SSE Math

We'd like to support SSE math operations on vectors of floats, etc.

Some places to look at for inspiration:

• vectorial
• Dart's SIMD APIs

One of the first steps to take will be to add a new 128 raw data type.

This will involve code in:

• sources/dfmc/back-end/raw-type-descriptors.dylan
• sources/dfmc/modeling/objects.dylan
• sources/dfmc/modeling/primitives.dylan
• sources/dfmc/modeling/raw-types.dylan

Other work will have to done in the backends, such as:

• sources/dfmc/llvm-back-end/llvm-back-end-types.dylan
• sources/dfmc/llvm-back-end/llvm-emit-object.dylan
• sources/dfmc/llvm-back-end/llvm-primitives-alloc.dylan
• sources/dfmc/llvm-back-end/llvm-primitives-float.dylan

You'll probably also want to update the C-FFI:

• sources/lib/c-ffi/pointer-ref.dylan

You can take guidance from the <raw-single-float> and <raw-double-float> types.

It is best to just do this for the LLVM and C backends initially. If it is done for HARP, it can be done in a slow way by compiling some C code into the run-time. HARP isn't important as we hope to phase it out entirely soon in favor of using LLVM.