cr($J, 3))
))
impl<S: Float>