AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

v3d/compiler: use mask for 16bit and 32bit return values

Browse Source 
There are only ever two possibilities here so lets use a mask.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34742>
This commit is contained in:
Ella Stanforth
2025-04-29 17:29:55 +01:00
committed by Marge Bot
parent bb07364c54
commit 9a71e6dcc2
6 changed files with 41 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/broadcom/compiler/v3d_compiler.h
+5 -3
View File
@@ -408,9 +408,11 @@ static inline uint8_t v3d_slot_get_component(struct v3d_varying_slot slot)
}
struct v3d_key {
struct {
uint8_t return_size;
} sampler[V3D_MAX_TEXTURE_SAMPLERS];
/* Mask of sampler return sizes.
* 0 is 16bit
* 1 is 32bit
*/
uint32_t sampler_is_32b;
uint8_t ucp_enables;
bool is_last_geometry_stage;
src/broadcom/compiler/v3d_tex.c
+1 -1
View File
@@ -239,7 +239,7 @@ v3d_vir_emit_tex(struct v3d_compile *c, nir_tex_instr *instr)
* parameter if the output is 32 bit
*/
bool output_type_32_bit =
c->key->sampler[sampler_idx].return_size == 32;
(c->key->sampler_is_32b & (1 << sampler_idx));
struct V3D42_TMU_CONFIG_PARAMETER_0 p0_unpacked = {
};
src/broadcom/compiler/vir.c
+2 -2
View File
@@ -601,8 +601,8 @@ lower_tex_packing_cb(const nir_tex_instr *tex, const void *data)
int sampler_index = nir_tex_instr_need_sampler(tex) ?
tex->sampler_index : tex->backend_flags;
return c->key->sampler[sampler_index].return_size == 16 ?
nir_lower_tex_packing_16 : nir_lower_tex_packing_none;
return (c->key->sampler_is_32b & (1 << sampler_index)) ?
nir_lower_tex_packing_none : nir_lower_tex_packing_16;
}
static bool
src/broadcom/vulkan/v3dv_pipeline.c
+28 -30
View File
@@ -396,11 +396,10 @@ descriptor_map_add(struct v3dv_descriptor_map *map,
int array_index,
int array_size,
int start_index,
uint8_t return_size,
bool sampler_is_32b,
uint8_t plane)
{
assert(array_index < array_size);
assert(return_size == 16 || return_size == 32);
unsigned index = start_index;
for (; index < map->num_desc; index++) {
@@ -410,14 +409,13 @@ descriptor_map_add(struct v3dv_descriptor_map *map,
array_index == map->array_index[index] &&
plane == map->plane[index]) {
assert(array_size == map->array_size[index]);
if (return_size != map->return_size[index]) {
/* It the return_size is different it means that the same sampler
* was used for operations with different precision
* requirement. In this case we need to ensure that we use the
* larger one.
*/
map->return_size[index] = 32;
}
/* It the return_size is different it means that the same sampler
* was used for operations with different precision
* requirement. In this case we need to ensure that we use the
* larger one.
*/
if (sampler_is_32b != map->sampler_is_32b[index])
map->sampler_is_32b[index] = true;
return index;
} else if (!map->used[index]) {
break;
@@ -432,7 +430,7 @@ descriptor_map_add(struct v3dv_descriptor_map *map,
map->binding[index] = binding;
map->array_index[index] = array_index;
map->array_size[index] = array_size;
map->return_size[index] = return_size;
map->sampler_is_32b[index] = sampler_is_32b;
map->plane[index] = plane;
map->num_desc = MAX2(map->num_desc, index + 1);
@@ -541,7 +539,7 @@ lower_vulkan_resource_index(nir_builder *b,
const_val->u32,
binding_layout->array_size,
start_index,
32 /* return_size: doesn't really apply for this case */,
true /* sampler_is_32b: doesn't really apply for this case */,
0);
break;
}
@@ -570,10 +568,10 @@ tex_instr_get_and_remove_plane_src(nir_tex_instr *tex)
return plane;
}
/* Returns return_size, so it could be used for the case of not having a
* sampler object
/* Returns true if we need 32bit, so we know what size to use when we do not
* have a sampler object
*/
static uint8_t
static bool
lower_tex_src(nir_builder *b,
nir_tex_instr *instr,
unsigned src_idx,
@@ -644,13 +642,13 @@ lower_tex_src(nir_builder *b,
struct v3dv_descriptor_set_binding_layout *binding_layout =
&set_layout->binding[binding];
uint8_t return_size;
uint8_t sampler_is_32b;
if (V3D_DBG(TMU_16BIT))
return_size = 16;
sampler_is_32b = false;
else if (V3D_DBG(TMU_32BIT))
return_size = 32;
sampler_is_32b = true;
else
return_size = relaxed_precision ? 16 : 32;
sampler_is_32b = !relaxed_precision;
struct v3dv_descriptor_map *map =
pipeline_get_descriptor_map(state->pipeline, binding_layout->type,
@@ -662,7 +660,7 @@ lower_tex_src(nir_builder *b,
base_index,
binding_layout->array_size,
0,
return_size,
sampler_is_32b,
plane);
if (is_sampler)
@@ -670,7 +668,7 @@ lower_tex_src(nir_builder *b,
else
instr->texture_index = desc_index;
return return_size;
return sampler_is_32b;
}
static bool
@@ -678,13 +676,13 @@ lower_sampler(nir_builder *b,
nir_tex_instr *instr,
struct lower_pipeline_layout_state *state)
{
uint8_t return_size = 0;
bool sampler_is_32b = false;
int texture_idx =
nir_tex_instr_src_index(instr, nir_tex_src_texture_deref);
if (texture_idx >= 0)
return_size = lower_tex_src(b, instr, texture_idx, state);
sampler_is_32b = lower_tex_src(b, instr, texture_idx, state);
int sampler_idx =
nir_tex_instr_src_index(instr, nir_tex_src_sampler_deref);
@@ -702,8 +700,8 @@ lower_sampler(nir_builder *b,
*/
if (sampler_idx < 0) {
state->needs_default_sampler_state = true;
instr->backend_flags = return_size == 16 ?
V3DV_NO_SAMPLER_16BIT_IDX : V3DV_NO_SAMPLER_32BIT_IDX;
instr->backend_flags = sampler_is_32b ?
V3DV_NO_SAMPLER_32BIT_IDX : V3DV_NO_SAMPLER_16BIT_IDX;
}
return true;
@@ -769,7 +767,7 @@ lower_image_deref(nir_builder *b,
base_index,
binding_layout->array_size,
0,
32 /* return_size: doesn't apply for textures */,
true /* return_size: doesn't apply for textures */,
0);
/* Note: we don't need to do anything here in relation to the precision and
@@ -965,8 +963,8 @@ pipeline_populate_v3d_key(struct v3d_key *key,
for (uint32_t sampler_idx = 0; sampler_idx < sampler_map->num_desc;
sampler_idx++) {
key->sampler[sampler_idx].return_size =
sampler_map->return_size[sampler_idx];
if (sampler_map->sampler_is_32b[sampler_idx])
key->sampler_is_32b |= 1 << sampler_idx;
}
switch (p_stage->stage) {
@@ -1767,10 +1765,10 @@ pipeline_lower_nir(struct v3dv_pipeline *pipeline,
pipeline->shared_data->maps[p_stage->stage];
UNUSED unsigned index;
index = descriptor_map_add(&maps->sampler_map, -1, -1, -1, 0, 0, 16, 0);
index = descriptor_map_add(&maps->sampler_map, -1, -1, -1, 0, 0, false, 0);
assert(index == V3DV_NO_SAMPLER_16BIT_IDX);
index = descriptor_map_add(&maps->sampler_map, -2, -2, -2, 0, 0, 32, 0);
index = descriptor_map_add(&maps->sampler_map, -2, -2, -2, 0, 0, true, 0);
assert(index == V3DV_NO_SAMPLER_32BIT_IDX);
/* Apply the actual pipeline layout to UBOs, SSBOs, and textures */
src/broadcom/vulkan/v3dv_private.h
+1 -1
View File
@@ -2175,7 +2175,7 @@ struct v3dv_descriptor_map {
/* NOTE: the following is only for sampler, but this is the easier place to
* put it.
*/
uint8_t return_size[DESCRIPTOR_MAP_SIZE];
bool sampler_is_32b[DESCRIPTOR_MAP_SIZE];
};
struct v3dv_sampler {
src/gallium/drivers/v3d/v3d_program.c
+4 -25
View File
@@ -43,10 +43,6 @@ v3d_get_compiled_shader(struct v3d_context *v3d,
struct v3d_key *key, size_t key_size,
struct v3d_uncompiled_shader *uncompiled);
static void
v3d_setup_shared_precompile_key(struct v3d_uncompiled_shader *uncompiled,
struct v3d_key *key);
static gl_varying_slot
v3d_get_slot_for_driver_location(nir_shader *s, uint32_t driver_location)
{
@@ -238,15 +234,12 @@ v3d_shader_precompile(struct v3d_context *v3d,
key.logicop_func = PIPE_LOGICOP_COPY;
v3d_setup_shared_precompile_key(so, &key.base);
v3d_get_compiled_shader(v3d, &key.base, sizeof(key), so);
} else if (s->info.stage == MESA_SHADER_GEOMETRY) {
struct v3d_gs_key key = {
.base.is_last_geometry_stage = true,
};
v3d_setup_shared_precompile_key(so, &key.base);
precompile_all_outputs(s,
key.used_outputs,
&key.num_used_outputs);
@@ -268,8 +261,6 @@ v3d_shader_precompile(struct v3d_context *v3d,
.base.is_last_geometry_stage = true,
};
v3d_setup_shared_precompile_key(so, &key.base);
precompile_all_outputs(s,
key.used_outputs,
&key.num_used_outputs);
@@ -288,7 +279,6 @@ v3d_shader_precompile(struct v3d_context *v3d,
} else {
assert(s->info.stage == MESA_SHADER_COMPUTE);
struct v3d_key key = { 0 };
v3d_setup_shared_precompile_key(so, &key);
v3d_get_compiled_shader(v3d, &key, sizeof(key), so);
}
}
@@ -582,23 +572,12 @@ v3d_setup_shared_key(struct v3d_context *v3d, struct v3d_key *key,
if (!sampler)
continue;
key->sampler[i].return_size =
uint8_t return_size =
v3d_get_tex_return_size(devinfo, sampler->format);
}
}
assert(return_size == 16 || return_size == 32);
static void
v3d_setup_shared_precompile_key(struct v3d_uncompiled_shader *uncompiled,
struct v3d_key *key)
{
nir_shader *s = uncompiled->base.ir.nir;
/* Note that below we access they key's texture and sampler fields
* using the same index. On OpenGL they are the same (they are
* combined)
*/
for (int i = 0; i < s->info.num_textures; i++) {
key->sampler[i].return_size = 16;
if (return_size == 32)
key->sampler_is_32b |= (1 << i);
}
}