From: Marek Olšák <marek.ol...@amd.com>
This is needed to get shader-db stats for LS,HS,ES,GS stages on gfx9.
---
src/gallium/drivers/radeonsi/si_shader.c | 33 +++++++++++++++----------
src/gallium/drivers/radeonsi/si_shader.h | 2 ++
src/gallium/drivers/radeonsi/si_state_shaders.c | 1 +
3 files changed, 23 insertions(+), 13 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_shader.c
b/src/gallium/drivers/radeonsi/si_shader.c
index f1ac94f..de1f725 100644
--- a/src/gallium/drivers/radeonsi/si_shader.c
+++ b/src/gallium/drivers/radeonsi/si_shader.c
@@ -5418,29 +5418,43 @@ static void si_calculate_max_simd_waves(struct
si_shader *shader)
max_simd_waves = MIN2(max_simd_waves, 256 / conf->num_vgprs);
/* LDS is 64KB per CU (4 SIMDs), which is 16KB per SIMD (usage above
* 16KB makes some SIMDs unoccupied). */
if (lds_per_wave)
max_simd_waves = MIN2(max_simd_waves, 16384 / lds_per_wave);
conf->max_simd_waves = max_simd_waves;
}
+void si_shader_dump_stats_for_shader_db(const struct si_shader *shader,
+ struct pipe_debug_callback *debug)
+{
+ const struct si_shader_config *conf = &shader->config;
+
+ pipe_debug_message(debug, SHADER_INFO,
+ "Shader Stats: SGPRS: %d VGPRS: %d Code Size: %d "
+ "LDS: %d Scratch: %d Max Waves: %d Spilled SGPRs: %d
"
+ "Spilled VGPRs: %d PrivMem VGPRs: %d",
+ conf->num_sgprs, conf->num_vgprs,
+ si_get_shader_binary_size(shader),
+ conf->lds_size, conf->scratch_bytes_per_wave,
+ conf->max_simd_waves, conf->spilled_sgprs,
+ conf->spilled_vgprs, conf->private_mem_vgprs);
+}
+
static void si_shader_dump_stats(struct si_screen *sscreen,
const struct si_shader *shader,
- struct pipe_debug_callback *debug,
unsigned processor,
FILE *file,
bool check_debug_option)
{
const struct si_shader_config *conf = &shader->config;
- unsigned code_size = si_get_shader_binary_size(shader);
if (!check_debug_option ||
si_can_dump_shader(sscreen, processor)) {
if (processor == PIPE_SHADER_FRAGMENT) {
fprintf(file, "*** SHADER CONFIG ***\n"
"SPI_PS_INPUT_ADDR = 0x%04x\n"
"SPI_PS_INPUT_ENA = 0x%04x\n",
conf->spi_ps_input_addr,
conf->spi_ps_input_ena);
}
@@ -5450,33 +5464,25 @@ static void si_shader_dump_stats(struct si_screen
*sscreen,
"Spilled SGPRs: %d\n"
"Spilled VGPRs: %d\n"
"Private memory VGPRs: %d\n"
"Code Size: %d bytes\n"
"LDS: %d blocks\n"
"Scratch: %d bytes per wave\n"
"Max Waves: %d\n"
"********************\n\n\n",
conf->num_sgprs, conf->num_vgprs,
conf->spilled_sgprs, conf->spilled_vgprs,
- conf->private_mem_vgprs, code_size,
+ conf->private_mem_vgprs,
+ si_get_shader_binary_size(shader),
conf->lds_size, conf->scratch_bytes_per_wave,
conf->max_simd_waves);
}
-
- pipe_debug_message(debug, SHADER_INFO,
- "Shader Stats: SGPRS: %d VGPRS: %d Code Size: %d "
- "LDS: %d Scratch: %d Max Waves: %d Spilled SGPRs: %d
"
- "Spilled VGPRs: %d PrivMem VGPRs: %d",
- conf->num_sgprs, conf->num_vgprs, code_size,
- conf->lds_size, conf->scratch_bytes_per_wave,
- conf->max_simd_waves, conf->spilled_sgprs,
- conf->spilled_vgprs, conf->private_mem_vgprs);
}
const char *si_get_shader_name(const struct si_shader *shader, unsigned
processor)
{
switch (processor) {
case PIPE_SHADER_VERTEX:
if (shader->key.as_es)
return "Vertex Shader as ES";
else if (shader->key.as_ls)
return "Vertex Shader as LS";
@@ -5540,21 +5546,21 @@ void si_shader_dump(struct si_screen *sscreen, const
struct si_shader *shader,
debug, "prolog2", file);
si_shader_dump_disassembly(&shader->binary, debug, "main",
file);
if (shader->epilog)
si_shader_dump_disassembly(&shader->epilog->binary,
debug, "epilog", file);
fprintf(file, "\n");
}
- si_shader_dump_stats(sscreen, shader, debug, processor, file,
+ si_shader_dump_stats(sscreen, shader, processor, file,
check_debug_option);
}
static int si_compile_llvm(struct si_screen *sscreen,
struct ac_shader_binary *binary,
struct si_shader_config *conf,
LLVMTargetMachineRef tm,
LLVMModuleRef mod,
struct pipe_debug_callback *debug,
unsigned processor,
@@ -6969,20 +6975,21 @@ int si_compile_tgsi_shader(struct si_screen *sscreen,
shader->info.ancillary_vgpr_index =
shader->info.num_input_vgprs;
shader->info.num_input_vgprs += 1;
}
if
(G_0286CC_SAMPLE_COVERAGE_ENA(shader->config.spi_ps_input_addr))
shader->info.num_input_vgprs += 1;
if (G_0286CC_POS_FIXED_PT_ENA(shader->config.spi_ps_input_addr))
shader->info.num_input_vgprs += 1;
}
si_calculate_max_simd_waves(shader);
+ si_shader_dump_stats_for_shader_db(shader, debug);
return 0;
}
/**
* Create, compile and return a shader part (prolog or epilog).
*
* \param sscreen screen
* \param list list of shader parts of the same category
* \param type shader type
* \param key shader part key
diff --git a/src/gallium/drivers/radeonsi/si_shader.h
b/src/gallium/drivers/radeonsi/si_shader.h
index 6ed1646..3cc49ca 100644
--- a/src/gallium/drivers/radeonsi/si_shader.h
+++ b/src/gallium/drivers/radeonsi/si_shader.h
@@ -634,20 +634,22 @@ int si_compile_tgsi_shader(struct si_screen *sscreen,
int si_shader_create(struct si_screen *sscreen, LLVMTargetMachineRef tm,
struct si_shader *shader,
struct pipe_debug_callback *debug);
void si_shader_destroy(struct si_shader *shader);
unsigned si_shader_io_get_unique_index_patch(unsigned semantic_name, unsigned
index);
unsigned si_shader_io_get_unique_index(unsigned semantic_name, unsigned index);
int si_shader_binary_upload(struct si_screen *sscreen, struct si_shader
*shader);
void si_shader_dump(struct si_screen *sscreen, const struct si_shader *shader,
struct pipe_debug_callback *debug, unsigned processor,
FILE *f, bool check_debug_option);
+void si_shader_dump_stats_for_shader_db(const struct si_shader *shader,
+ struct pipe_debug_callback *debug);
void si_multiwave_lds_size_workaround(struct si_screen *sscreen,
unsigned *lds_size);
void si_shader_apply_scratch_relocs(struct si_shader *shader,
uint64_t scratch_va);
void si_shader_binary_read_config(struct ac_shader_binary *binary,
struct si_shader_config *conf,
unsigned symbol_offset);
const char *si_get_shader_name(const struct si_shader *shader, unsigned
processor);
/* si_shader_nir.c */
diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c
b/src/gallium/drivers/radeonsi/si_state_shaders.c
index ecb8a0d..959aead 100644
--- a/src/gallium/drivers/radeonsi/si_state_shaders.c
+++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
@@ -1816,20 +1816,21 @@ static void si_init_shader_selector_async(void *job,
int thread_index)
if (sel->tokens)
tgsi_binary = si_get_tgsi_binary(sel);
/* Try to load the shader from the shader cache. */
mtx_lock(&sscreen->shader_cache_mutex);
if (tgsi_binary &&
si_shader_cache_load_shader(sscreen, tgsi_binary, shader)) {
mtx_unlock(&sscreen->shader_cache_mutex);
+ si_shader_dump_stats_for_shader_db(shader, debug);
} else {
mtx_unlock(&sscreen->shader_cache_mutex);
/* Compile the shader if it hasn't been loaded from the
cache. */
if (si_compile_tgsi_shader(sscreen, tm, shader, false,
debug) != 0) {
FREE(shader);
FREE(tgsi_binary);
fprintf(stderr, "radeonsi: can't compile a main
shader part\n");
return;
--
2.7.4
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
