RE: [Algorithms] non-branching zbuf code tricks?

[<Chr...@Pl...>]

Tom Forsyth wrote:

>> >zbuf[x] = z < zbuf[x] ? z : zbuf[x];
>> [...]
>Indeed - this is just a branch hiding in an expression.

On my platform gcc compiles this to:

lw   v0, 0(a0)
slt  v1,a0,v0
movn v0,a1,v1
sw   v0,0(a0)

which (for once, when it comes to gcc) is optimal code.
Clearly not a branch in sight. Can't a PC do something
similar?

Christer Ericson
SCEA, Santa Monica