2015-11-12 21:04:50 +00:00
|
|
|
/*
|
|
|
|
* Copyright 2015 Advanced Micro Devices, Inc.
|
|
|
|
*
|
|
|
|
* Permission is hereby granted, free of charge, to any person obtaining a
|
|
|
|
* copy of this software and associated documentation files (the "Software"),
|
|
|
|
* to deal in the Software without restriction, including without limitation
|
|
|
|
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
|
|
|
|
* and/or sell copies of the Software, and to permit persons to whom the
|
|
|
|
* Software is furnished to do so, subject to the following conditions:
|
|
|
|
*
|
|
|
|
* The above copyright notice and this permission notice (including the next
|
|
|
|
* paragraph) shall be included in all copies or substantial portions of the
|
|
|
|
* Software.
|
|
|
|
*
|
|
|
|
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
|
|
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
|
|
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
|
|
|
|
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
|
|
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
|
|
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
|
|
* SOFTWARE.
|
|
|
|
*
|
|
|
|
* Authors:
|
|
|
|
* Nicolai Hähnle <nicolai.haehnle@amd.com>
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef R600_QUERY_H
|
|
|
|
#define R600_QUERY_H
|
|
|
|
|
|
|
|
#include "pipe/p_defines.h"
|
2016-09-14 08:46:36 +01:00
|
|
|
#include "pipe/p_state.h"
|
2015-11-18 10:59:21 +00:00
|
|
|
#include "util/list.h"
|
2015-11-12 21:04:50 +00:00
|
|
|
|
2015-11-25 14:30:03 +00:00
|
|
|
struct pipe_context;
|
|
|
|
struct pipe_query;
|
2016-09-14 11:00:50 +01:00
|
|
|
struct pipe_resource;
|
2015-11-25 14:30:03 +00:00
|
|
|
|
2015-11-18 10:40:00 +00:00
|
|
|
struct r600_common_context;
|
2015-11-25 14:30:03 +00:00
|
|
|
struct r600_common_screen;
|
2015-11-18 10:40:00 +00:00
|
|
|
struct r600_query;
|
2015-11-18 11:05:11 +00:00
|
|
|
struct r600_query_hw;
|
2015-11-18 10:59:21 +00:00
|
|
|
struct r600_resource;
|
2015-11-18 10:40:00 +00:00
|
|
|
|
2016-04-16 12:35:08 +01:00
|
|
|
enum {
|
|
|
|
R600_QUERY_DRAW_CALLS = PIPE_QUERY_DRIVER_SPECIFIC,
|
2016-06-09 22:16:43 +01:00
|
|
|
R600_QUERY_SPILL_DRAW_CALLS,
|
|
|
|
R600_QUERY_COMPUTE_CALLS,
|
|
|
|
R600_QUERY_SPILL_COMPUTE_CALLS,
|
|
|
|
R600_QUERY_DMA_CALLS,
|
2016-10-28 00:55:17 +01:00
|
|
|
R600_QUERY_CP_DMA_CALLS,
|
2016-08-23 14:17:35 +01:00
|
|
|
R600_QUERY_NUM_VS_FLUSHES,
|
|
|
|
R600_QUERY_NUM_PS_FLUSHES,
|
|
|
|
R600_QUERY_NUM_CS_FLUSHES,
|
2016-12-25 18:48:55 +00:00
|
|
|
R600_QUERY_NUM_FB_CACHE_FLUSHES,
|
|
|
|
R600_QUERY_NUM_L2_INVALIDATES,
|
|
|
|
R600_QUERY_NUM_L2_WRITEBACKS,
|
2017-02-11 19:24:07 +00:00
|
|
|
R600_QUERY_CS_THREAD_BUSY,
|
2016-04-16 12:35:08 +01:00
|
|
|
R600_QUERY_REQUESTED_VRAM,
|
|
|
|
R600_QUERY_REQUESTED_GTT,
|
2016-08-05 00:38:23 +01:00
|
|
|
R600_QUERY_MAPPED_VRAM,
|
|
|
|
R600_QUERY_MAPPED_GTT,
|
2016-04-16 12:35:08 +01:00
|
|
|
R600_QUERY_BUFFER_WAIT_TIME,
|
2017-01-23 20:44:45 +00:00
|
|
|
R600_QUERY_NUM_MAPPED_BUFFERS,
|
2016-12-27 16:54:18 +00:00
|
|
|
R600_QUERY_NUM_GFX_IBS,
|
2016-12-27 17:35:32 +00:00
|
|
|
R600_QUERY_NUM_SDMA_IBS,
|
2016-04-16 12:35:08 +01:00
|
|
|
R600_QUERY_NUM_BYTES_MOVED,
|
2016-08-18 00:18:14 +01:00
|
|
|
R600_QUERY_NUM_EVICTIONS,
|
2016-04-16 12:35:08 +01:00
|
|
|
R600_QUERY_VRAM_USAGE,
|
2017-01-25 15:56:46 +00:00
|
|
|
R600_QUERY_VRAM_VIS_USAGE,
|
2016-04-16 12:35:08 +01:00
|
|
|
R600_QUERY_GTT_USAGE,
|
|
|
|
R600_QUERY_GPU_TEMPERATURE,
|
|
|
|
R600_QUERY_CURRENT_GPU_SCLK,
|
|
|
|
R600_QUERY_CURRENT_GPU_MCLK,
|
|
|
|
R600_QUERY_GPU_LOAD,
|
2017-01-15 22:59:00 +00:00
|
|
|
R600_QUERY_GPU_SHADERS_BUSY,
|
2017-01-20 18:21:12 +00:00
|
|
|
R600_QUERY_GPU_TA_BUSY,
|
|
|
|
R600_QUERY_GPU_GDS_BUSY,
|
|
|
|
R600_QUERY_GPU_VGT_BUSY,
|
|
|
|
R600_QUERY_GPU_IA_BUSY,
|
|
|
|
R600_QUERY_GPU_SX_BUSY,
|
|
|
|
R600_QUERY_GPU_WD_BUSY,
|
|
|
|
R600_QUERY_GPU_BCI_BUSY,
|
|
|
|
R600_QUERY_GPU_SC_BUSY,
|
|
|
|
R600_QUERY_GPU_PA_BUSY,
|
|
|
|
R600_QUERY_GPU_DB_BUSY,
|
|
|
|
R600_QUERY_GPU_CP_BUSY,
|
|
|
|
R600_QUERY_GPU_CB_BUSY,
|
2017-01-30 11:52:24 +00:00
|
|
|
R600_QUERY_GPU_SDMA_BUSY,
|
2017-01-30 11:52:56 +00:00
|
|
|
R600_QUERY_GPU_PFP_BUSY,
|
|
|
|
R600_QUERY_GPU_MEQ_BUSY,
|
|
|
|
R600_QUERY_GPU_ME_BUSY,
|
|
|
|
R600_QUERY_GPU_SURF_SYNC_BUSY,
|
|
|
|
R600_QUERY_GPU_DMA_BUSY,
|
|
|
|
R600_QUERY_GPU_SCRATCH_RAM_BUSY,
|
|
|
|
R600_QUERY_GPU_CE_BUSY,
|
2016-04-16 12:35:08 +01:00
|
|
|
R600_QUERY_NUM_COMPILATIONS,
|
|
|
|
R600_QUERY_NUM_SHADERS_CREATED,
|
2016-06-21 18:36:14 +01:00
|
|
|
R600_QUERY_BACK_BUFFER_PS_DRAW_RATIO,
|
2016-02-11 23:58:46 +00:00
|
|
|
R600_QUERY_NUM_SHADER_CACHE_HITS,
|
2016-04-16 12:35:08 +01:00
|
|
|
R600_QUERY_GPIN_ASIC_ID,
|
|
|
|
R600_QUERY_GPIN_NUM_SIMD,
|
|
|
|
R600_QUERY_GPIN_NUM_RB,
|
|
|
|
R600_QUERY_GPIN_NUM_SPI,
|
|
|
|
R600_QUERY_GPIN_NUM_SE,
|
|
|
|
|
|
|
|
R600_QUERY_FIRST_PERFCOUNTER = PIPE_QUERY_DRIVER_SPECIFIC + 100,
|
|
|
|
};
|
2015-11-12 21:04:50 +00:00
|
|
|
|
2015-12-11 20:39:23 +00:00
|
|
|
enum {
|
|
|
|
R600_QUERY_GROUP_GPIN = 0,
|
|
|
|
R600_NUM_SW_QUERY_GROUPS
|
|
|
|
};
|
|
|
|
|
2015-11-18 10:40:00 +00:00
|
|
|
struct r600_query_ops {
|
|
|
|
void (*destroy)(struct r600_common_context *, struct r600_query *);
|
2016-06-21 20:29:39 +01:00
|
|
|
bool (*begin)(struct r600_common_context *, struct r600_query *);
|
2016-04-20 15:43:09 +01:00
|
|
|
bool (*end)(struct r600_common_context *, struct r600_query *);
|
2016-06-21 20:29:39 +01:00
|
|
|
bool (*get_result)(struct r600_common_context *,
|
|
|
|
struct r600_query *, bool wait,
|
|
|
|
union pipe_query_result *result);
|
2016-09-14 11:00:50 +01:00
|
|
|
void (*get_result_resource)(struct r600_common_context *,
|
|
|
|
struct r600_query *, bool wait,
|
|
|
|
enum pipe_query_value_type result_type,
|
|
|
|
int index,
|
|
|
|
struct pipe_resource *resource,
|
|
|
|
unsigned offset);
|
2015-11-18 10:40:00 +00:00
|
|
|
};
|
|
|
|
|
2015-11-18 10:59:21 +00:00
|
|
|
struct r600_query {
|
|
|
|
struct r600_query_ops *ops;
|
|
|
|
|
|
|
|
/* The type of query */
|
|
|
|
unsigned type;
|
|
|
|
};
|
|
|
|
|
2015-11-18 11:05:11 +00:00
|
|
|
enum {
|
|
|
|
R600_QUERY_HW_FLAG_NO_START = (1 << 0),
|
2016-09-15 14:58:36 +01:00
|
|
|
/* gap */
|
2016-06-21 14:49:25 +01:00
|
|
|
/* whether begin_query doesn't clear the result */
|
|
|
|
R600_QUERY_HW_FLAG_BEGIN_RESUMES = (1 << 2),
|
2015-11-18 11:05:11 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
struct r600_query_hw_ops {
|
2017-03-01 18:24:42 +00:00
|
|
|
bool (*prepare_buffer)(struct r600_common_screen *,
|
2015-11-18 11:05:11 +00:00
|
|
|
struct r600_query_hw *,
|
|
|
|
struct r600_resource *);
|
|
|
|
void (*emit_start)(struct r600_common_context *,
|
|
|
|
struct r600_query_hw *,
|
|
|
|
struct r600_resource *buffer, uint64_t va);
|
|
|
|
void (*emit_stop)(struct r600_common_context *,
|
|
|
|
struct r600_query_hw *,
|
|
|
|
struct r600_resource *buffer, uint64_t va);
|
2015-11-12 23:27:34 +00:00
|
|
|
void (*clear_result)(struct r600_query_hw *, union pipe_query_result *);
|
|
|
|
void (*add_result)(struct r600_common_context *ctx,
|
|
|
|
struct r600_query_hw *, void *buffer,
|
|
|
|
union pipe_query_result *result);
|
2015-11-18 11:05:11 +00:00
|
|
|
};
|
|
|
|
|
2015-11-18 10:59:21 +00:00
|
|
|
struct r600_query_buffer {
|
|
|
|
/* The buffer where query results are stored. */
|
|
|
|
struct r600_resource *buf;
|
|
|
|
/* Offset of the next free result after current query data */
|
|
|
|
unsigned results_end;
|
|
|
|
/* If a query buffer is full, a new buffer is created and the old one
|
|
|
|
* is put in here. When we calculate the result, we sum up the samples
|
|
|
|
* from all buffers. */
|
|
|
|
struct r600_query_buffer *previous;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct r600_query_hw {
|
|
|
|
struct r600_query b;
|
2015-11-18 11:05:11 +00:00
|
|
|
struct r600_query_hw_ops *ops;
|
|
|
|
unsigned flags;
|
2015-11-18 10:59:21 +00:00
|
|
|
|
|
|
|
/* The query buffer and how many results are in it. */
|
|
|
|
struct r600_query_buffer buffer;
|
|
|
|
/* Size of the result in memory for both begin_query and end_query,
|
|
|
|
* this can be one or two numbers, or it could even be a size of a structure. */
|
|
|
|
unsigned result_size;
|
|
|
|
/* The number of dwords for begin_query or end_query. */
|
2015-11-12 23:38:36 +00:00
|
|
|
unsigned num_cs_dw_begin;
|
|
|
|
unsigned num_cs_dw_end;
|
2015-11-18 10:59:21 +00:00
|
|
|
/* Linked list of queries */
|
|
|
|
struct list_head list;
|
|
|
|
/* For transform feedback: which stream the query is for */
|
|
|
|
unsigned stream;
|
|
|
|
};
|
|
|
|
|
2017-03-01 18:24:42 +00:00
|
|
|
bool r600_query_hw_init(struct r600_common_screen *rscreen,
|
2016-06-21 20:29:39 +01:00
|
|
|
struct r600_query_hw *query);
|
2015-11-18 10:59:21 +00:00
|
|
|
void r600_query_hw_destroy(struct r600_common_context *rctx,
|
|
|
|
struct r600_query *rquery);
|
2016-06-21 20:29:39 +01:00
|
|
|
bool r600_query_hw_begin(struct r600_common_context *rctx,
|
|
|
|
struct r600_query *rquery);
|
2016-04-20 15:43:09 +01:00
|
|
|
bool r600_query_hw_end(struct r600_common_context *rctx,
|
2015-11-18 11:06:58 +00:00
|
|
|
struct r600_query *rquery);
|
2016-06-21 20:29:39 +01:00
|
|
|
bool r600_query_hw_get_result(struct r600_common_context *rctx,
|
|
|
|
struct r600_query *rquery,
|
|
|
|
bool wait,
|
|
|
|
union pipe_query_result *result);
|
2015-11-18 10:59:21 +00:00
|
|
|
|
2015-11-25 14:30:03 +00:00
|
|
|
/* Performance counters */
|
|
|
|
enum {
|
|
|
|
/* This block is part of the shader engine */
|
|
|
|
R600_PC_BLOCK_SE = (1 << 0),
|
|
|
|
|
|
|
|
/* Expose per-instance groups instead of summing all instances (within
|
|
|
|
* an SE). */
|
|
|
|
R600_PC_BLOCK_INSTANCE_GROUPS = (1 << 1),
|
|
|
|
|
|
|
|
/* Expose per-SE groups instead of summing instances across SEs. */
|
|
|
|
R600_PC_BLOCK_SE_GROUPS = (1 << 2),
|
|
|
|
|
|
|
|
/* Shader block */
|
|
|
|
R600_PC_BLOCK_SHADER = (1 << 3),
|
|
|
|
|
|
|
|
/* Non-shader block with perfcounters windowed by shaders. */
|
|
|
|
R600_PC_BLOCK_SHADER_WINDOWED = (1 << 4),
|
|
|
|
};
|
|
|
|
|
|
|
|
/* Describes a hardware block with performance counters. Multiple instances of
|
|
|
|
* each block, possibly per-SE, may exist on the chip. Depending on the block
|
|
|
|
* and on the user's configuration, we either
|
|
|
|
* (a) expose every instance as a performance counter group,
|
|
|
|
* (b) expose a single performance counter group that reports the sum over all
|
|
|
|
* instances, or
|
|
|
|
* (c) expose one performance counter group per instance, but summed over all
|
|
|
|
* shader engines.
|
|
|
|
*/
|
|
|
|
struct r600_perfcounter_block {
|
|
|
|
const char *basename;
|
|
|
|
unsigned flags;
|
|
|
|
unsigned num_counters;
|
|
|
|
unsigned num_selectors;
|
|
|
|
unsigned num_instances;
|
|
|
|
|
|
|
|
unsigned num_groups;
|
|
|
|
char *group_names;
|
|
|
|
unsigned group_name_stride;
|
|
|
|
|
|
|
|
char *selector_names;
|
|
|
|
unsigned selector_name_stride;
|
|
|
|
|
|
|
|
void *data;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct r600_perfcounters {
|
|
|
|
unsigned num_groups;
|
|
|
|
unsigned num_blocks;
|
|
|
|
struct r600_perfcounter_block *blocks;
|
|
|
|
|
|
|
|
unsigned num_start_cs_dwords;
|
|
|
|
unsigned num_stop_cs_dwords;
|
|
|
|
unsigned num_instance_cs_dwords;
|
|
|
|
unsigned num_shaders_cs_dwords;
|
|
|
|
|
2015-12-11 20:58:11 +00:00
|
|
|
unsigned num_shader_types;
|
|
|
|
const char * const *shader_type_suffixes;
|
|
|
|
const unsigned *shader_type_bits;
|
|
|
|
|
2015-11-25 14:30:03 +00:00
|
|
|
void (*get_size)(struct r600_perfcounter_block *,
|
|
|
|
unsigned count, unsigned *selectors,
|
|
|
|
unsigned *num_select_dw, unsigned *num_read_dw);
|
|
|
|
|
|
|
|
void (*emit_instance)(struct r600_common_context *,
|
|
|
|
int se, int instance);
|
|
|
|
void (*emit_shaders)(struct r600_common_context *, unsigned shaders);
|
|
|
|
void (*emit_select)(struct r600_common_context *,
|
|
|
|
struct r600_perfcounter_block *,
|
|
|
|
unsigned count, unsigned *selectors);
|
|
|
|
void (*emit_start)(struct r600_common_context *,
|
|
|
|
struct r600_resource *buffer, uint64_t va);
|
|
|
|
void (*emit_stop)(struct r600_common_context *,
|
|
|
|
struct r600_resource *buffer, uint64_t va);
|
|
|
|
void (*emit_read)(struct r600_common_context *,
|
|
|
|
struct r600_perfcounter_block *,
|
|
|
|
unsigned count, unsigned *selectors,
|
|
|
|
struct r600_resource *buffer, uint64_t va);
|
|
|
|
|
|
|
|
void (*cleanup)(struct r600_common_screen *);
|
|
|
|
|
2016-06-21 20:29:39 +01:00
|
|
|
bool separate_se;
|
|
|
|
bool separate_instance;
|
2015-11-25 14:30:03 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
struct pipe_query *r600_create_batch_query(struct pipe_context *ctx,
|
|
|
|
unsigned num_queries,
|
|
|
|
unsigned *query_types);
|
|
|
|
|
|
|
|
int r600_get_perfcounter_info(struct r600_common_screen *,
|
|
|
|
unsigned index,
|
|
|
|
struct pipe_driver_query_info *info);
|
|
|
|
int r600_get_perfcounter_group_info(struct r600_common_screen *,
|
|
|
|
unsigned index,
|
|
|
|
struct pipe_driver_query_group_info *info);
|
|
|
|
|
2016-06-21 20:29:39 +01:00
|
|
|
bool r600_perfcounters_init(struct r600_perfcounters *, unsigned num_blocks);
|
2015-11-25 11:32:23 +00:00
|
|
|
void r600_perfcounters_add_block(struct r600_common_screen *,
|
|
|
|
struct r600_perfcounters *,
|
|
|
|
const char *name, unsigned flags,
|
|
|
|
unsigned counters, unsigned selectors,
|
|
|
|
unsigned instances, void *data);
|
2015-11-25 14:30:03 +00:00
|
|
|
void r600_perfcounters_do_destroy(struct r600_perfcounters *);
|
2016-06-27 18:46:39 +01:00
|
|
|
void r600_query_hw_reset_buffers(struct r600_common_context *rctx,
|
|
|
|
struct r600_query_hw *query);
|
2015-11-25 14:30:03 +00:00
|
|
|
|
2016-09-14 08:46:36 +01:00
|
|
|
struct r600_qbo_state {
|
|
|
|
void *saved_compute;
|
|
|
|
struct pipe_constant_buffer saved_const0;
|
|
|
|
struct pipe_shader_buffer saved_ssbo[3];
|
|
|
|
};
|
|
|
|
|
2015-11-12 21:04:50 +00:00
|
|
|
#endif /* R600_QUERY_H */
|