This is no longer true. I added SIMD to SwiftGL. Here's one million mat4x4
multiplications:
SIMD module direct : 1.489s
SwiftGL + SIMD : 1.487s
SwiftGL plain Swift : 6.492s <- you'll get this on Linux
Everything will continue to work on Linux. It's just slower. Better than
nothing.
-david
···
On Tue, Jan 5, 2016 at 12:52 PM, David Turnbull <dturnbull@gmail.com> wrote:
It's not as fast as packed SIMD but it's still really fast. As fast as C.