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

st/mesa: add VAO fast path C++ template variants for st_update_array callback

Browse Source 
This way we execute 1 half of setup_arrays with the fast path enabled,
and the other half with the fast path disabled, so it's not that much
of code duplication, and it will facilitate further optimizations.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27731>
This commit is contained in:
Marek Olšák
2024-01-04 20:49:38 -05:00
committed by Marge Bot
parent 7a5ddd29c2
commit 045b1cda57
3 changed files with 53 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/mesa/state_tracker/st_atom.h
+1 -1
View File
@@ -63,7 +63,7 @@ st_setup_current_user(struct st_context *st,
struct pipe_vertex_buffer *vbuffer, unsigned *num_vbuffers);
void
st_update_array_with_popcnt(struct st_context *st);
st_init_update_array(struct st_context *st);
struct pipe_vertex_state *
st_create_gallium_vertex_state(struct gl_context *ctx,
src/mesa/state_tracker/st_atom_array.cpp
+51 -19
View File
@@ -41,6 +41,7 @@
#include "st_program.h"
#include "cso_cache/cso_context.h"
#include "util/u_cpu_detect.h"
#include "util/u_math.h"
#include "util/u_upload_mgr.h"
#include "main/bufferobj.h"
@@ -48,6 +49,11 @@
#include "main/varray.h"
#include "main/arrayobj.h"
enum st_use_vao_fast_path {
VAO_FAST_PATH_OFF, /* more complicated version (slower) */
VAO_FAST_PATH_ON, /* always works (faster) */
};
enum st_update_velems {
UPDATE_VELEMS_OFF, /* specialized version (faster) */
UPDATE_VELEMS_ON, /* always works */
@@ -75,18 +81,19 @@ init_velement(struct pipe_vertex_element *velements,
/* ALWAYS_INLINE helps the compiler realize that most of the parameters are
* on the stack.
*/
template<util_popcnt POPCNT, st_update_velems UPDATE_VELEMS> void ALWAYS_INLINE
template<util_popcnt POPCNT,
st_use_vao_fast_path USE_VAO_FAST_PATH,
st_update_velems UPDATE_VELEMS> void ALWAYS_INLINE
setup_arrays(struct gl_context *ctx,
const struct gl_vertex_array_object *vao,
const GLbitfield dual_slot_inputs,
const GLbitfield inputs_read,
GLbitfield mask,
struct cso_velems_state *velements,
struct pipe_vertex_buffer *vbuffer, unsigned *num_vbuffers,
bool use_vao_fast_path)
struct pipe_vertex_buffer *vbuffer, unsigned *num_vbuffers)
{
/* Set up enabled vertex arrays. */
if (use_vao_fast_path) {
if (USE_VAO_FAST_PATH) {
const GLubyte *attribute_map =
_mesa_vao_attribute_map[vao->_AttributeMapMode];
@@ -124,6 +131,11 @@ setup_arrays(struct gl_context *ctx,
return;
}
/* The slow path needs more fields initialized, which is not done if it's
* disabled.
*/
assert(!ctx->Const.UseVAOFastPath || vao->SharedAndImmutable);
while (mask) {
/* The attribute index to start pulling a binding */
const gl_vert_attrib i = (gl_vert_attrib)(ffs(mask) - 1);
@@ -180,11 +192,11 @@ st_setup_arrays(struct st_context *st,
struct gl_context *ctx = st->ctx;
GLbitfield enabled_arrays = _mesa_get_enabled_vertex_arrays(ctx);
setup_arrays<POPCNT_NO, UPDATE_VELEMS_ON>
setup_arrays<POPCNT_NO, VAO_FAST_PATH_ON, UPDATE_VELEMS_ON>
(ctx, ctx->Array._DrawVAO, vp->Base.DualSlotInputs,
vp_variant->vert_attrib_mask,
vp_variant->vert_attrib_mask & enabled_arrays,
velements, vbuffer, num_vbuffers, ctx->Const.UseVAOFastPath);
velements, vbuffer, num_vbuffers);
}
/* ALWAYS_INLINE helps the compiler realize that most of the parameters are
@@ -295,7 +307,9 @@ st_setup_current_user(struct st_context *st,
}
}
template<util_popcnt POPCNT, st_update_velems UPDATE_VELEMS> void ALWAYS_INLINE
template<util_popcnt POPCNT,
st_use_vao_fast_path USE_VAO_FAST_PATH,
st_update_velems UPDATE_VELEMS> void ALWAYS_INLINE
st_update_array_templ(struct st_context *st,
const GLbitfield enabled_arrays,
const GLbitfield enabled_user_arrays,
@@ -322,10 +336,9 @@ st_update_array_templ(struct st_context *st,
/* ST_NEW_VERTEX_ARRAYS */
/* Setup arrays */
setup_arrays<POPCNT, UPDATE_VELEMS>
setup_arrays<POPCNT, USE_VAO_FAST_PATH, UPDATE_VELEMS>
(ctx, ctx->Array._DrawVAO, dual_slot_inputs, inputs_read,
inputs_read & enabled_arrays, &velements, vbuffer, &num_vbuffers,
ctx->Const.UseVAOFastPath);
inputs_read & enabled_arrays, &velements, vbuffer, &num_vbuffers);
/* _NEW_CURRENT_ATTRIB */
/* Setup zero-stride attribs. */
@@ -352,7 +365,8 @@ st_update_array_templ(struct st_context *st,
}
}
template<util_popcnt POPCNT> void ALWAYS_INLINE
template<util_popcnt POPCNT,
st_use_vao_fast_path USE_VAO_FAST_PATH> void ALWAYS_INLINE
st_update_array_impl(struct st_context *st)
{
struct gl_context *ctx = st->ctx;
@@ -364,7 +378,7 @@ st_update_array_impl(struct st_context *st)
assert(vao->_EnabledWithMapMode ==
_mesa_vao_enable_to_vp_inputs(vao->_AttributeMapMode, vao->Enabled));
if (!ctx->Const.UseVAOFastPath && !vao->SharedAndImmutable)
if (!USE_VAO_FAST_PATH && !vao->SharedAndImmutable)
_mesa_update_vao_derived_arrays(ctx, vao, false);
_mesa_get_derived_vao_masks(ctx, enabled_arrays, &enabled_user_arrays,
@@ -377,24 +391,39 @@ st_update_array_impl(struct st_context *st)
if (ctx->Array.NewVertexElements ||
st->uses_user_vertex_buffers !=
!!(st->vp_variant->vert_attrib_mask & enabled_user_arrays)) {
st_update_array_templ<POPCNT, UPDATE_VELEMS_ON>
st_update_array_templ<POPCNT, USE_VAO_FAST_PATH, UPDATE_VELEMS_ON>
(st, enabled_arrays, enabled_user_arrays, nonzero_divisor_arrays);
} else {
st_update_array_templ<POPCNT, UPDATE_VELEMS_OFF>
st_update_array_templ<POPCNT, USE_VAO_FAST_PATH, UPDATE_VELEMS_OFF>
(st, enabled_arrays, enabled_user_arrays, nonzero_divisor_arrays);
}
}
/* The default callback that must be present before st_init_update_array
* selects the driver-dependent variant.
*/
void
st_update_array(struct st_context *st)
{
st_update_array_impl<POPCNT_NO>(st);
unreachable("st_init_update_array not called");
}
void
st_update_array_with_popcnt(struct st_context *st)
st_init_update_array(struct st_context *st)
{
st_update_array_impl<POPCNT_YES>(st);
st_update_func_t *func = &st->update_functions[ST_NEW_VERTEX_ARRAYS_INDEX];
if (util_get_cpu_caps()->has_popcnt) {
if (st->ctx->Const.UseVAOFastPath)
*func = st_update_array_impl<POPCNT_YES, VAO_FAST_PATH_ON>;
else
*func = st_update_array_impl<POPCNT_YES, VAO_FAST_PATH_OFF>;
} else {
if (st->ctx->Const.UseVAOFastPath)
*func = st_update_array_impl<POPCNT_NO, VAO_FAST_PATH_ON>;
else
*func = st_update_array_impl<POPCNT_NO, VAO_FAST_PATH_OFF>;
}
}
struct pipe_vertex_state *
@@ -410,9 +439,12 @@ st_create_gallium_vertex_state(struct gl_context *ctx,
unsigned num_vbuffers = 0;
struct cso_velems_state velements;
setup_arrays<POPCNT_NO, UPDATE_VELEMS_ON>
/* This should use the slow path because there is only 1 interleaved
* vertex buffers.
*/
setup_arrays<POPCNT_NO, VAO_FAST_PATH_OFF, UPDATE_VELEMS_ON>
(ctx, vao, dual_slot_inputs, inputs_read, inputs_read, &velements,
vbuffer, &num_vbuffers, false);
vbuffer, &num_vbuffers);
if (num_vbuffers != 1) {
assert(!"this should never happen with display lists");
src/mesa/state_tracker/st_context.c
+1 -5
View File
@@ -477,11 +477,6 @@ st_create_context_priv(struct gl_context *ctx, struct pipe_context *pipe,
#include "st_atom_list.h"
#undef ST_STATE
if (util_get_cpu_caps()->has_popcnt) {
st->update_functions[ST_NEW_VERTEX_ARRAYS_INDEX] =
st_update_array_with_popcnt;
}
st_init_clear(st);
{
enum pipe_texture_transfer_mode val = screen->get_param(screen, PIPE_CAP_TEXTURE_TRANSFER_MODES);
@@ -775,6 +770,7 @@ st_create_context_priv(struct gl_context *ctx, struct pipe_context *pipe,
_vbo_CreateContext(ctx);
st_init_driver_flags(st);
st_init_update_array(st);
/* Initialize context's winsys buffers list */
list_inithead(&st->winsys_buffers);