Skip to content

Introduce Subgroup Operations Extension #1459

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.

Sign up for GitHub

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

Closed
wants to merge 1 commit into from
Closed

Introduce Subgroup Operations Extension #1459

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
5 changes: 5 additions & 0 deletions spec/index.bs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1456,6 +1456,7 @@ enum GPUFeatureName {
"pipeline-statistics-query",
"texture-compression-bc",
"timestamp-query",
"subgroup-operations"
};
</script>

Expand Down Expand Up @@ -7706,6 +7707,10 @@ The following enums are supported if and only if the {{GPUFeatureName/"pipeline-
* {{GPUQueryType/"pipeline-statistics"}}
</dl>

## <dfn dfn-type=enum-value dfn-for=GPUFeatureName>subgroup-operations</dfn> ## {#subgroup-operations}

Issue: Define functionality when the {{GPUFeatureName/"subgroup-operations"}} [=feature=] is enabled.


## <dfn dfn-type=enum-value dfn-for=GPUFeatureName>texture-compression-bc</dfn> ## {#texture-compression-bc}

Expand Down
44 changes: 44 additions & 0 deletions wgsl/index.bs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4600,6 +4600,19 @@ See [[#builtin-inputs-outputs]] for how to declare a built-in variable.
<td>vec3&lt;u32&gt;
<td width="50%">The [=workgroup_size=] of the current entry point.

<tr><td>`subgroup_size`
<td>compute
<td>in
<td>u32
<td width="50%">The subgroup size of the current entry point.

<tr><td>`subgroup_invocation_index`
<td>compute
<td>in
<td>u32
<td width="50%">The current invocation's subgroup invocation index.
Must be in range [0, `subgroup_size`-1].

<tr><td>`sample_index`
<td>fragment
<td>in
Expand Down Expand Up @@ -5787,6 +5800,37 @@ reduce a shader's memory bandwidth demand.
See [[#floating-point-conversion]] for edge case behaviour.
</table>

## Subgroup built-in functions ## {#subgroup-builtin-functions}

<table class='data'>
<thead>
<tr><td>Subgroup built-in functions<td>SPIR-V
</thead>
<tr><td>subgroupIsFirst() -&gt; bool<td>OpGroupNonUniformElect
<tr><td>subgroupAll(bool) -&gt; bool<td>OpGroupNonUniformAll
<tr><td>subgroupAny(bool) -&gt; bool<td>OpGroupNonUniformAny
<tr><td>subgroupBallot(bool) -&gt; vec4&lt;u32&gt;<td>OpGroupNonUniformBallot
<tr><td>subgroupBroadcastFirst(Integral) -&gt; Integral<td>OpGroupNonUniformBroadcastFirst
<tr><td>subgroupBroadcastFirst(Floating) -&gt; Floating<td>OpGroupNonUniformBroadcastFirst
<tr><td>subgroupAdd(Integral) -&gt; Integral<td>OpGroupNonUniformIAdd with Reduce
<tr><td>subgroupAdd(Floating) -&gt; Floating<td>OpGroupNonUniformFAdd with Reduce
<tr><td>subgroupMul(Integral) -&gt; Integral<td>OpGroupNonUniformIMul with Reduce
<tr><td>subgroupMul(Floating) -&gt; Floating<td>OpGroupNonUniformFMul with Reduce
<tr><td>subgroupMin(Integral) -&gt; Integral<td>OpGroupNonUniformUMin or OpGroupNonUniformSMin with Reduce
<tr><td>subgroupMin(Floating) -&gt; Floating<td>OpGroupNonUniformFMin with Reduce
<tr><td>subgroupMax(Integral) -&gt; Integral<td>OpGroupNonUniformUMax or OpGroupNonUniformSMax with Reduce
<tr><td>subgroupMax(Floating) -&gt; Floating<td>OpGroupNonUniformFMax with Reduce
<tr><td>subgroupAnd(Integral) -&gt; Integral<td>OpGroupNonUniformBitwiseAnd
<tr><td>subgroupOr(Integral) -&gt; Integral<td>OpGroupNonUniformBitwiseOr
<tr><td>subgroupXor(Integral) -&gt; Integral<td>OpGroupNonUniformBitwiseXor
<tr><td>subgroupPrefixAdd(Integral) -&gt; Integral<td>OpGroupNonUniformIAdd with ExclusiveScan
<tr><td>subgroupPrefixAdd(Floating) -&gt; Floating<td>OpGroupNonUniformFAdd with ExclusiveScan
<tr><td>subgroupPrefixMul(Integral) -&gt; Integral<td>OpGroupNonUniformIMul with ExclusiveScan
<tr><td>subgroupPrefixMul(Floating) -&gt; Floating<td>OpGroupNonUniformFMul with ExclusiveScan
</table>

Note: Subgroup built-in functions exist if "subgroup-operations" is enabled in requestDevice.

# Glossary # {#glossary}

TODO: Remove terms unused in the rest of the specification.
Expand Down