[Mesa-dev] [PATCH 3/8] i965/fs: Implement integer quotient and remainder math operations.
Kenneth Graunke
kenneth at whitecape.org
Wed Sep 7 07:39:12 PDT 2011
Signed-off-by: Kenneth Graunke <kenneth at whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick at intel.com> [v1]
Reviewed-by: Eric Anholt <eric at anholt.net> [v1]
---
src/mesa/drivers/dri/i965/brw_defines.h | 2 ++
src/mesa/drivers/dri/i965/brw_eu_emit.c | 2 ++
src/mesa/drivers/dri/i965/brw_fs.cpp | 12 +++++++++++-
src/mesa/drivers/dri/i965/brw_fs.h | 2 ++
src/mesa/drivers/dri/i965/brw_fs_emit.cpp | 2 ++
.../dri/i965/brw_fs_schedule_instructions.cpp | 2 ++
src/mesa/drivers/dri/i965/brw_fs_visitor.cpp | 10 ++++++++--
src/mesa/drivers/dri/i965/brw_shader.cpp | 4 ++++
8 files changed, 33 insertions(+), 3 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index 5f34939..4e127f3 100644
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -622,6 +622,8 @@ enum opcode {
SHADER_OPCODE_EXP2,
SHADER_OPCODE_LOG2,
SHADER_OPCODE_POW,
+ SHADER_OPCODE_INT_QUOTIENT,
+ SHADER_OPCODE_INT_REMAINDER,
SHADER_OPCODE_SIN,
SHADER_OPCODE_COS,
FS_OPCODE_DDX,
diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c
index f9f8d49..9068b14 100644
--- a/src/mesa/drivers/dri/i965/brw_eu_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c
@@ -1479,6 +1479,7 @@ void brw_math( struct brw_compile *p,
assert(!src.abs);
if (function != BRW_MATH_FUNCTION_INT_DIV_QUOTIENT &&
+ function != BRW_MATH_FUNCTION_INT_DIV_REMAINDER &&
function != BRW_MATH_FUNCTION_INT_DIV_QUOTIENT_AND_REMAINDER) {
assert(src.type == BRW_REGISTER_TYPE_F);
}
@@ -1539,6 +1540,7 @@ void brw_math2(struct brw_compile *p,
assert(src1.hstride == BRW_HORIZONTAL_STRIDE_1);
if (function != BRW_MATH_FUNCTION_INT_DIV_QUOTIENT &&
+ function != BRW_MATH_FUNCTION_INT_DIV_REMAINDER &&
function != BRW_MATH_FUNCTION_INT_DIV_QUOTIENT_AND_REMAINDER) {
assert(src0.type == BRW_REGISTER_TYPE_F);
assert(src1.type == BRW_REGISTER_TYPE_F);
diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
index 83737bc..6d6f26c 100644
--- a/src/mesa/drivers/dri/i965/brw_fs.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
@@ -152,6 +152,8 @@ fs_visitor::implied_mrf_writes(fs_inst *inst)
case SHADER_OPCODE_COS:
return 1 * c->dispatch_width / 8;
case SHADER_OPCODE_POW:
+ case SHADER_OPCODE_INT_QUOTIENT:
+ case SHADER_OPCODE_INT_REMAINDER:
return 2 * c->dispatch_width / 8;
case FS_OPCODE_TEX:
case FS_OPCODE_TXB:
@@ -575,7 +577,15 @@ fs_visitor::emit_math(enum opcode opcode, fs_reg dst, fs_reg src0, fs_reg src1)
int base_mrf = 2;
fs_inst *inst;
- assert(opcode == SHADER_OPCODE_POW);
+ switch (opcode) {
+ case SHADER_OPCODE_POW:
+ case SHADER_OPCODE_INT_QUOTIENT:
+ case SHADER_OPCODE_INT_REMAINDER:
+ break;
+ default:
+ assert(!"not reached: unsupported binary math opcode");
+ return NULL;
+ }
if (intel->gen >= 6) {
/* Can't do hstride == 0 args to gen6 math, so expand it out.
diff --git a/src/mesa/drivers/dri/i965/brw_fs.h b/src/mesa/drivers/dri/i965/brw_fs.h
index 6f0c802..dbe4471 100644
--- a/src/mesa/drivers/dri/i965/brw_fs.h
+++ b/src/mesa/drivers/dri/i965/brw_fs.h
@@ -304,6 +304,8 @@ public:
opcode == SHADER_OPCODE_LOG2 ||
opcode == SHADER_OPCODE_SIN ||
opcode == SHADER_OPCODE_COS ||
+ opcode == SHADER_OPCODE_INT_QUOTIENT ||
+ opcode == SHADER_OPCODE_INT_REMAINDER ||
opcode == SHADER_OPCODE_POW);
}
diff --git a/src/mesa/drivers/dri/i965/brw_fs_emit.cpp b/src/mesa/drivers/dri/i965/brw_fs_emit.cpp
index dc6a211..a24f202 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_emit.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_emit.cpp
@@ -796,6 +796,8 @@ fs_visitor::generate_code()
generate_math1_gen4(inst, dst, src[0]);
}
break;
+ case SHADER_OPCODE_INT_QUOTIENT:
+ case SHADER_OPCODE_INT_REMAINDER:
case SHADER_OPCODE_POW:
if (intel->gen >= 6) {
generate_math2_gen6(inst, dst, src[0], src[1]);
diff --git a/src/mesa/drivers/dri/i965/brw_fs_schedule_instructions.cpp b/src/mesa/drivers/dri/i965/brw_fs_schedule_instructions.cpp
index 1f83ee2..910f329 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_schedule_instructions.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_schedule_instructions.cpp
@@ -75,11 +75,13 @@ public:
case SHADER_OPCODE_RSQ:
this->latency = 2 * chans * math_latency;
break;
+ case SHADER_OPCODE_INT_QUOTIENT:
case SHADER_OPCODE_SQRT:
case SHADER_OPCODE_LOG2:
/* full precision log. partial is 2. */
this->latency = 3 * chans * math_latency;
break;
+ case SHADER_OPCODE_INT_REMAINDER:
case SHADER_OPCODE_EXP2:
/* full precision. partial is 3, same throughput. */
this->latency = 4 * chans * math_latency;
diff --git a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
index fc9d0e7..4c2a449 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
@@ -294,10 +294,16 @@ fs_visitor::visit(ir_expression *ir)
}
break;
case ir_binop_div:
- assert(!"not reached: should be handled by ir_div_to_mul_rcp");
+ /* Floating point should be lowered by DIV_TO_MUL_RCP in the compiler. */
+ assert(ir->type->is_integer());
+
+ emit_math(SHADER_OPCODE_INT_QUOTIENT, this->result, op[0], op[1]);
break;
case ir_binop_mod:
- assert(!"ir_binop_mod should have been converted to b * fract(a/b)");
+ /* Floating point should be lowered by MOD_TO_FRACT in the compiler. */
+ assert(ir->type->is_integer());
+
+ emit_math(SHADER_OPCODE_INT_REMAINDER, this->result, op[0], op[1]);
break;
case ir_binop_less:
diff --git a/src/mesa/drivers/dri/i965/brw_shader.cpp b/src/mesa/drivers/dri/i965/brw_shader.cpp
index a6ed810..c938c75 100644
--- a/src/mesa/drivers/dri/i965/brw_shader.cpp
+++ b/src/mesa/drivers/dri/i965/brw_shader.cpp
@@ -227,6 +227,10 @@ brw_math_function(enum opcode op)
return BRW_MATH_FUNCTION_SIN;
case SHADER_OPCODE_COS:
return BRW_MATH_FUNCTION_COS;
+ case SHADER_OPCODE_INT_QUOTIENT:
+ return BRW_MATH_FUNCTION_INT_DIV_QUOTIENT;
+ case SHADER_OPCODE_INT_REMAINDER:
+ return BRW_MATH_FUNCTION_INT_DIV_REMAINDER;
default:
assert(!"not reached: unknown math function");
return 0;
--
1.7.6.1
More information about the mesa-dev
mailing list