T0b1/wasmtime
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[RFC] Dynamic Vector Support (#4200)

Browse Source 
Introduce a new concept in the IR that allows a producer to create
dynamic vector types. An IR function can now contain global value(s)
that represent a dynamic scaling factor, for a given fixed-width
vector type. A dynamic type is then created by 'multiplying' the
corresponding global value with a fixed-width type. These new types
can be used just like the existing types and the type system has a
set of hard-coded dynamic types, such as I32X4XN, which the user
defined types map onto. The dynamic types are also used explicitly
to create dynamic stack slots, which have no set size like their
existing counterparts. New IR instructions are added to access these
new stack entities.

Currently, during codegen, the dynamic scaling factor has to be
lowered to a constant so the dynamic slots do eventually have a
compile-time known size, as do spill slots.

The current lowering for aarch64 just targets Neon, using a dynamic
scale of 1.

Copyright (c) 2022, Arm Limited.
This commit is contained in:
Sam Parker
2022-07-07 20:54:39 +01:00
committed by GitHub
parent 9ae060a12a
commit 9c43749dfe
69 changed files with 2422 additions and 294 deletions
Download Patch File Download Diff File Expand all files Collapse all files

70
cranelift/codegen/src/ir/instructions.rs
View File

@@ -633,6 +633,8 @@ pub struct ValueTypeSet {
pub bools: BitSet8,
/// Allowed ref widths
pub refs: BitSet8,
/// Allowed dynamic vectors minimum lane sizes
pub dynamic_lanes: BitSet16,
}
impl ValueTypeSet {
@@ -656,8 +658,13 @@ impl ValueTypeSet {
/// Does `typ` belong to this set?
pub fn contains(self, typ: Type) -> bool {
let l2l = typ.log2_lane_count();
self.lanes.contains(l2l) && self.is_base_type(typ.lane_type())
if typ.is_dynamic_vector() {
let l2l = typ.log2_min_lane_count();
self.dynamic_lanes.contains(l2l) && self.is_base_type(typ.lane_type())
} else {
let l2l = typ.log2_lane_count();
self.lanes.contains(l2l) && self.is_base_type(typ.lane_type())
}
}
/// Get an example member of this type set.
@@ -712,6 +719,9 @@ enum OperandConstraint {
/// This operand is `ctrlType.merge_lanes()`.
MergeLanes,
/// This operands is `ctrlType.dynamic_to_vector()`.
DynamicToVector,
}
impl OperandConstraint {
@@ -738,15 +748,48 @@ impl OperandConstraint {
.expect("invalid type for half_vector"),
),
DoubleVector => Bound(ctrl_type.by(2).expect("invalid type for double_vector")),
SplitLanes => Bound(
SplitLanes => {
if ctrl_type.is_dynamic_vector() {
Bound(
ctrl_type
.dynamic_to_vector()
.expect("invalid type for dynamic_to_vector")
.split_lanes()
.expect("invalid type for split_lanes")
.vector_to_dynamic()
.expect("invalid dynamic type"),
)
} else {
Bound(
ctrl_type
.split_lanes()
.expect("invalid type for split_lanes"),
)
}
}
MergeLanes => {
if ctrl_type.is_dynamic_vector() {
Bound(
ctrl_type
.dynamic_to_vector()
.expect("invalid type for dynamic_to_vector")
.merge_lanes()
.expect("invalid type for merge_lanes")
.vector_to_dynamic()
.expect("invalid dynamic type"),
)
} else {
Bound(
ctrl_type
.merge_lanes()
.expect("invalid type for merge_lanes"),
)
}
}
DynamicToVector => Bound(
ctrl_type
.split_lanes()
.expect("invalid type for split_lanes"),
),
MergeLanes => Bound(
ctrl_type
.merge_lanes()
.expect("invalid type for merge_lanes"),
.dynamic_to_vector()
.expect("invalid type for dynamic_to_vector"),
),
}
}
@@ -860,11 +903,13 @@ mod tests {
floats: BitSet8::from_range(0, 0),
bools: BitSet8::from_range(3, 7),
refs: BitSet8::from_range(5, 7),
dynamic_lanes: BitSet16::from_range(0, 4),
};
assert!(!vts.contains(I8));
assert!(vts.contains(I32));
assert!(vts.contains(I64));
assert!(vts.contains(I32X4));
assert!(vts.contains(I32X4XN));
assert!(!vts.contains(F32));
assert!(!vts.contains(B1));
assert!(vts.contains(B8));
@@ -879,6 +924,7 @@ mod tests {
floats: BitSet8::from_range(5, 7),
bools: BitSet8::from_range(3, 7),
refs: BitSet8::from_range(0, 0),
dynamic_lanes: BitSet16::from_range(0, 8),
};
assert_eq!(vts.example().to_string(), "f32");
@@ -888,6 +934,7 @@ mod tests {
floats: BitSet8::from_range(5, 7),
bools: BitSet8::from_range(3, 7),
refs: BitSet8::from_range(0, 0),
dynamic_lanes: BitSet16::from_range(0, 8),
};
assert_eq!(vts.example().to_string(), "f32x2");
@@ -897,9 +944,11 @@ mod tests {
floats: BitSet8::from_range(0, 0),
bools: BitSet8::from_range(3, 7),
refs: BitSet8::from_range(0, 0),
dynamic_lanes: BitSet16::from_range(0, 8),
};
assert!(!vts.contains(B32X2));
assert!(vts.contains(B32X4));
assert!(vts.contains(B16X4XN));
assert_eq!(vts.example().to_string(), "b32x4");
let vts = ValueTypeSet {
@@ -909,6 +958,7 @@ mod tests {
floats: BitSet8::from_range(0, 0),
bools: BitSet8::from_range(0, 0),
refs: BitSet8::from_range(0, 0),
dynamic_lanes: BitSet16::from_range(0, 8),
};
assert!(vts.contains(I32));
assert!(vts.contains(I32X4));