Skip to content

Instantly share code, notes, and snippets.

@FrankNiemeyer

FrankNiemeyer/gist:893436328a0a3b9a4aad Secret

Created June 9, 2015 09:01
Download ZIP

Revisions

  1. FrankNiemeyer created this gist Jun 9, 2015.
    8 changes: 8 additions & 0 deletions gistfile1.fs
    View file
    Original file line number Diff line number Diff line change
    @@ -0,0 +1,8 @@
    for j = 0 to (m / 4) - 1 do // assume m % 4 = 0
    let i = 4 * j
    // load four x components [x0, x1, x2, x3] into a SIMD register
    let xs = u.x.[i .. i+3]
    let ys = u.y.[i .. i+3]
    let zs = u.z.[i .. i+3]
    // component-wise multiplication and addition -> 4 DPs
    dp.[i .. i+3] <- xs * xs + ys * ys + zs * zs