You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Does it work on the backend architecture? Are there any resources in this regard, what is the purpose of this translator project, and does it work on the back-end architecture? Or something else?
The text was updated successfully, but these errors were encountered:
The first one that generates SPIR-V. Across compilers for heterogeneous languages it's used a FE to generate SPIR-V IR. There are several types of such FE, for example GLSL and Vulkan have their own SPIR-V generators. What to do with such SPIR-V IR - is up to the appropriate device's backend.
The second part of the translator generates back LLVM IR. It can be used (though not necessarily) by various SPIR-V consumers across different devices' backends, which are on their own are built on LLVM infrastructure. Yet again, you can already generate machine code right from SPIR-V, but AFAIK several vendors are transforming SPIR-V IR to LLVM IR to be used it later in their backend, for example Intel Graphics Compiler does that, see https://github.com/intel/intel-graphics-compiler .
Does it work on the backend architecture? Are there any resources in this regard, what is the purpose of this translator project, and does it work on the back-end architecture? Or something else?
The text was updated successfully, but these errors were encountered: