Drivers will probably want to be able to take UBO references in a
shader like:
uniform ubo1 {
float a;
float b;
float c;
float d;
}
void main() {
gl_FragColor = vec4(a, b, c, d);
}
and generate a single aligned vec4 load out of the UBO. For intel,
this involves recognizing the shared offset of the aligned loads and
CSEing them out. Obviously that involves breaking things down to
loads from an offset from a particular UBO first. Thus, the driver
doesn't want to see
variable_ref(ir_variable(a)),
and even more so does it not want to see
array_ref(record_ref(variable_ref(ir_variable(a)),
field1), variable_ref(ir_variable(i))).
where a.field1[i] is a row_major matrix.
Instead, we're going to make a lowering pass to break UBO references
down to expressions that are obvious to codegen, and amenable to
merging through CSE.
Reviewed-by: Ian Romanick ian.d.roman...@intel.com
---
src/glsl/ir.cpp |1 +
src/glsl/ir.h| 11 ++-
src/glsl/ir_validate.cpp |7 +++
src/mesa/drivers/dri/i965/brw_fs_channel_expressions.cpp |5 +
src/mesa/drivers/dri/i965/brw_fs_visitor.cpp |4
src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp |4
src/mesa/program/ir_to_mesa.cpp |4
src/mesa/state_tracker/st_glsl_to_tgsi.cpp |4
8 files changed, 39 insertions(+), 1 deletion(-)
diff --git a/src/glsl/ir.cpp b/src/glsl/ir.cpp
index b0e38d8..f59cdd2 100644
--- a/src/glsl/ir.cpp
+++ b/src/glsl/ir.cpp
@@ -480,6 +480,7 @@ static const char *const operator_strs[] = {
min,
max,
pow,
+ ubo_load,
vector,
};
diff --git a/src/glsl/ir.h b/src/glsl/ir.h
index f019837..2807ba6 100644
--- a/src/glsl/ir.h
+++ b/src/glsl/ir.h
@@ -1018,9 +1018,18 @@ enum ir_expression_operation {
ir_binop_pow,
/**
+* Load a value the size of a given GLSL type from a uniform block.
+*
+* operand0 is the uniform block index in the linked shader.
+* operand1 is a constant or variable byte offset within the
+* uniform block.
+*/
+ ir_binop_ubo_load,
+
+ /**
* A sentinel marking the last of the binary operations.
*/
- ir_last_binop = ir_binop_pow,
+ ir_last_binop = ir_binop_ubo_load,
ir_quadop_vector,
diff --git a/src/glsl/ir_validate.cpp b/src/glsl/ir_validate.cpp
index 191d398..af0b576 100644
--- a/src/glsl/ir_validate.cpp
+++ b/src/glsl/ir_validate.cpp
@@ -423,6 +423,13 @@ ir_validate::visit_leave(ir_expression *ir)
assert(ir-operands[0]-type == ir-operands[1]-type);
break;
+ case ir_binop_ubo_load:
+ assert(ir-operands[0]-as_constant());
+ assert(ir-operands[0]-type == glsl_type::uint_type);
+
+ assert(ir-operands[1]-type == glsl_type::uint_type);
+ break;
+
case ir_quadop_vector:
/* The vector operator collects some number of scalars and generates a
* vector from them.
diff --git a/src/mesa/drivers/dri/i965/brw_fs_channel_expressions.cpp
b/src/mesa/drivers/dri/i965/brw_fs_channel_expressions.cpp
index 983d92e..58521ee 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_channel_expressions.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_channel_expressions.cpp
@@ -337,6 +337,11 @@ ir_channel_expressions_visitor::visit_leave(ir_assignment
*ir)
case ir_unop_noise:
assert(!noise should have been broken down to function call);
break;
+
+ case ir_binop_ubo_load:
+ assert(!not yet supported);
+ break;
+
case ir_quadop_vector:
assert(!should have been lowered);
break;
diff --git a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
index fefe2c7..34f00ca 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
@@ -571,6 +571,10 @@ fs_visitor::visit(ir_expression *ir)
else
inst = emit(BRW_OPCODE_SHR, this-result, op[0], op[1]);
break;
+
+ case ir_binop_ubo_load:
+ assert(!not yet supported);
+ break;
}
}
diff --git a/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
b/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
index c77dc91..d6a786f 100644
--- a/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
@@ -1314,6 +1314,10 @@ vec4_visitor::visit(ir_expression *ir)
inst = emit(BRW_OPCODE_SHR, result_dst, op[0], op[1]);
break;
+ case ir_binop_ubo_load:
+ assert(!not yet supported);
+ break;
+
case ir_quadop_vector:
assert(!not reached: should be handled by lower_quadop_vector);
break;
diff --git a/src/mesa/program/ir_to_mesa.cpp b/src/mesa/program/ir_to_mesa.cpp
index 255a8a7..70c4cc8 100644
---
