Implement new functions from gpurt

[jiaolu]

Add AmdTraceRayGetKnownSetRayFlags/AmdTraceRayGetKnownUnsetRayFlags implementations

[amdvlk-admin]

Test summary for commit 2b38a27

CTS tests (Failed: 0/138443)

• Built with version 1.3.5.2
Ubuntu navi3x, Srdcvk
• Passed: 35211/69228 (50.9%)
• Failed: 0/69228 (0.0%)
• Not Supported: 34017/69228 (49.1%)
• Warnings: 0/69228 (0.0%)
Ubuntu navi2x, Srdcvk
• Passed: 35242/69215 (50.9%)
• Failed: 0/69215 (0.0%)
• Not Supported: 33973/69215 (49.1%)
• Warnings: 0/69215 (0.0%)

[jasilvanus]

Are we not analyzing constant TraceRay flags in LLPC, and optimize Traversal for it?

[jiaolu]

Are we not analyzing constant TraceRay flags in LLPC, and optimize Traversal for it?

this PR is originated from internal project souce.

[LLJJDD]

Are we not analyzing constant TraceRay flags in LLPC, and optimize Traversal for it?

We aren't so far, I suppose this would rely on Nicolai's internal proposal about library.

[jasilvanus]

Are we not analyzing constant TraceRay flags in LLPC, and optimize Traversal for it?

We aren't so far,

Interesting.

I suppose this would rely on Nicolai's internal proposal about library.

Only when compiling libraries, correct? For full pipelines, we should have everything we need available.

[LLJJDD]

I suppose this would rely on Nicolai's internal proposal about library.

Only when compiling libraries, correct? For full pipelines, we should have everything we need available.

Oh yes, it should be doable for full pipelines for now.

[jiaolu]

For full pipelines, we should have everything we need available.

A full pipeline can be very large with thousands of shaders. Probably not good idea to iterate every shader for constant flags analysis.

[jiaolu]

retest this please

[amdvlk-admin]

Test summary for commit 62b90cf

CTS tests (Failed: 0/138443)

• Built with version 1.3.5.2
Ubuntu navi3x, Srdcvk
• Passed: 35211/69228 (50.9%)
• Failed: 0/69228 (0.0%)
• Not Supported: 34017/69228 (49.1%)
• Warnings: 0/69228 (0.0%)
Ubuntu navi2x, Srdcvk
• Passed: 35242/69215 (50.9%)
• Failed: 0/69215 (0.0%)
• Not Supported: 33973/69215 (49.1%)
• Warnings: 0/69215 (0.0%)

[nhaehnle]

Yes, we certainly should apply such optimizations in the future, but the point of this PR is primarily to allow us to have the GPURT change move forward. LGTM.