[Mesa-dev] [PATCH 24/47] i965/fs: Add byte scattered write message and fs support
Alejandro Piñeiro
apinheiro at igalia.com
Thu Aug 24 13:54:39 UTC 2017
From: Jose Maria Casanova Crespo <jmcasanova at igalia.com>
Signed-off-by: Jose Maria Casanova Crespo <jmcasanova at igalia.com>
Signed-off-by: Alejandro Piñeiro <apinheiro at igalia.com>
---
src/intel/compiler/brw_eu.h | 6 ++
src/intel/compiler/brw_eu_defines.h | 17 +++++
src/intel/compiler/brw_eu_emit.c | 89 ++++++++++++++++++++++++++
src/intel/compiler/brw_fs.cpp | 10 +++
src/intel/compiler/brw_fs_copy_propagation.cpp | 2 +
src/intel/compiler/brw_fs_generator.cpp | 5 ++
src/intel/compiler/brw_fs_surface_builder.cpp | 17 +++++
src/intel/compiler/brw_fs_surface_builder.h | 9 +++
src/intel/compiler/brw_shader.cpp | 7 ++
9 files changed, 162 insertions(+)
diff --git a/src/intel/compiler/brw_eu.h b/src/intel/compiler/brw_eu.h
index 0a7f8020398..228c1bf6e9e 100644
--- a/src/intel/compiler/brw_eu.h
+++ b/src/intel/compiler/brw_eu.h
@@ -476,6 +476,12 @@ brw_typed_surface_write(struct brw_codegen *p,
unsigned num_channels);
void
+brw_byte_scattered_write(struct brw_codegen *p,
+ struct brw_reg payload,
+ struct brw_reg surface,
+ unsigned msg_length);
+
+void
brw_memory_fence(struct brw_codegen *p,
struct brw_reg dst);
diff --git a/src/intel/compiler/brw_eu_defines.h b/src/intel/compiler/brw_eu_defines.h
index 50435df2fcf..7f76cd06527 100644
--- a/src/intel/compiler/brw_eu_defines.h
+++ b/src/intel/compiler/brw_eu_defines.h
@@ -388,6 +388,16 @@ enum opcode {
SHADER_OPCODE_TYPED_SURFACE_WRITE,
SHADER_OPCODE_TYPED_SURFACE_WRITE_LOGICAL,
+ /**
+ * Byte scattered write/read opcodes.
+ *
+ * LOGICAL opcodes are eventually translated to the matching non-LOGICAL
+ * opcode, but instead of taking a single payload blog they expect their
+ * arguments separately as individual sources, like untyped write/read.
+ */
+ SHADER_OPCODE_BYTE_SCATTERED_WRITE,
+ SHADER_OPCODE_BYTE_SCATTERED_WRITE_LOGICAL,
+
SHADER_OPCODE_RND_MODE_RTE,
SHADER_OPCODE_RND_MODE_RTZ,
@@ -1242,4 +1252,11 @@ enum PACKED brw_rnd_mode {
BRW_RND_MODE_RTZ,
};
+/* MDC_DS - Data Size Message Descriptor Control Field */
+enum PACKED brw_data_size {
+ GEN7_BYTE_SCATTERED_DATA_SIZE_BYTE = 0,
+ GEN7_BYTE_SCATTERED_DATA_SIZE_WORD = 1,
+ GEN7_BYTE_SCATTERED_DATA_SIZE_DWORD = 2
+};
+
#endif /* BRW_EU_DEFINES_H */
diff --git a/src/intel/compiler/brw_eu_emit.c b/src/intel/compiler/brw_eu_emit.c
index 07ad3d9384b..a6acadbbf5f 100644
--- a/src/intel/compiler/brw_eu_emit.c
+++ b/src/intel/compiler/brw_eu_emit.c
@@ -2756,6 +2756,49 @@ brw_send_indirect_surface_message(struct brw_codegen *p,
return insn;
}
+
+static struct brw_inst *
+brw_send_indirect_scattered_message(struct brw_codegen *p,
+ unsigned sfid,
+ struct brw_reg dst,
+ struct brw_reg payload,
+ struct brw_reg surface,
+ unsigned message_len,
+ unsigned response_len,
+ bool header_present)
+{
+ const struct gen_device_info *devinfo = p->devinfo;
+ struct brw_inst *insn;
+
+ if (surface.file != BRW_IMMEDIATE_VALUE) {
+ struct brw_reg addr = retype(brw_address_reg(0), BRW_REGISTER_TYPE_UD);
+
+ brw_push_insn_state(p);
+ brw_set_default_access_mode(p, BRW_ALIGN_1);
+ brw_set_default_mask_control(p, BRW_MASK_DISABLE);
+ brw_set_default_predicate_control(p, BRW_PREDICATE_NONE);
+
+ /* Mask out invalid bits from the surface index to avoid hangs e.g. when
+ * some surface array is accessed out of bounds.
+ */
+ insn = brw_AND(p, addr,
+ suboffset(vec1(retype(surface, BRW_REGISTER_TYPE_UD)),
+ BRW_GET_SWZ(surface.swizzle, 0)),
+ brw_imm_ud(0xff));
+
+ brw_pop_insn_state(p);
+
+ surface = addr;
+ }
+
+ insn = brw_send_indirect_message(p, sfid, dst, payload, surface);
+ brw_inst_set_mlen(devinfo, insn, message_len);
+ brw_inst_set_rlen(devinfo, insn, response_len);
+ brw_inst_set_header_present(devinfo, insn, header_present);
+
+ return insn;
+}
+
static bool
while_jumps_before_offset(const struct gen_device_info *devinfo,
brw_inst *insn, int while_offset, int start_offset)
@@ -3160,6 +3203,52 @@ brw_untyped_surface_write(struct brw_codegen *p,
}
static void
+brw_set_dp_byte_scattered_write(struct brw_codegen *p,
+ struct brw_inst *insn)
+{
+ const struct gen_device_info *devinfo = p->devinfo;
+
+ /* Although we could configure this message to write BYTE, WORD, or DWORD,
+ * it was added for the need of writing WORD sizes, so we use directly that
+ * size. This could be revisited on the future.
+ */
+ unsigned msg_control = GEN7_BYTE_SCATTERED_DATA_SIZE_WORD << 2;
+
+ assert(brw_inst_access_mode(devinfo, p->current) == BRW_ALIGN_1);
+ if (brw_inst_exec_size(devinfo, p->current) == BRW_EXECUTE_16)
+ msg_control |= 1;
+ else
+ msg_control |= 0;
+
+ brw_inst_set_dp_msg_type(devinfo, insn,
+ devinfo->gen >= 8 || devinfo->is_haswell ?
+ HSW_DATAPORT_DC_PORT0_BYTE_SCATTERED_WRITE :
+ GEN7_DATAPORT_DC_BYTE_SCATTERED_WRITE);
+ brw_inst_set_dp_msg_control(devinfo, insn, msg_control);
+}
+
+
+void
+brw_byte_scattered_write(struct brw_codegen *p,
+ struct brw_reg payload,
+ struct brw_reg surface,
+ unsigned msg_length)
+{
+ const struct gen_device_info *devinfo = p->devinfo;
+ const unsigned sfid = GEN7_SFID_DATAPORT_DATA_CACHE;
+ const bool align1 = brw_inst_access_mode(devinfo, p->current) == BRW_ALIGN_1;
+ /* Mask out unused components -- See comment in brw_untyped_atomic(). */
+ const unsigned mask = devinfo->gen == 7 && !devinfo->is_haswell && !align1 ?
+ WRITEMASK_X : WRITEMASK_XYZW;
+ struct brw_inst *insn = brw_send_indirect_scattered_message(
+ p, sfid, brw_writemask(brw_null_reg(), mask),
+ payload, surface, msg_length, 0, align1);
+
+ brw_set_dp_byte_scattered_write(p, insn);
+
+}
+
+static void
brw_set_dp_typed_atomic_message(struct brw_codegen *p,
struct brw_inst *insn,
unsigned atomic_op,
diff --git a/src/intel/compiler/brw_fs.cpp b/src/intel/compiler/brw_fs.cpp
index e5ad553c11b..3e3c372a848 100644
--- a/src/intel/compiler/brw_fs.cpp
+++ b/src/intel/compiler/brw_fs.cpp
@@ -250,6 +250,7 @@ fs_inst::is_send_from_grf() const
case SHADER_OPCODE_UNTYPED_ATOMIC:
case SHADER_OPCODE_UNTYPED_SURFACE_READ:
case SHADER_OPCODE_UNTYPED_SURFACE_WRITE:
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE:
case SHADER_OPCODE_TYPED_ATOMIC:
case SHADER_OPCODE_TYPED_SURFACE_READ:
case SHADER_OPCODE_TYPED_SURFACE_WRITE:
@@ -743,6 +744,7 @@ fs_inst::components_read(unsigned i) const
case SHADER_OPCODE_UNTYPED_SURFACE_WRITE_LOGICAL:
case SHADER_OPCODE_TYPED_SURFACE_WRITE_LOGICAL:
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE_LOGICAL:
assert(src[3].file == IMM &&
src[4].file == IMM);
/* Surface coordinates. */
@@ -796,6 +798,7 @@ fs_inst::size_read(int arg) const
case SHADER_OPCODE_TYPED_SURFACE_READ:
case SHADER_OPCODE_TYPED_SURFACE_WRITE:
case FS_OPCODE_INTERPOLATE_AT_PER_SLOT_OFFSET:
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE:
if (arg == 0)
return mlen * REG_SIZE;
break;
@@ -4529,6 +4532,12 @@ fs_visitor::lower_logical_sends()
ibld.sample_mask_reg());
break;
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE_LOGICAL:
+ lower_surface_logical_send(ibld, inst,
+ SHADER_OPCODE_BYTE_SCATTERED_WRITE,
+ ibld.sample_mask_reg());
+ break;
+
case SHADER_OPCODE_UNTYPED_ATOMIC_LOGICAL:
lower_surface_logical_send(ibld, inst,
SHADER_OPCODE_UNTYPED_ATOMIC,
@@ -5013,6 +5022,7 @@ get_lowered_simd_width(const struct gen_device_info *devinfo,
case SHADER_OPCODE_UNTYPED_ATOMIC_LOGICAL:
case SHADER_OPCODE_UNTYPED_SURFACE_READ_LOGICAL:
case SHADER_OPCODE_UNTYPED_SURFACE_WRITE_LOGICAL:
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE_LOGICAL:
return MIN2(16, inst->exec_size);
case SHADER_OPCODE_URB_READ_SIMD8:
diff --git a/src/intel/compiler/brw_fs_copy_propagation.cpp b/src/intel/compiler/brw_fs_copy_propagation.cpp
index 05691b4369e..71a880e3996 100644
--- a/src/intel/compiler/brw_fs_copy_propagation.cpp
+++ b/src/intel/compiler/brw_fs_copy_propagation.cpp
@@ -655,6 +655,7 @@ fs_visitor::try_constant_propagate(fs_inst *inst, acp_entry *entry)
case SHADER_OPCODE_TYPED_ATOMIC:
case SHADER_OPCODE_TYPED_SURFACE_READ:
case SHADER_OPCODE_TYPED_SURFACE_WRITE:
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE:
/* We only propagate into the surface argument of the
* instruction. Everything else goes through LOAD_PAYLOAD.
*/
@@ -694,6 +695,7 @@ fs_visitor::try_constant_propagate(fs_inst *inst, acp_entry *entry)
case SHADER_OPCODE_TYPED_ATOMIC_LOGICAL:
case SHADER_OPCODE_TYPED_SURFACE_READ_LOGICAL:
case SHADER_OPCODE_TYPED_SURFACE_WRITE_LOGICAL:
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE_LOGICAL:
inst->src[i] = val;
progress = true;
break;
diff --git a/src/intel/compiler/brw_fs_generator.cpp b/src/intel/compiler/brw_fs_generator.cpp
index e0bd191ea7e..90daf38e6b4 100644
--- a/src/intel/compiler/brw_fs_generator.cpp
+++ b/src/intel/compiler/brw_fs_generator.cpp
@@ -2049,6 +2049,11 @@ fs_generator::generate_code(const cfg_t *cfg, int dispatch_width)
inst->mlen, src[2].ud);
break;
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE:
+ assert(src[2].file == BRW_IMMEDIATE_VALUE);
+ brw_byte_scattered_write(p, src[0], src[1], inst->mlen);
+ break;
+
case SHADER_OPCODE_TYPED_ATOMIC:
assert(src[2].file == BRW_IMMEDIATE_VALUE);
brw_typed_atomic(p, dst, src[0], src[1],
diff --git a/src/intel/compiler/brw_fs_surface_builder.cpp b/src/intel/compiler/brw_fs_surface_builder.cpp
index 8990a5ca710..19e218c4b00 100644
--- a/src/intel/compiler/brw_fs_surface_builder.cpp
+++ b/src/intel/compiler/brw_fs_surface_builder.cpp
@@ -1192,3 +1192,20 @@ namespace brw {
}
}
}
+
+namespace brw {
+ namespace scattered_access {
+ void
+ emit_byte_scattered_write(const fs_builder &bld, const fs_reg &surface,
+ const fs_reg &addr, const fs_reg &src,
+ unsigned dims, unsigned size,
+ brw_predicate pred)
+ {
+ using namespace surface_access;
+
+ emit_send(bld, SHADER_OPCODE_BYTE_SCATTERED_WRITE_LOGICAL,
+ addr, src, surface, dims, size, 0, pred);
+ }
+
+ }
+}
diff --git a/src/intel/compiler/brw_fs_surface_builder.h b/src/intel/compiler/brw_fs_surface_builder.h
index 32b56d387f6..913ffaee72c 100644
--- a/src/intel/compiler/brw_fs_surface_builder.h
+++ b/src/intel/compiler/brw_fs_surface_builder.h
@@ -84,5 +84,14 @@ namespace brw {
unsigned surf_dims, unsigned arr_dims,
unsigned rsize, unsigned op);
}
+
+ namespace scattered_access {
+ void
+ emit_byte_scattered_write(const fs_builder &bld, const fs_reg &surface,
+ const fs_reg &addr, const fs_reg &src,
+ unsigned dims, unsigned size,
+ brw_predicate pred = BRW_PREDICATE_NONE);
+
+ }
}
#endif
diff --git a/src/intel/compiler/brw_shader.cpp b/src/intel/compiler/brw_shader.cpp
index f22e204e262..cd6dcb96160 100644
--- a/src/intel/compiler/brw_shader.cpp
+++ b/src/intel/compiler/brw_shader.cpp
@@ -297,6 +297,11 @@ brw_instruction_name(const struct gen_device_info *devinfo, enum opcode op)
case SHADER_OPCODE_MEMORY_FENCE:
return "memory_fence";
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE:
+ return "byte_scattered_write";
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE_LOGICAL:
+ return "byte_scattered_write_logical";
+
case SHADER_OPCODE_LOAD_PAYLOAD:
return "load_payload";
case FS_OPCODE_PACK:
@@ -995,6 +1000,8 @@ backend_instruction::has_side_effects() const
case SHADER_OPCODE_GEN4_SCRATCH_WRITE:
case SHADER_OPCODE_UNTYPED_SURFACE_WRITE:
case SHADER_OPCODE_UNTYPED_SURFACE_WRITE_LOGICAL:
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE:
+ case SHADER_OPCODE_BYTE_SCATTERED_WRITE_LOGICAL:
case SHADER_OPCODE_TYPED_ATOMIC:
case SHADER_OPCODE_TYPED_ATOMIC_LOGICAL:
case SHADER_OPCODE_TYPED_SURFACE_WRITE:
--
2.11.0
More information about the mesa-dev
mailing list