brw_clip_unfilled.c \
brw_clip_util.c \
brw_compute.c \
+ brw_conditional_render.c \
brw_context.c \
brw_context.h \
brw_cs.cpp \
--- /dev/null
+/*
+ * Copyright © 2014 Intel Corporation
+ *
+ * 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:
+ * Neil Roberts <neil@linux.intel.com>
+ */
+
+/** @file brw_conditional_render.c
+ *
+ * Support for conditional rendering based on query objects
+ * (GL_NV_conditional_render, GL_ARB_conditional_render_inverted) on Gen7+.
+ */
+
+#include "main/imports.h"
+#include "main/condrender.h"
+
+#include "brw_context.h"
+#include "brw_defines.h"
+#include "intel_batchbuffer.h"
+
+static void
+set_predicate_enable(struct brw_context *brw,
+ bool value)
+{
+ if (value)
+ brw->predicate.state = BRW_PREDICATE_STATE_RENDER;
+ else
+ brw->predicate.state = BRW_PREDICATE_STATE_DONT_RENDER;
+}
+
+static void
+set_predicate_for_result(struct brw_context *brw,
+ struct brw_query_object *query,
+ bool inverted)
+{
+ int load_op;
+
+ assert(query->bo != NULL);
+
+ brw_load_register_mem64(brw,
+ MI_PREDICATE_SRC0,
+ query->bo,
+ I915_GEM_DOMAIN_INSTRUCTION,
+ 0, /* write domain */
+ 0 /* offset */);
+ brw_load_register_mem64(brw,
+ MI_PREDICATE_SRC1,
+ query->bo,
+ I915_GEM_DOMAIN_INSTRUCTION,
+ 0, /* write domain */
+ 8 /* offset */);
+
+ if (inverted)
+ load_op = MI_PREDICATE_LOADOP_LOAD;
+ else
+ load_op = MI_PREDICATE_LOADOP_LOADINV;
+
+ BEGIN_BATCH(1);
+ OUT_BATCH(GEN7_MI_PREDICATE |
+ load_op |
+ MI_PREDICATE_COMBINEOP_SET |
+ MI_PREDICATE_COMPAREOP_SRCS_EQUAL);
+ ADVANCE_BATCH();
+
+ brw->predicate.state = BRW_PREDICATE_STATE_USE_BIT;
+}
+
+static void
+brw_begin_conditional_render(struct gl_context *ctx,
+ struct gl_query_object *q,
+ GLenum mode)
+{
+ struct brw_context *brw = brw_context(ctx);
+ struct brw_query_object *query = (struct brw_query_object *) q;
+ bool inverted;
+
+ if (!brw->predicate.supported)
+ return;
+
+ switch (mode) {
+ case GL_QUERY_WAIT:
+ case GL_QUERY_NO_WAIT:
+ case GL_QUERY_BY_REGION_WAIT:
+ case GL_QUERY_BY_REGION_NO_WAIT:
+ inverted = false;
+ break;
+ case GL_QUERY_WAIT_INVERTED:
+ case GL_QUERY_NO_WAIT_INVERTED:
+ case GL_QUERY_BY_REGION_WAIT_INVERTED:
+ case GL_QUERY_BY_REGION_NO_WAIT_INVERTED:
+ inverted = true;
+ break;
+ default:
+ unreachable("Unexpected conditional render mode");
+ }
+
+ /* If there are already samples from a BLT operation or if the query object
+ * is ready then we can avoid looking at the values in the buffer and just
+ * decide whether to draw using the CPU without stalling.
+ */
+ if (query->Base.Result || query->Base.Ready)
+ set_predicate_enable(brw, (query->Base.Result != 0) ^ inverted);
+ else
+ set_predicate_for_result(brw, query, inverted);
+}
+
+static void
+brw_end_conditional_render(struct gl_context *ctx,
+ struct gl_query_object *q)
+{
+ struct brw_context *brw = brw_context(ctx);
+
+ /* When there is no longer a conditional render in progress it should
+ * always render.
+ */
+ brw->predicate.state = BRW_PREDICATE_STATE_RENDER;
+}
+
+void
+brw_init_conditional_render_functions(struct dd_function_table *functions)
+{
+ functions->BeginConditionalRender = brw_begin_conditional_render;
+ functions->EndConditionalRender = brw_end_conditional_render;
+}
+
+bool
+brw_check_conditional_render(struct brw_context *brw)
+{
+ if (brw->predicate.supported) {
+ /* In some cases it is possible to determine that the primitives should
+ * be skipped without needing the predicate enable bit and still without
+ * stalling.
+ */
+ return brw->predicate.state != BRW_PREDICATE_STATE_DONT_RENDER;
+ } else if (brw->ctx.Query.CondRenderQuery) {
+ perf_debug("Conditional rendering is implemented in software and may "
+ "stall.\n");
+ return _mesa_check_conditional_render(&brw->ctx);
+ } else {
+ return true;
+ }
+}
else
gen4_init_queryobj_functions(functions);
brw_init_compute_functions(functions);
+ if (brw->gen >= 7)
+ brw_init_conditional_render_functions(functions);
functions->QuerySamplesForFormat = brw_query_samples_for_format;
brw->gs.enabled = false;
brw->sf.viewport_transform_enable = true;
+ brw->predicate.state = BRW_PREDICATE_STATE_RENDER;
+
ctx->VertexProgram._MaintainTnlProgram = true;
ctx->FragmentProgram._MaintainTexEnvProgram = true;
uint32_t sampler_offset;
};
+enum brw_predicate_state {
+ /* The first two states are used if we can determine whether to draw
+ * without having to look at the values in the query object buffer. This
+ * will happen if there is no conditional render in progress, if the query
+ * object is already completed or if something else has already added
+ * samples to the preliminary result such as via a BLT command.
+ */
+ BRW_PREDICATE_STATE_RENDER,
+ BRW_PREDICATE_STATE_DONT_RENDER,
+ /* In this case whether to draw or not depends on the result of an
+ * MI_PREDICATE command so the predicate enable bit needs to be checked.
+ */
+ BRW_PREDICATE_STATE_USE_BIT
+};
/**
* brw_context is derived from gl_context.
bool begin_emitted;
} query;
+ struct {
+ enum brw_predicate_state state;
+ bool supported;
+ } predicate;
+
struct {
/** A map from pipeline statistics counter IDs to MMIO addresses. */
const int *statistics_registers;
void brw_store_register_mem64(struct brw_context *brw,
drm_intel_bo *bo, uint32_t reg, int idx);
+/** brw_conditional_render.c */
+void brw_init_conditional_render_functions(struct dd_function_table *functions);
+bool brw_check_conditional_render(struct brw_context *brw);
+
/** intel_batchbuffer.c */
void brw_load_register_mem(struct brw_context *brw,
uint32_t reg,
# define GEN4_3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL (0 << 15)
# define GEN4_3DPRIM_VERTEXBUFFER_ACCESS_RANDOM (1 << 15)
# define GEN7_3DPRIM_INDIRECT_PARAMETER_ENABLE (1 << 10)
+# define GEN7_3DPRIM_PREDICATE_ENABLE (1 << 8)
/* DW1 */
# define GEN7_3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL (0 << 8)
# define GEN7_3DPRIM_VERTEXBUFFER_ACCESS_RANDOM (1 << 8)
int verts_per_instance;
int vertex_access_type;
int indirect_flag;
+ int predicate_enable;
DBG("PRIM: %s %d %d\n", _mesa_lookup_enum_by_nr(prim->mode),
prim->start, prim->count);
indirect_flag = 0;
}
-
if (brw->gen >= 7) {
+ if (brw->predicate.state == BRW_PREDICATE_STATE_USE_BIT)
+ predicate_enable = GEN7_3DPRIM_PREDICATE_ENABLE;
+ else
+ predicate_enable = 0;
+
BEGIN_BATCH(7);
- OUT_BATCH(CMD_3D_PRIM << 16 | (7 - 2) | indirect_flag);
+ OUT_BATCH(CMD_3D_PRIM << 16 | (7 - 2) | indirect_flag | predicate_enable);
OUT_BATCH(hw_prim | vertex_access_type);
} else {
BEGIN_BATCH(6);
assert(unused_tfb_object == NULL);
- if (ctx->Query.CondRenderQuery) {
- perf_debug("Conditional rendering is implemented in software and may "
- "stall. This should be fixed in the driver.\n");
- }
-
- if (!_mesa_check_conditional_render(ctx))
+ if (!brw_check_conditional_render(brw))
return;
/* Handle primitive restart if needed */
void
brw_write_depth_count(struct brw_context *brw, drm_intel_bo *query_bo, int idx)
{
- brw_emit_pipe_control_write(brw,
- PIPE_CONTROL_WRITE_DEPTH_COUNT
- | PIPE_CONTROL_DEPTH_STALL,
- query_bo, idx * sizeof(uint64_t), 0, 0);
+ uint32_t flags;
+
+ flags = (PIPE_CONTROL_WRITE_DEPTH_COUNT |
+ PIPE_CONTROL_DEPTH_STALL);
+
+ /* Needed to ensure the memory is coherent for the MI_LOAD_REGISTER_MEM
+ * command when loading the values into the predicate source registers for
+ * conditional rendering.
+ */
+ if (brw->predicate.supported)
+ flags |= PIPE_CONTROL_FLUSH_ENABLE;
+
+ brw_emit_pipe_control_write(brw, flags, query_bo,
+ idx * sizeof(uint64_t), 0, 0);
}
/**
}
}
+ brw->predicate.supported = false;
+
if (brw->gen >= 7) {
ctx->Extensions.ARB_conservative_depth = true;
ctx->Extensions.ARB_derivative_control = true;
ctx->Extensions.ARB_transform_feedback2 = true;
ctx->Extensions.ARB_transform_feedback3 = true;
ctx->Extensions.ARB_transform_feedback_instanced = true;
+
+ if (brw->intelScreen->cmd_parser_version >= 2)
+ brw->predicate.supported = true;
}
/* Only enable this in core profile because other parts of Mesa behave
#define GEN7_MI_LOAD_REGISTER_MEM (CMD_MI | (0x29 << 23))
# define MI_LOAD_REGISTER_MEM_USE_GGTT (1 << 22)
+/* Manipulate the predicate bit based on some register values. Only on Gen7+ */
+#define GEN7_MI_PREDICATE (CMD_MI | (0xC << 23))
+# define MI_PREDICATE_LOADOP_KEEP (0 << 6)
+# define MI_PREDICATE_LOADOP_LOAD (2 << 6)
+# define MI_PREDICATE_LOADOP_LOADINV (3 << 6)
+# define MI_PREDICATE_COMBINEOP_SET (0 << 3)
+# define MI_PREDICATE_COMBINEOP_AND (1 << 3)
+# define MI_PREDICATE_COMBINEOP_OR (2 << 3)
+# define MI_PREDICATE_COMBINEOP_XOR (3 << 3)
+# define MI_PREDICATE_COMPAREOP_TRUE (0 << 0)
+# define MI_PREDICATE_COMPAREOP_FALSE (1 << 0)
+# define MI_PREDICATE_COMPAREOP_SRCS_EQUAL (2 << 0)
+# define MI_PREDICATE_COMPAREOP_DELTAS_EQUAL (3 << 0)
+
/** @{
*
* PIPE_CONTROL operation, a combination MI_FLUSH and register write with
#define PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE (1 << 10) /* GM45+ only */
#define PIPE_CONTROL_ISP_DIS (1 << 9)
#define PIPE_CONTROL_INTERRUPT_ENABLE (1 << 8)
+#define PIPE_CONTROL_FLUSH_ENABLE (1 << 7) /* Gen7+ only */
/* GT */
#define PIPE_CONTROL_DATA_CACHE_INVALIDATE (1 << 5)
#define PIPE_CONTROL_VF_CACHE_INVALIDATE (1 << 4)
# define GEN9_PARTIAL_RESOLVE_DISABLE_IN_VC (1 << 1)
# define GEN8_HIZ_PMA_MASK_BITS \
((GEN8_HIZ_NP_PMA_FIX_ENABLE | GEN8_HIZ_NP_EARLY_Z_FAILS_DISABLE) << 16)
+
+/* Predicate registers */
+#define MI_PREDICATE_SRC0 0x2400
+#define MI_PREDICATE_SRC1 0x2408
+#define MI_PREDICATE_DATA 0x2410
+#define MI_PREDICATE_RESULT 0x2418
+#define MI_PREDICATE_RESULT_1 0x241C
+#define MI_PREDICATE_RESULT_2 0x2214
projects / mesa.git / commitdiff