[Mesa-dev] [PATCH v5 33/70] i965/vec4: Introduce VEC4 IR builder.
Iago Toral Quiroga
itoral at igalia.com
Thu Sep 10 06:35:49 PDT 2015
From: Francisco Jerez <currojerez at riseup.net>
See "i965/fs: Introduce FS IR builder." for the rationale.
v2: Drop scalarizing VEC4 builder.
v3: Take a backend_shader as constructor argument. Improve handling
of debug annotations and execution control flags. Rename "instr"
variable. Initialize cursor to NULL by default and add method to
explicitly point the builder at the end of the program.
---
src/mesa/drivers/dri/i965/Makefile.sources | 1 +
src/mesa/drivers/dri/i965/brw_vec4_builder.h | 602 +++++++++++++++++++++++++++
2 files changed, 603 insertions(+)
create mode 100644 src/mesa/drivers/dri/i965/brw_vec4_builder.h
diff --git a/src/mesa/drivers/dri/i965/Makefile.sources b/src/mesa/drivers/dri/i965/Makefile.sources
index dfdad75..88e64fb 100644
--- a/src/mesa/drivers/dri/i965/Makefile.sources
+++ b/src/mesa/drivers/dri/i965/Makefile.sources
@@ -115,6 +115,7 @@ i965_FILES = \
brw_urb.c \
brw_util.c \
brw_util.h \
+ brw_vec4_builder.h \
brw_vec4_copy_propagation.cpp \
brw_vec4.cpp \
brw_vec4_cse.cpp \
diff --git a/src/mesa/drivers/dri/i965/brw_vec4_builder.h b/src/mesa/drivers/dri/i965/brw_vec4_builder.h
new file mode 100644
index 0000000..a90cadb
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/brw_vec4_builder.h
@@ -0,0 +1,602 @@
+/* -*- c++ -*- */
+/*
+ * Copyright © 2010-2015 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.
+ */
+
+#ifndef BRW_VEC4_BUILDER_H
+#define BRW_VEC4_BUILDER_H
+
+#include "brw_ir_vec4.h"
+#include "brw_ir_allocator.h"
+#include "brw_context.h"
+
+namespace brw {
+ /**
+ * Toolbox to assemble a VEC4 IR program out of individual instructions.
+ *
+ * This object is meant to have an interface consistent with
+ * brw::fs_builder. They cannot be fully interchangeable because
+ * brw::fs_builder generates scalar code while brw::vec4_builder generates
+ * vector code.
+ */
+ class vec4_builder {
+ public:
+ /** Type used in this IR to represent a source of an instruction. */
+ typedef brw::src_reg src_reg;
+
+ /** Type used in this IR to represent the destination of an instruction. */
+ typedef brw::dst_reg dst_reg;
+
+ /** Type used in this IR to represent an instruction. */
+ typedef vec4_instruction instruction;
+
+ /**
+ * Construct a vec4_builder that inserts instructions into \p shader.
+ */
+ vec4_builder(backend_shader *shader) :
+ shader(shader), block(NULL), cursor(NULL),
+ force_writemask_all(false),
+ annotation()
+ {
+ }
+
+ /**
+ * Construct a vec4_builder that inserts instructions into \p shader
+ * before instruction \p inst in basic block \p block. The default
+ * execution controls and debug annotation are initialized from the
+ * instruction passed as argument.
+ */
+ vec4_builder(backend_shader *shader, bblock_t *block, instruction *inst) :
+ shader(shader), block(block), cursor(inst)
+ {
+ annotation.str = inst->annotation;
+ annotation.ir = inst->ir;
+ }
+
+ /**
+ * Construct a vec4_builder that inserts instructions before \p cursor
+ * in basic block \p block, inheriting other code generation parameters
+ * from this.
+ */
+ vec4_builder
+ at(bblock_t *block, exec_node *cursor) const
+ {
+ vec4_builder bld = *this;
+ bld.block = block;
+ bld.cursor = cursor;
+ return bld;
+ }
+
+ /**
+ * Construct a vec4_builder appending instructions at the end of the
+ * instruction list of the shader, inheriting other code generation
+ * parameters from this.
+ */
+ vec4_builder
+ at_end() const
+ {
+ return at(NULL, (exec_node *)&shader->instructions.tail);
+ }
+
+ /**
+ * Construct a builder with per-channel control flow execution masking
+ * disabled if \p b is true. If control flow execution masking is
+ * already disabled this has no effect.
+ */
+ vec4_builder
+ exec_all(bool b = true) const
+ {
+ vec4_builder bld = *this;
+ if (b)
+ bld.force_writemask_all = true;
+ return bld;
+ }
+
+ /**
+ * Construct a builder with the given debug annotation info.
+ */
+ vec4_builder
+ annotate(const char *str, const void *ir = NULL) const
+ {
+ vec4_builder bld = *this;
+ bld.annotation.str = str;
+ bld.annotation.ir = ir;
+ return bld;
+ }
+
+ /**
+ * Get the SIMD width in use.
+ */
+ unsigned
+ dispatch_width() const
+ {
+ return 8;
+ }
+
+ /**
+ * Allocate a virtual register of natural vector size (four for this IR)
+ * and SIMD width. \p n gives the amount of space to allocate in
+ * dispatch_width units (which is just enough space for four logical
+ * components in this IR).
+ */
+ dst_reg
+ vgrf(enum brw_reg_type type, unsigned n = 1) const
+ {
+ assert(dispatch_width() <= 32);
+
+ if (n > 0)
+ return retype(dst_reg(GRF, shader->alloc.allocate(
+ n * DIV_ROUND_UP(type_sz(type), 4))),
+ type);
+ else
+ return retype(null_reg_ud(), type);
+ }
+
+ /**
+ * Create a null register of floating type.
+ */
+ dst_reg
+ null_reg_f() const
+ {
+ return dst_reg(retype(brw_null_vec(dispatch_width()),
+ BRW_REGISTER_TYPE_F));
+ }
+
+ /**
+ * Create a null register of signed integer type.
+ */
+ dst_reg
+ null_reg_d() const
+ {
+ return dst_reg(retype(brw_null_vec(dispatch_width()),
+ BRW_REGISTER_TYPE_D));
+ }
+
+ /**
+ * Create a null register of unsigned integer type.
+ */
+ dst_reg
+ null_reg_ud() const
+ {
+ return dst_reg(retype(brw_null_vec(dispatch_width()),
+ BRW_REGISTER_TYPE_UD));
+ }
+
+ /**
+ * Insert an instruction into the program.
+ */
+ instruction *
+ emit(const instruction &inst) const
+ {
+ return emit(new(shader->mem_ctx) instruction(inst));
+ }
+
+ /**
+ * Create and insert a nullary control instruction into the program.
+ */
+ instruction *
+ emit(enum opcode opcode) const
+ {
+ return emit(instruction(opcode));
+ }
+
+ /**
+ * Create and insert a nullary instruction into the program.
+ */
+ instruction *
+ emit(enum opcode opcode, const dst_reg &dst) const
+ {
+ return emit(instruction(opcode, dst));
+ }
+
+ /**
+ * Create and insert a unary instruction into the program.
+ */
+ instruction *
+ emit(enum opcode opcode, const dst_reg &dst, const src_reg &src0) const
+ {
+ switch (opcode) {
+ case SHADER_OPCODE_RCP:
+ case SHADER_OPCODE_RSQ:
+ case SHADER_OPCODE_SQRT:
+ case SHADER_OPCODE_EXP2:
+ case SHADER_OPCODE_LOG2:
+ case SHADER_OPCODE_SIN:
+ case SHADER_OPCODE_COS:
+ return fix_math_instruction(
+ emit(instruction(opcode, dst,
+ fix_math_operand(src0))));
+
+ default:
+ return emit(instruction(opcode, dst, src0));
+ }
+ }
+
+ /**
+ * Create and insert a binary instruction into the program.
+ */
+ instruction *
+ emit(enum opcode opcode, const dst_reg &dst, const src_reg &src0,
+ const src_reg &src1) const
+ {
+ switch (opcode) {
+ case SHADER_OPCODE_POW:
+ case SHADER_OPCODE_INT_QUOTIENT:
+ case SHADER_OPCODE_INT_REMAINDER:
+ return fix_math_instruction(
+ emit(instruction(opcode, dst,
+ fix_math_operand(src0),
+ fix_math_operand(src1))));
+
+ default:
+ return emit(instruction(opcode, dst, src0, src1));
+ }
+ }
+
+ /**
+ * Create and insert a ternary instruction into the program.
+ */
+ instruction *
+ emit(enum opcode opcode, const dst_reg &dst, const src_reg &src0,
+ const src_reg &src1, const src_reg &src2) const
+ {
+ switch (opcode) {
+ case BRW_OPCODE_BFE:
+ case BRW_OPCODE_BFI2:
+ case BRW_OPCODE_MAD:
+ case BRW_OPCODE_LRP:
+ return emit(instruction(opcode, dst,
+ fix_3src_operand(src0),
+ fix_3src_operand(src1),
+ fix_3src_operand(src2)));
+
+ default:
+ return emit(instruction(opcode, dst, src0, src1, src2));
+ }
+ }
+
+ /**
+ * Insert a preallocated instruction into the program.
+ */
+ instruction *
+ emit(instruction *inst) const
+ {
+ inst->force_writemask_all = force_writemask_all;
+ inst->annotation = annotation.str;
+ inst->ir = annotation.ir;
+
+ if (block)
+ static_cast<instruction *>(cursor)->insert_before(block, inst);
+ else
+ cursor->insert_before(inst);
+
+ return inst;
+ }
+
+ /**
+ * Select \p src0 if the comparison of both sources with the given
+ * conditional mod evaluates to true, otherwise select \p src1.
+ *
+ * Generally useful to get the minimum or maximum of two values.
+ */
+ void
+ emit_minmax(const dst_reg &dst, const src_reg &src0,
+ const src_reg &src1, brw_conditional_mod mod) const
+ {
+ if (shader->devinfo->gen >= 6) {
+ set_condmod(mod, SEL(dst, fix_unsigned_negate(src0),
+ fix_unsigned_negate(src1)));
+ } else {
+ CMP(null_reg_d(), src0, src1, mod);
+ set_predicate(BRW_PREDICATE_NORMAL,
+ SEL(dst, src0, src1));
+ }
+ }
+
+ /**
+ * Copy any live channel from \p src to the first channel of the result.
+ */
+ src_reg
+ emit_uniformize(const src_reg &src) const
+ {
+ const vec4_builder ubld = exec_all();
+ const dst_reg chan_index =
+ writemask(vgrf(BRW_REGISTER_TYPE_UD), WRITEMASK_X);
+ const dst_reg dst = vgrf(src.type);
+
+ ubld.emit(SHADER_OPCODE_FIND_LIVE_CHANNEL, chan_index);
+ ubld.emit(SHADER_OPCODE_BROADCAST, dst, src, src_reg(chan_index));
+
+ return src_reg(dst);
+ }
+
+ /**
+ * Assorted arithmetic ops.
+ * @{
+ */
+#define ALU1(op) \
+ instruction * \
+ op(const dst_reg &dst, const src_reg &src0) const \
+ { \
+ return emit(BRW_OPCODE_##op, dst, src0); \
+ }
+
+#define ALU2(op) \
+ instruction * \
+ op(const dst_reg &dst, const src_reg &src0, const src_reg &src1) const \
+ { \
+ return emit(BRW_OPCODE_##op, dst, src0, src1); \
+ }
+
+#define ALU2_ACC(op) \
+ instruction * \
+ op(const dst_reg &dst, const src_reg &src0, const src_reg &src1) const \
+ { \
+ instruction *inst = emit(BRW_OPCODE_##op, dst, src0, src1); \
+ inst->writes_accumulator = true; \
+ return inst; \
+ }
+
+#define ALU3(op) \
+ instruction * \
+ op(const dst_reg &dst, const src_reg &src0, const src_reg &src1, \
+ const src_reg &src2) const \
+ { \
+ return emit(BRW_OPCODE_##op, dst, src0, src1, src2); \
+ }
+
+ ALU2(ADD)
+ ALU2_ACC(ADDC)
+ ALU2(AND)
+ ALU2(ASR)
+ ALU2(AVG)
+ ALU3(BFE)
+ ALU2(BFI1)
+ ALU3(BFI2)
+ ALU1(BFREV)
+ ALU1(CBIT)
+ ALU2(CMPN)
+ ALU3(CSEL)
+ ALU2(DP2)
+ ALU2(DP3)
+ ALU2(DP4)
+ ALU2(DPH)
+ ALU1(F16TO32)
+ ALU1(F32TO16)
+ ALU1(FBH)
+ ALU1(FBL)
+ ALU1(FRC)
+ ALU2(LINE)
+ ALU1(LZD)
+ ALU2(MAC)
+ ALU2_ACC(MACH)
+ ALU3(MAD)
+ ALU1(MOV)
+ ALU2(MUL)
+ ALU1(NOT)
+ ALU2(OR)
+ ALU2(PLN)
+ ALU1(RNDD)
+ ALU1(RNDE)
+ ALU1(RNDU)
+ ALU1(RNDZ)
+ ALU2(SAD2)
+ ALU2_ACC(SADA2)
+ ALU2(SEL)
+ ALU2(SHL)
+ ALU2(SHR)
+ ALU2_ACC(SUBB)
+ ALU2(XOR)
+
+#undef ALU3
+#undef ALU2_ACC
+#undef ALU2
+#undef ALU1
+ /** @} */
+
+ /**
+ * CMP: Sets the low bit of the destination channels with the result
+ * of the comparison, while the upper bits are undefined, and updates
+ * the flag register with the packed 16 bits of the result.
+ */
+ instruction *
+ CMP(const dst_reg &dst, const src_reg &src0, const src_reg &src1,
+ brw_conditional_mod condition) const
+ {
+ /* Take the instruction:
+ *
+ * CMP null<d> src0<f> src1<f>
+ *
+ * Original gen4 does type conversion to the destination type
+ * before comparison, producing garbage results for floating
+ * point comparisons.
+ *
+ * The destination type doesn't matter on newer generations,
+ * so we set the type to match src0 so we can compact the
+ * instruction.
+ */
+ return set_condmod(condition,
+ emit(BRW_OPCODE_CMP, retype(dst, src0.type),
+ fix_unsigned_negate(src0),
+ fix_unsigned_negate(src1)));
+ }
+
+ /**
+ * Gen4 predicated IF.
+ */
+ instruction *
+ IF(brw_predicate predicate) const
+ {
+ return set_predicate(predicate, emit(BRW_OPCODE_IF));
+ }
+
+ /**
+ * Gen6 IF with embedded comparison.
+ */
+ instruction *
+ IF(const src_reg &src0, const src_reg &src1,
+ brw_conditional_mod condition) const
+ {
+ assert(shader->devinfo->gen == 6);
+ return set_condmod(condition,
+ emit(BRW_OPCODE_IF,
+ null_reg_d(),
+ fix_unsigned_negate(src0),
+ fix_unsigned_negate(src1)));
+ }
+
+ /**
+ * Emit a linear interpolation instruction.
+ */
+ instruction *
+ LRP(const dst_reg &dst, const src_reg &x, const src_reg &y,
+ const src_reg &a) const
+ {
+ if (shader->devinfo->gen >= 6) {
+ /* The LRP instruction actually does op1 * op0 + op2 * (1 - op0), so
+ * we need to reorder the operands.
+ */
+ return emit(BRW_OPCODE_LRP, dst, a, y, x);
+
+ } else {
+ /* We can't use the LRP instruction. Emit x*(1-a) + y*a. */
+ const dst_reg y_times_a = vgrf(dst.type);
+ const dst_reg one_minus_a = vgrf(dst.type);
+ const dst_reg x_times_one_minus_a = vgrf(dst.type);
+
+ MUL(y_times_a, y, a);
+ ADD(one_minus_a, negate(a), src_reg(1.0f));
+ MUL(x_times_one_minus_a, x, src_reg(one_minus_a));
+ return ADD(dst, src_reg(x_times_one_minus_a), src_reg(y_times_a));
+ }
+ }
+
+ backend_shader *shader;
+
+ protected:
+ /**
+ * Workaround for negation of UD registers. See comment in
+ * fs_generator::generate_code() for the details.
+ */
+ src_reg
+ fix_unsigned_negate(const src_reg &src) const
+ {
+ if (src.type == BRW_REGISTER_TYPE_UD && src.negate) {
+ dst_reg temp = vgrf(BRW_REGISTER_TYPE_UD);
+ MOV(temp, src);
+ return src_reg(temp);
+ } else {
+ return src;
+ }
+ }
+
+ /**
+ * Workaround for register access modes not supported by the ternary
+ * instruction encoding.
+ */
+ src_reg
+ fix_3src_operand(const src_reg &src) const
+ {
+ /* Using vec4 uniforms in SIMD4x2 programs is difficult. You'd like to be
+ * able to use vertical stride of zero to replicate the vec4 uniform, like
+ *
+ * g3<0;4,1>:f - [0, 4][1, 5][2, 6][3, 7]
+ *
+ * But you can't, since vertical stride is always four in three-source
+ * instructions. Instead, insert a MOV instruction to do the replication so
+ * that the three-source instruction can consume it.
+ */
+
+ /* The MOV is only needed if the source is a uniform or immediate. */
+ if (src.file != UNIFORM && src.file != IMM)
+ return src;
+
+ if (src.file == UNIFORM && brw_is_single_value_swizzle(src.swizzle))
+ return src;
+
+ const dst_reg expanded = vgrf(src.type);
+ emit(VEC4_OPCODE_UNPACK_UNIFORM, expanded, src);
+ return src_reg(expanded);
+ }
+
+ /**
+ * Workaround for register access modes not supported by the math
+ * instruction.
+ */
+ src_reg
+ fix_math_operand(const src_reg &src) const
+ {
+ /* The gen6 math instruction ignores the source modifiers --
+ * swizzle, abs, negate, and at least some parts of the register
+ * region description.
+ *
+ * Rather than trying to enumerate all these cases, *always* expand the
+ * operand to a temp GRF for gen6.
+ *
+ * For gen7, keep the operand as-is, except if immediate, which gen7 still
+ * can't use.
+ */
+ if (shader->devinfo->gen == 6 ||
+ (shader->devinfo->gen == 7 && src.file == IMM)) {
+ const dst_reg tmp = vgrf(src.type);
+ MOV(tmp, src);
+ return src_reg(tmp);
+ } else {
+ return src;
+ }
+ }
+
+ /**
+ * Workaround other weirdness of the math instruction.
+ */
+ instruction *
+ fix_math_instruction(instruction *inst) const
+ {
+ if (shader->devinfo->gen == 6 &&
+ inst->dst.writemask != WRITEMASK_XYZW) {
+ const dst_reg tmp = vgrf(inst->dst.type);
+ MOV(inst->dst, src_reg(tmp));
+ inst->dst = tmp;
+
+ } else if (shader->devinfo->gen < 6) {
+ const unsigned sources = (inst->src[1].file == BAD_FILE ? 1 : 2);
+ inst->base_mrf = 1;
+ inst->mlen = sources;
+ }
+
+ return inst;
+ }
+
+ bblock_t *block;
+ exec_node *cursor;
+
+ bool force_writemask_all;
+
+ /** Debug annotation info. */
+ struct {
+ const char *str;
+ const void *ir;
+ } annotation;
+ };
+}
+
+#endif
--
1.9.1
More information about the mesa-dev
mailing list