wine/vkd3d
0
0
Fork 0
mirror of https://gitlab.winehq.org/wine/vkd3d.git synced 2024-11-21 16:46:41 -08:00
Code Issues Packages Projects Releases Wiki Activity

vkd3d-shader/hlsl: Optimize interstage signatures.

Browse Source
This commit is contained in:
Francisco Casas 2024-09-25 16:48:00 -03:00 committed by Henri Verbeet
parent 88dd082160
commit 51d05c1844
Notes: Henri Verbeet 2024-10-22 20:54:15 +02:00 
Approved-by: Elizabeth Figura (@zfigura)
Approved-by: Henri Verbeet (@hverbeet)
Merge-Request: https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/1193
2 changed files with 25 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
libs/vkd3d-shader/hlsl_codegen.c
View File

@ -5253,7 +5253,7 @@ enum vkd3d_shader_interpolation_mode sm4_get_interpolation_mode(struct hlsl_type
} }
static void allocate_semantic_register(struct hlsl_ctx *ctx, struct hlsl_ir_var *var, static void allocate_semantic_register(struct hlsl_ctx *ctx, struct hlsl_ir_var *var,
struct register_allocator *allocator, bool output, bool is_patch_constant_func) struct register_allocator *allocator, bool output, bool optimize, bool is_patch_constant_func)
{ {
static const char *const shader_names[] = static const char *const shader_names[] =
{ {
@ -5312,6 +5312,13 @@ static void allocate_semantic_register(struct hlsl_ctx *ctx, struct hlsl_ir_var
if ((builtin = sm4_register_from_semantic_name(&version, var->semantic.name, output, &type, &has_idx))) if ((builtin = sm4_register_from_semantic_name(&version, var->semantic.name, output, &type, &has_idx)))
reg = has_idx ? var->semantic.index : 0; reg = has_idx ? var->semantic.index : 0;
if (semantic == VKD3D_SHADER_SV_TESS_FACTOR_TRIINT)
{
/* While SV_InsideTessFactor can be declared as 'float' for "tri"
* domains, it is allocated as if it was 'float[1]'. */
var->force_align = true;
}
} }
if (builtin) if (builtin)
@ -5323,18 +5330,21 @@ static void allocate_semantic_register(struct hlsl_ctx *ctx, struct hlsl_ir_var
{ {
int mode = (ctx->profile->major_version < 4) int mode = (ctx->profile->major_version < 4)
? 0 : sm4_get_interpolation_mode(var->data_type, var->storage_modifiers); ? 0 : sm4_get_interpolation_mode(var->data_type, var->storage_modifiers);
unsigned int reg_size = optimize ? var->data_type->dimx : 4;
var->regs[HLSL_REGSET_NUMERIC] = allocate_register(ctx, allocator, 1, var->regs[HLSL_REGSET_NUMERIC] = allocate_register(ctx, allocator, 1,
UINT_MAX, 4, var->data_type->dimx, mode, var->force_align); UINT_MAX, reg_size, var->data_type->dimx, mode, var->force_align);
TRACE("Allocated %s to %s.\n", var->name, debug_register(output ? 'o' : 'v', TRACE("Allocated %s to %s (mode %d).\n", var->name, debug_register(output ? 'o' : 'v',
var->regs[HLSL_REGSET_NUMERIC], var->data_type)); var->regs[HLSL_REGSET_NUMERIC], var->data_type), mode);
} }
} }
static void allocate_semantic_registers(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry_func) static void allocate_semantic_registers(struct hlsl_ctx *ctx, struct hlsl_ir_function_decl *entry_func)
{ {
struct register_allocator input_allocator = {0}, output_allocator = {0}; struct register_allocator input_allocator = {0}, output_allocator = {0};
bool is_vertex_shader = ctx->profile->type == VKD3D_SHADER_TYPE_VERTEX;
bool is_pixel_shader = ctx->profile->type == VKD3D_SHADER_TYPE_PIXEL;
bool is_patch_constant_func = entry_func == ctx->patch_constant_func; bool is_patch_constant_func = entry_func == ctx->patch_constant_func;
struct hlsl_ir_var *var; struct hlsl_ir_var *var;
@ -5344,9 +5354,9 @@ static void allocate_semantic_registers(struct hlsl_ctx *ctx, struct hlsl_ir_fun
LIST_FOR_EACH_ENTRY(var, &entry_func->extern_vars, struct hlsl_ir_var, extern_entry) LIST_FOR_EACH_ENTRY(var, &entry_func->extern_vars, struct hlsl_ir_var, extern_entry)
{ {
if (var->is_input_semantic) if (var->is_input_semantic)
allocate_semantic_register(ctx, var, &input_allocator, false, is_patch_constant_func); allocate_semantic_register(ctx, var, &input_allocator, false, !is_vertex_shader, is_patch_constant_func);
if (var->is_output_semantic) if (var->is_output_semantic)
allocate_semantic_register(ctx, var, &output_allocator, true, is_patch_constant_func); allocate_semantic_register(ctx, var, &output_allocator, true, !is_pixel_shader, is_patch_constant_func);
} }
vkd3d_free(input_allocator.allocations); vkd3d_free(input_allocator.allocations);

16
tests/hlsl/interface-packing.shader_test
View File

@ -116,14 +116,14 @@ float4 main(data input) : SV_Target
[test] [test]
draw triangle list 3 draw triangle list 3
probe (0, 0) rgba ( 0.0, 0.0, 0.0, 0.0) probe (0, 0) rgba ( 0.0, 0.0, 0.0, 0.0)
todo probe (1, 0) rgba ( 1.0, 12.0, 13.0, 14.0) probe (1, 0) rgba ( 1.0, 12.0, 13.0, 14.0)
todo probe (2, 0) rgba ( 2.0, 15.0, 16.0, 22.0) probe (2, 0) rgba ( 2.0, 15.0, 16.0, 22.0)
todo probe (3, 0) rgba ( 3.0, 4.0, 5.0, 26.0) probe (3, 0) rgba ( 3.0, 4.0, 5.0, 26.0)
todo probe (4, 0) rgba ( 6.0, 17.0, 20.0, 21.0) probe (4, 0) rgba ( 6.0, 17.0, 20.0, 21.0)
todo probe (5, 0) rgba ( 7.0, 18.0, 19.0, 0.0) probe (5, 0) rgba ( 7.0, 18.0, 19.0, 0.0)
todo probe (6, 0) rgba (27.0, 28.0, 0.0, 0.0) probe (6, 0) rgba (27.0, 28.0, 0.0, 0.0)
todo probe (7, 0) rgba ( 8.0, 9.0, 10.0, 11.0) probe (7, 0) rgba ( 8.0, 9.0, 10.0, 11.0)
todo probe (8, 0) rgba (23.0, 24.0, 25.0, 0.0) probe (8, 0) rgba (23.0, 24.0, 25.0, 0.0)
[require] [require]
shader model >= 3.0 shader model >= 3.0