Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add design doc for atomic<T> type. #5101

Merged
merged 5 commits into from
Sep 19, 2024
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
116 changes: 116 additions & 0 deletions docs/proposals/003-atomic-t.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,116 @@
SP #003 - `Atomic<T>` type
==============


Status
------

Author: Yong He

Status: Design Discussion.

Implementation: N/A

Reviewed by: N/A

Background
----------

HLSL defines atomic intrinsics to work on free references to ordinary values such as `int` and `float`. However, this doesn't translate well to Metal and WebGPU,
which defines `atomic<T>` type and only allow atomic operations to be applied on values of `atomic<T>` types.

Slang's Metal backend follows the same technique in SPIRV-Cross and DXIL->Metal converter that relies on a C++ undefined behavior that casts an ordinary `int*` pointer to a `atomic<int>*` pointer
and then call atomic intrinsic on the reinterpreted pointer. This is fragile and not guaranteed to work in the future.

To make the situation worse, WebGPU bans all possible ways to cast a normal pointer into an `atomic` pointer. In order to provide a truly portable way to define
atomic operations and allow them to be translatable to all targets, we will also need an `atomic<T>` type in Slang that maps to `atomic<T>` in WGSL and Metal, and maps to
`T` for HLSL/SPIRV.


Proposed Approach
-----------------

We define an `Atomic<T>` type that functions as a wrapper of `T` and provides atomic operations:
```
[sealed] interface IAtomicable {}
[sealed] interface IArithmeticAtomicable : IAtomicable {}
[sealed] interface IBitAtomicable : IArithmeticAtomicable {}

extension int : IArithmeticAtomicable {}
extension uint : IArithmeticAtomicable {}
extension int64_t : IBitAtomicable {}
extension uint64_t : IBitAtomicable {}
extension float : IArithmeticAtomicable {}
extension half : IArithmeticAtomicable {}

struct Atomic<T : IAtomicable>
{
T load();

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do you intend to support syntactic sugar where you can use the Atomic as an lvalue/rvalue and assign without using load/store? This is pretty common.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes we will need that.

[ref] void store(T newValue); // Question: do we really need this?

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, you surely want this, atomic stores can often be lowered to a much cheaper hardware instruction than whatever you would replace it with (e.g. atomic exchange).

[ref] T exchange(T newValue); // returns old value
[ref] T compareExchange(T compareValue, T newValue); // returns old value.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are sometimes two forms of compareExchange (strong vs weak). Should say which this is.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

According to SPIRV spec, the OpAtomicCompareExchangeWeak is deprecated and we only have OpAtomicCompareExchange, so perhaps it is fine to just assume it means weak here?

}

extension<T:IArithmeticAtomicable> Atomic<T>
{
[ref] T atomicAdd(T value); // returns original value
[ref] T atomicSub(T value); // returns original value
[ref] T atomicMax(T value); // returns original value
[ref] T atomicMin(T value); // returns original value
[ref] T atomicIncrement();
[ref] T atomicDecrement();
}

extension<T:IBitAtomicable> Atomic<T>
{
[ref] T atomicAnd(T value); // returns original value
[ref] T atomicOr(T value); // returns original value
[ref] T atomicXor(T value); // returns original value
}
```

We allow `Atomic<T>` to be defined anywhere: as struct fields, as array elements, as elements of `RWStructuredBuffer` types,
or as local, global and groupshared variable types or function parameter types. For example, in global memory:

```hlsl
struct MyType
{
int ordinaryValue;
Atomic<int> atomicValue;
}

RWStructuredBuffer<MyType> atomicBuffer;

void main()
{
atomicBuffer[0].atomicValue.atomicAdd(1);
printf("%d", atomicBuffer[0].atomicValue.load());
}
```

In groupshared memory:

```hlsl
void main()
{
groupshared atomic<int> c;
c.atomicAdd(1);
}
```

When generating WGSL code where `atomic<T>` isn't allowed on local variables or other illegal address spaces, we will lower the type
into its underlying type. The use of atomic type in these positions will simply have no meaning. A caveat is what the semantics should be
when there is a function that takes `inout Atomic<T>` as parameter. This likely need to be a warning or error.

This should be handled by a legalization pass similar to `lowerBufferElementTypeToStorageType` but operates
in the opposite direction: the "loaded" value from a buffer is converted into an atomic-free type, and storing a value leads to an
atomic store at the corresponding locations.

For non-WGSL/Metal targets, we can simply lower the type out of existence into its underlying type.

# Related Work

`Atomic<T>` type exists in almost all CPU programming languages and is the proven way to express atomic operations over different
architectures that have different memory models. WGSL and Metal follows this trend to require atomic operations being expressed
this way. This proposal is to make Slang follow this trend and make `Atomic<T>` the recommened way to express atomic operation
going forward.