Go back to running both sides of 'if' statements with masking and without
branching if we can determine that the code is relatively simple (as per
the simple cost model), and is safe to run even if the mask is 'all off'.
This gives a bit of a performance improvement for some of the examples
(most notably, the ray tracer), and is the code that one wants generated
in this case anyhow.
This is currently only used to decide whether it's worth doing an
"are all lanes running" check at the start of functions--for small
functions, it's not worth the overhead.
The cost is estimated relatively early in compilation (e.g. before
we know if an array access is a scatter/gather or not, before
constant folding, etc.), so there are many known shortcomings.
Fixes issue #73. Previously, if we had e.g. an int16 type that was being shifted
left by 1, then the constant integer 1 would come in as an int32, we'd convert
the int16 to an int32, and then we'd do the shift. Now, for shifts, the type
of the expression is always the same as the type of the value being shifted.
This commit adds support for swizzles like "foo.zy" (if "foo" is,
for example, a float<3> type) as rvalues. (Still need support for
swizzles as lvalues.)
