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.

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

Adjust fragment state blend factor and write mask validation #3217

Merged
merged 2 commits into from
Dec 8, 2023
Merged

Adjust fragment state blend factor and write mask validation #3217

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d817016
Adjust fragment state blend factor and write mask validation
shrekshao Dec 7, 2023
299ae3a
fix
shrekshao Dec 8, 2023
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
15 changes: 13 additions & 2 deletions src/webgpu/api/validation/render_pipeline/fragment_state.spec.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -366,7 +366,7 @@ g.test('pipeline_output_targets')
g.test('pipeline_output_targets,blend')
.desc(
`On top of requirements from pipeline_output_targets, when blending is enabled and alpha channel is read indicated by any blend factor, an extra requirement is added:
- fragment output must be vec4.
- fragment output must have an alpha channel (i.e. it must be vec4), otherwise writeMask must be 0.
`
)
.params(u =>
Expand All @@ -382,6 +382,15 @@ g.test('pipeline_output_targets,blend')
...u.combine('alphaSrcFactor', kBlendFactors),
...u.combine('alphaDstFactor', kBlendFactors),
] as const)
.expand('writeMask', function* (p) {
yield 0;
for (let i = 1; i <= p.componentCount; i++) {
yield (1 << i) - 1;
shrekshao marked this conversation as resolved.
Show resolved Hide resolved
}
if (p.componentCount < 4) {
yield 0xf;
}
})
)
.beforeAllSubcases(t => {
const { format } = t.params;
Expand All @@ -398,6 +407,7 @@ g.test('pipeline_output_targets,blend')
colorDstFactor,
alphaSrcFactor,
alphaDstFactor,
writeMask,
} = t.params;
const info = kTextureFormatInfo[format];

Expand All @@ -409,6 +419,7 @@ g.test('pipeline_output_targets,blend')
color: { srcFactor: colorSrcFactor, dstFactor: colorDstFactor },
alpha: { srcFactor: alphaSrcFactor, dstFactor: alphaDstFactor },
},
writeMask,
},
],
fragmentShaderCode: getFragmentShaderCodeWithOutput([
Expand All @@ -422,6 +433,6 @@ g.test('pipeline_output_targets,blend')
const _success =
info.color.type === sampleType &&
componentCount >= kTexelRepresentationInfo[format].componentOrder.length &&
meetsExtraBlendingRequirement;
(meetsExtraBlendingRequirement || writeMask === 0);
Copy link
Collaborator

@kainino0x kainino0x Dec 8, 2023 •
edited
Loading

Choose a reason for hiding this comment

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

Sorry, I just reread the spec and I think I misinterpreted this the other day.
IIUC actually the only case where writeMask=0 makes it valid is when the fragment shader has no output to that location. If it has output to that location, then the writeMask isn't checked by the validation in the spec.
https://gpuweb.github.io/gpuweb/#abstract-opdef-validating-gpufragmentstate

I do believe we could loosen the validation in the way that you've done here, but it would require a spec change.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

I think you are right. The "Otherwise..." is at the same level as

"If get the entry point(FRAGMENT, descriptor) has a shader stage output value output with location attribute equal to index:"

Copy link
Collaborator

Choose a reason for hiding this comment

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

I do believe we could loosen the validation in the way that you've done here, but it would require a spec change.

For reference the spec issue was gpuweb/gpuweb#2013

t.doCreateRenderPipelineTest(isAsync, _success, descriptor);
});