[Mesa-dev] [PATCH 01/16] st/nine: Handle special LIT case

[Ilia Mirkin]

On Fri, Apr 24, 2015 at 5:08 PM, Axel Davy <axel.davy at ens.fr> wrote:
> Le 24/04/2015 23:04, Ilia Mirkin a écrit :
>>
>> On Fri, Apr 24, 2015 at 4:09 PM, Axel Davy <axel.davy at ens.fr> wrote:
>>>
>>> Signed-off-by: Axel Davy <axel.davy at ens.fr>
>>> ---
>>>   src/gallium/state_trackers/nine/nine_shader.c | 19 ++++++++++++++++++-
>>>   1 file changed, 18 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/src/gallium/state_trackers/nine/nine_shader.c
>>> b/src/gallium/state_trackers/nine/nine_shader.c
>>> index 2ba625e..0fd3d37 100644
>>> --- a/src/gallium/state_trackers/nine/nine_shader.c
>>> +++ b/src/gallium/state_trackers/nine/nine_shader.c
>>> @@ -2040,6 +2040,23 @@ DECL_SPECIAL(LOG)
>>>       return D3D_OK;
>>>   }
>>>
>>> +DECL_SPECIAL(LIT)
>>> +{
>>> +    struct ureg_program *ureg = tx->ureg;
>>> +    struct ureg_dst tmp = tx_scratch(tx);
>>> +    struct ureg_dst dst = tx_dst_param(tx, &tx->insn.dst[0]);
>>> +    struct ureg_src src = tx_src_param(tx, &tx->insn.src[0]);
>>> +    ureg_LIT(ureg, tmp, src);
>>> +    /* d3d9 LIT is the same than gallium LIT. One difference is that
>>> d3d9
>>> +     * states that dst.z is 0 when src.y <= 0. Gallium definition can
>>> assign
>>> +     * it 0^0 if src.w=0, which value is driver dependent. */
>>> +    ureg_CMP(ureg, ureg_writemask(dst, TGSI_WRITEMASK_Z),
>>> +             ureg_negate(ureg_scalar(src, TGSI_SWIZZLE_Y)),
>>> +             ureg_src(tmp), ureg_imm1f(ureg, 0.0f));
>>> +    ureg_MOV(ureg, ureg_writemask(dst, TGSI_WRITEMASK_XYW),
>>> ureg_src(tmp));
>>
>> Why use a separate temp? Couldn't you just
>>
>> ureg_LIT(ureg, dst, src);
>> ureg_CMP(ureg, writemask(dst, Z), -src.y, ureg_src(dst), ureg_imm1f(0.0))
>> ?
>>
> Isn't there a restriction in the case dst is an output register of the
> shader ? Reading from it would be forbidden.
> Also likely it is easier to optimiser for compilers with the temp version.

Ah yes. Usually that doesn't come in st/mesa, but I guess it can
happen with d3d9. OK.

Reviewed-by: Ilia Mirkin <imirkin at alum.mit.edu>