Commit Graph

474 Commits

Author SHA1 Message Date
Francisco Casas
7cb1970952 vkd3d-shader/hlsl: Don't cast all expressions to float.
Instead, allow half, bool, int, and uint, expressions that map to the
same float op to reach sm1_generate_vsir_instr_expr().
2025-03-03 17:26:15 +01:00
Elizabeth Figura
2989373212 vkd3d-shader/hlsl: Add a hlsl_block_add_binary_expr() helper. 2025-02-24 16:19:07 +01:00
Elizabeth Figura
858b6a3e0b vkd3d-shader/hlsl: Add a hlsl_block_add_cast() helper. 2025-02-24 16:19:07 +01:00
Elizabeth Figura
2e09257d94 vkd3d-shader/hlsl: Add a hlsl_block_add_unary_expr() helper. 2025-02-24 16:19:07 +01:00
Elizabeth Figura
b7ea23303e vkd3d-shader/hlsl: Add a hlsl_block_add_float_constant() helper. 2025-02-24 16:19:07 +01:00
Elizabeth Figura
d5a2ff5c12 vkd3d-shader/hlsl: Add a hlsl_block_add_int_constant() helper. 2025-02-20 15:56:31 +01:00
Elizabeth Figura
992d20def3 vkd3d-shader/hlsl: Add a hlsl_block_add_uint_constant() helper. 2025-02-20 15:50:13 +01:00
Francisco Casas
321fda9c26 vkd3d-shader/hlsl: Only use the temp copy for variables that are written.
This can save a significant amount of temp registers because it allows to
avoid referencing the temp (and having to store it) when not needed.

For instance, this patch lowers the number of required temps for the
following ps_2_0 shader from 24 to 19:

    int i;
    float3x3 mats[4];

    float4 main() : sv_target
    {
        return mul(mats[i], float3(1, 2, 3)).xyzz;
    }

Also, it is needed for SM1 vertex shader relative addressing since
non-constant loads are required to be directly on the uniform ('c'
registers) instead of the temp, and non-constant loads cannot be
transformed by copy propagation.
2025-02-20 15:44:09 +01:00
Shaun Ren
f5d216835a vkd3d-shader/hlsl: Add an "is_patch_constant_func" field to struct hlsl_ctx.
In anticipation of the need for is_patch_constant_func in
sm4_generate_vsir_reg_from_deref(), in order to generate vsir
registers from InputPatch/OutputPatch dereferences.
2025-02-03 16:00:38 +01:00
Elizabeth Figura
fbd17266cf vkd3d-shader/hlsl: Do not abort on variable redefinition.
There is no harm in defining two variables with the same name.
2025-01-29 17:58:00 +01:00
Elizabeth Figura
2863d86bcc vkd3d-shader/hlsl: Do not define a separate scope for function parameters.
Remove the no longer needed code to detect redefinition in this case.
2025-01-29 17:58:00 +01:00
Shaun Ren
f064a4022a vkd3d-shader/hlsl: Validate and record InputPatch/OutputPatch types. 2025-01-29 17:45:46 +01:00
Shaun Ren
f4d5e05d96 vkd3d-shader/hlsl: Parse InputPatch and OutputPatch objects. 2025-01-29 17:41:51 +01:00
Shaun Ren
802d7317cb vkd3d-shader/hlsl: Implement the InterlockedXor() intrinsic. 2025-01-20 15:42:45 +01:00
Shaun Ren
cbd504e888 vkd3d-shader/hlsl: Implement the InterlockedOr() intrinsic. 2025-01-20 15:42:45 +01:00
Shaun Ren
5b9634a7b7 vkd3d-shader/hlsl: Implement the InterlockedMin() intrinsic. 2025-01-20 15:42:43 +01:00
Shaun Ren
63fbe161f2 vkd3d-shader/hlsl: Implement the InterlockedMax() intrinsic. 2025-01-20 15:42:07 +01:00
Shaun Ren
b447fdce51 vkd3d-shader/hlsl: Implement the InterlockedExchange() intrinsic. 2025-01-20 15:41:07 +01:00
Shaun Ren
22ab08f4d8 vkd3d-shader/hlsl: Implement the InterlockedCompareExchange() intrinsic. 2025-01-20 15:41:07 +01:00
Shaun Ren
3b19a4aaf3 vkd3d-shader/hlsl: Implement the InterlockedAnd() intrinsic. 2025-01-20 15:41:07 +01:00
Shaun Ren
660a71ef22 vkd3d-shader/hlsl: Implement the InterlockedAdd() intrinsic. 2025-01-20 15:40:34 +01:00
Nikolay Sivov
a082daeb56 vkd3d-shader/hlsl: Implement the isinf() intrinsic.
Signed-off-by: Nikolay Sivov <nsivov@codeweavers.com>
2025-01-16 19:25:02 +01:00
Henri Verbeet
4227858cfe vkd3d-shader/hlsl: Move RDEF generation to hlsl_codegen.c. 2025-01-10 20:03:14 +01:00
Elizabeth Figura
29bac62ba2 vkd3d-shader/hlsl: Move the "dimx" and "dimy" fields to the type-specific union. 2025-01-10 19:38:13 +01:00
Henri Verbeet
0236c480e3 vkd3d-shader/hlsl: Make sm4_get_interpolation_mode() static. 2024-12-19 21:09:32 +01:00