Mirrors/Ryujinx
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Optimize LSRA (#2563)

Browse source 
* Optimize `TryAllocateRegWithtoutSpill` a bit

* Add a fast path for when all registers are live.
* Do not query `GetOverlapPosition` if the register is already in use
  (i.e: free position is 0).

* Do not allocate child split list if not parent

* Turn `LiveRange` into a reference struct

`LiveRange` is now a reference wrapping struct like `Operand` and
`Operation`.

It has also been changed into a singly linked-list. In micro-benchmarks
traversing the linked-list was faster than binary search on `List<T>`.
Even for quite large input sizes (e.g: 1,000,000), surprisingly.

Could be because the code gen for traversing the linked-list is much
much cleaner and there is no virtual dispatch happening when checking if
intervals overlaps.

* Turn `LiveInterval` into an iterator

The LSRA allocates in forward order and never inspect previous
`LiveInterval` once they are expired. Something similar can be done for
the `LiveRange`s within the `LiveInterval`s themselves.

The `LiveInterval` is turned into a iterator which expires `LiveRange`
within it. The iterator is moved forward along with interval walking
code, i.e: AllocateInterval(context, interval, cIndex).

* Remove `LinearScanAllocator.Sources`

Local methods are less susceptible to do allocations than lambdas.

* Optimize `GetOverlapPosition(interval)` a bit

Time complexity should be in O(n+m) instead of O(nm) now.

* Optimize `NumberLocals` a bit

Use the same idea as in `HybridAllocator` to store the visited state
in the MSB of the Operand's value instead of using a `HashSet<T>`.

* Optimize `InsertSplitCopies` a bit

Avoid allocating a redundant `CopyResolver`.

* Optimize `InsertSplitCopiesAtEdges` a bit

Avoid redundant allocations of `CopyResolver`.

* Use stack allocation for `freePositions`

Avoid redundant computations.

* Add `UseList`

Replace `SortedIntegerList` with an even more specialized data
structure. It allocates memory on the arena allocators and does not
require copying use positions when splitting it.

* Turn `LiveInterval` into a reference struct

`LiveInterval` is now a reference wrapping struct like `Operand` and
`Operation`.

The rationale behind turning this in a reference wrapping struct is
because a `LiveInterval` is associated with each local variable, and
these intervals may themselves be split further. I've seen translations
having up to 8000 local variables.

To make the `LiveInterval` unmanaged, a new data structure called
`LiveIntervalList` was added to store child splits. This differs from
`SortedList<,>` because it can contain intervals with the same start
position.

Really wished we got some more of C++ template in C#. :^(

* Optimize `GetChildSplit` a bit

No need to inspect the remaining ranges if we've reached a range which
starts after position, since the split list is ordered.

* Optimize `CopyResolver` a bit

Lazily allocate the fill, spill and parallel copy structures since most
of the time only one of them is needed.

* Optimize `BitMap.Enumerator` a bit

Marking `MoveNext` as `AggressiveInlining` allows RyuJIT to promote the
`Enumerator` struct into registers completely, reducing load/store code
a lot since it does not have to store the struct on the stack for ABI
purposes.

* Use stack allocation for `use/blockedPositions`

* Optimize `AllocateWithSpill` a bit

* Address feedback

* Make `LiveInterval.AddRange(,)` more conservative

Produces no diff against master, but just for good measure.
This commit is contained in:
FICTURE7 2021-10-09 01:15:44 +04:00 • committed by GitHub
parent c54a14d0b8
commit 69093cf2d6
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
9 changed files with 675 additions and 496 deletions
Download patch file Download diff file Expand all files Collapse all files

51
ARMeilleure/CodeGen/RegisterAllocators/CopyResolver.cs
View file

@ -1,6 +1,7 @@
using ARMeilleure.IntermediateRepresentation;
using System;
using System.Collections.Generic;
using static ARMeilleure.IntermediateRepresentation.Operand.Factory;
using static ARMeilleure.IntermediateRepresentation.Operation.Factory;
@ -25,7 +26,7 @@ namespace ARMeilleure.CodeGen.RegisterAllocators
}
}
private List<Copy> _copies;
private readonly List<Copy> _copies;
public int Count => _copies.Count;
@ -146,21 +147,12 @@ namespace ARMeilleure.CodeGen.RegisterAllocators
}
}
private Queue<Operation> _fillQueue = new Queue<Operation>();
private Queue<Operation> _spillQueue = new Queue<Operation>();
private ParallelCopy _parallelCopy;
private Queue<Operation> _fillQueue = null;
private Queue<Operation> _spillQueue = null;
private ParallelCopy _parallelCopy = null;
public bool HasCopy { get; private set; }
public CopyResolver()
{
_fillQueue = new Queue<Operation>();
_spillQueue = new Queue<Operation>();
_parallelCopy = new ParallelCopy();
}
public void AddSplit(LiveInterval left, LiveInterval right)
{
if (left.Local != right.Local)
@ -194,8 +186,12 @@ namespace ARMeilleure.CodeGen.RegisterAllocators
private void AddSplitFill(LiveInterval left, LiveInterval right, OperandType type)
{
Operand register = GetRegister(right.Register, type);
if (_fillQueue == null)
{
_fillQueue = new Queue<Operation>();
}
Operand register = GetRegister(right.Register, type);
Operand offset = Const(left.SpillOffset);
_fillQueue.Enqueue(Operation(Instruction.Fill, register, offset));
@ -205,8 +201,12 @@ namespace ARMeilleure.CodeGen.RegisterAllocators
private void AddSplitSpill(LiveInterval left, LiveInterval right, OperandType type)
{
Operand offset = Const(right.SpillOffset);
if (_spillQueue == null)
{
_spillQueue = new Queue<Operation>();
}
Operand offset = Const(right.SpillOffset);
Operand register = GetRegister(left.Register, type);
_spillQueue.Enqueue(Operation(Instruction.Spill, default, offset, register));
@ -216,6 +216,11 @@ namespace ARMeilleure.CodeGen.RegisterAllocators
private void AddSplitCopy(LiveInterval left, LiveInterval right, OperandType type)
{
if (_parallelCopy == null)
{
_parallelCopy = new ParallelCopy();
}
_parallelCopy.AddCopy(right.Register, left.Register, type);
HasCopy = true;
@ -225,16 +230,22 @@ namespace ARMeilleure.CodeGen.RegisterAllocators
{
List<Operation> sequence = new List<Operation>();
while (_spillQueue.TryDequeue(out Operation spillOp))
if (_spillQueue != null)
{
sequence.Add(spillOp);
while (_spillQueue.TryDequeue(out Operation spillOp))
{
sequence.Add(spillOp);
}
}
_parallelCopy.Sequence(sequence);
_parallelCopy?.Sequence(sequence);
while (_fillQueue.TryDequeue(out Operation fillOp))
if (_fillQueue != null)
{
sequence.Add(fillOp);
while (_fillQueue.TryDequeue(out Operation fillOp))
{
sequence.Add(fillOp);
}
}
return sequence.ToArray();