[llvm-commits] [llvm] r107852 - in /llvm/trunk: lib/Target/ARM/ARMISelLowering.cpp lib/Target/ARM/ARMISelLowering.h test/CodeGen/ARM/fpcmp-opt.ll

[Evan Cheng]

On Jul 7, 2010, at 10:43 PM, Anton Korobeynikov wrote:

> Hello, Evan
> 
>> Optimize some vfp comparisons to integer ones. This patch implements the simplest case when the following conditions are met:
> Any reasons of doing this? This looks like a pretty expensive change
> and ARM manuals explicitly tell not to do so.

Which manual? Why would it be expensive?

before:
        vldr.32 s0, [r1]
        vldr.32 s1, [r0]
        vcmpe.f32       s1, s0
        vmrs    apsr_nzcv, fpscr
        beq     LBB0_2

after:
	ldr     r1, [r1]
	ldr     r0, [r0]
	cmp     r0, r1
	beq     LBB0_2

On Cortex-a8 vmrs is very expensive. But even on a9 the integer version should be faster.

Evan

> 
> -- 
> With best regards, Anton Korobeynikov
> Faculty of Mathematics and Mechanics, Saint Petersburg State University