Nathan Gauër
Nathan Gauër
spirv-tools fix in, now I need to bring it to DXC
> This should be an `OpConvertUToF` and not a bitcast. The bitcast of 49,600 results in subnormal, which can be flushed to 0. Yep, seems like this is the reason....
Hello! I looked at the SPIR-V file in this PR (`relaxed_demo.comp.spv`) and it seams it has no `OpExtInstWithForwardRef` instructions. I don't know if GLSL emits those, but you can have...
Short repro for this issue: ```hlsl // RUN: %dxc %s -T cs_6_8 -spirv -fspv-target-env=vulkan1.3 -E main struct B { float c; }; struct A { B b; }; float EvaluateLighting(A...
> I need more context on the purpose of this PR. Does it fix an issue? The main goal of this PR is to reduce the amount of special cases...
> `vk::biding` is not inline spir-v. That's a fair split. I'm fine removing the `vk::binding` and `vk::location` normalization from the PR as we could argue those are reserved for "native"...
This is a trim_capability issue: For Float16, the pass looks if any OpTypeFloat with 16 as size exists, and for storage if such type is used. But it doesn't do...
The `globallycoherent` bit seems possible, but I'll need to figure out how to correctly pass the attribute to the implicit cast, which for now only takes the type of the...
Took a look at this today, seems like we have 2 things to solve: 1. Support reinterpret/static pointer casts in SemaHLSL This part is straightforward, add code to check the...
Thanks for the issue! Sent a PR 😊 Most of the team being OOO, there might be some delays, but it's on its way.