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

radeonsi: store shader variants in an array 

Instead of storing them in a linked list, put them in an array
in si_shader_selector. The keys are also stored separatly, to
avoid pointer chasing when searching a variant in si_shader_select_with_key.

This main point here is to simplify the code by storing everything
in the selector instead of splitting the list storage between the selector
and the shaders; this shouldn't affect performance in a meaningful way.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16273>
This commit is contained in:
Pierre-Eric Pelloux-Prayer 2022-05-02 11:15:44 +02:00 committed by Marge Bot
parent e549b6fe42
commit d4e4aa997a
2 changed files with 40 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
src/gallium/drivers/radeonsi/si_shader.h
View File

@ -473,8 +473,10 @@ struct si_shader_selector {
gl_shader_stage stage;
simple_mtx_t mutex;
struct si_shader *first_variant; /* immutable after the first variant */
struct si_shader *last_variant; /* mutable */
union si_shader_key *keys;
unsigned variants_count;
unsigned variants_max_count;
struct si_shader **variants;
/* The compiled NIR shader without a prolog and/or epilog (not
* uploaded to a buffer object).
@ -817,7 +819,6 @@ struct si_shader {
struct si_shader_selector *selector;
struct si_shader_selector *previous_stage_sel; /* for refcounting */
struct si_shader *next_variant;
struct si_shader_part *prolog;
struct si_shader *previous_stage; /* for GFX9 */

64
src/gallium/drivers/radeonsi/si_state_shaders.cpp
View File

@ -2559,7 +2559,7 @@ static int si_shader_select_with_key(struct si_context *sctx, struct si_shader_c
struct si_shader_selector *sel = state->cso;
struct si_shader_selector *previous_stage_sel = NULL;
struct si_shader *current = state->current;
struct si_shader *iter, *shader = NULL;
struct si_shader *shader = NULL;
const SHADER_KEY_TYPE *zeroed_key = (SHADER_KEY_TYPE*)&zeroed;
/* "opt" must be the last field and "inlined_uniform_values" must be the last field inside opt.
@ -2620,10 +2620,13 @@ current_not_ready:
const int max_inline_uniforms_variants = 5;
/* Find the shader variant. */
for (iter = sel->first_variant; iter; iter = iter->next_variant) {
const SHADER_KEY_TYPE *iter_key = (const SHADER_KEY_TYPE *)&iter->key;
const unsigned cnt = sel->variants_count;
for (unsigned i = 0; i < cnt; i++) {
const SHADER_KEY_TYPE *iter_key = (const SHADER_KEY_TYPE *)&sel->keys[i];
if (memcmp(iter_key, key, key_size_no_uniforms) == 0) {
struct si_shader *iter = sel->variants[i];
/* Check the inlined uniform values separately, and count
* the number of variants based on them.
*/
@ -2662,7 +2665,7 @@ current_not_ready:
return -1; /* skip the draw call */
}
state->current = iter;
state->current = sel->variants[i];
return 0;
}
}
@ -2747,6 +2750,14 @@ current_not_ready:
}
}
if (sel->variants_count == sel->variants_max_count) {
sel->variants_max_count += 2;
sel->variants = (struct si_shader**)
realloc(sel->variants, sel->variants_max_count * sizeof(struct si_shader*));
sel->keys = (union si_shader_key*)
realloc(sel->keys, sel->variants_max_count * sizeof(union si_shader_key));
}
/* Keep the reference to the 1st shader of merged shaders, so that
* Gallium can't destroy it before we destroy the 2nd shader.
*
@ -2771,13 +2782,9 @@ current_not_ready:
/* Add only after the ready fence was reset, to guard against a
* race with si_bind_XX_shader. */
if (!sel->last_variant) {
sel->first_variant = shader;
sel->last_variant = shader;
} else {
sel->last_variant->next_variant = shader;
sel->last_variant = shader;
}
sel->variants[sel->variants_count] = shader;
sel->keys[sel->variants_count] = shader->key;
sel->variants_count++;
/* Use the default (unoptimized) shader for now. */
key = use_local_key_copy(key, &local_key, key_size);
@ -2793,13 +2800,9 @@ current_not_ready:
/* Reset the fence before adding to the variant list. */
util_queue_fence_reset(&shader->ready);
if (!sel->last_variant) {
sel->first_variant = shader;
sel->last_variant = shader;
} else {
sel->last_variant->next_variant = shader;
sel->last_variant = shader;
}
sel->variants[sel->variants_count] = shader;
sel->keys[sel->variants_count] = shader->key;
sel->variants_count++;
simple_mtx_unlock(&sel->mutex);
@ -3092,6 +3095,11 @@ static void *si_create_shader_selector(struct pipe_context *ctx,
sel->screen = sscreen;
sel->compiler_ctx_state.debug = sctx->debug;
sel->compiler_ctx_state.is_debug_context = sctx->is_debug;
sel->variants_max_count = 2;
sel->keys = (union si_shader_key *)
realloc(NULL, sel->variants_max_count * sizeof(union si_shader_key));
sel->variants = (struct si_shader **)
realloc(NULL, sel->variants_max_count * sizeof(struct si_shader *));
if (state->type == PIPE_SHADER_IR_TGSI) {
sel->nir = tgsi_to_nir(state->tokens, ctx->screen, true);
@ -3292,7 +3300,7 @@ static void si_bind_vs_shader(struct pipe_context *ctx, void *state)
return;
sctx->shader.vs.cso = sel;
sctx->shader.vs.current = sel ? sel->first_variant : NULL;
sctx->shader.vs.current = (sel && sel->variants_count) ? sel->variants[0] : NULL;
sctx->num_vs_blit_sgprs = sel ? sel->info.base.vs.blit_sgprs_amd : 0;
sctx->vs_uses_draw_id = sel ? sel->info.uses_drawid : false;
sctx->fixed_func_tcs_shader.key.ge.mono.u.ff_tcs_inputs_to_copy = sel ? sel->info.outputs_written : 0;
@ -3381,7 +3389,7 @@ static void si_bind_gs_shader(struct pipe_context *ctx, void *state)
return;
sctx->shader.gs.cso = sel;
sctx->shader.gs.current = sel ? sel->first_variant : NULL;
sctx->shader.gs.current = (sel && sel->variants_count) ? sel->variants[0] : NULL;
sctx->ia_multi_vgt_param_key.u.uses_gs = sel != NULL;
si_update_common_shader_state(sctx, sel, PIPE_SHADER_GEOMETRY);
@ -3412,7 +3420,7 @@ static void si_bind_tcs_shader(struct pipe_context *ctx, void *state)
return;
sctx->shader.tcs.cso = sel;
sctx->shader.tcs.current = sel ? sel->first_variant : NULL;
sctx->shader.tcs.current = (sel && sel->variants_count) ? sel->variants[0] : NULL;
sctx->shader.tcs.key.ge.part.tcs.epilog.invoc0_tess_factors_are_def =
sel ? sel->info.tessfactors_are_def_in_all_invocs : 0;
si_update_tess_uses_prim_id(sctx);
@ -3435,7 +3443,7 @@ static void si_bind_tes_shader(struct pipe_context *ctx, void *state)
return;
sctx->shader.tes.cso = sel;
sctx->shader.tes.current = sel ? sel->first_variant : NULL;
sctx->shader.tes.current = (sel && sel->variants_count) ? sel->variants[0] : NULL;
sctx->ia_multi_vgt_param_key.u.uses_tess = sel != NULL;
si_update_tess_uses_prim_id(sctx);
@ -3509,7 +3517,7 @@ static void si_bind_ps_shader(struct pipe_context *ctx, void *state)
return;
sctx->shader.ps.cso = sel;
sctx->shader.ps.current = sel ? sel->first_variant : NULL;
sctx->shader.ps.current = (sel && sel->variants_count) ? sel->variants[0] : NULL;
si_update_common_shader_state(sctx, sel, PIPE_SHADER_FRAGMENT);
if (sel) {
@ -3614,7 +3622,6 @@ static void si_destroy_shader_selector(struct pipe_context *ctx, void *cso)
{
struct si_context *sctx = (struct si_context *)ctx;
struct si_shader_selector *sel = (struct si_shader_selector *)cso;
struct si_shader *p = sel->first_variant, *c;
enum pipe_shader_type type = pipe_shader_type_from_mesa(sel->stage);
util_queue_drop_job(&sctx->screen->shader_compiler_queue, &sel->ready);
@ -3624,10 +3631,8 @@ static void si_destroy_shader_selector(struct pipe_context *ctx, void *cso)
sctx->shaders[type].current = NULL;
}
while (p) {
c = p->next_variant;
si_delete_shader(sctx, p);
p = c;
for (unsigned i = 0; i < sel->variants_count; i++) {
si_delete_shader(sctx, sel->variants[i]);
}
if (sel->main_shader_part)
@ -3639,6 +3644,9 @@ static void si_destroy_shader_selector(struct pipe_context *ctx, void *cso)
if (sel->main_shader_part_ngg)
si_delete_shader(sctx, sel->main_shader_part_ngg);
free(sel->keys);
free(sel->variants);
util_queue_fence_destroy(&sel->ready);
simple_mtx_destroy(&sel->mutex);
ralloc_free(sel->nir);