-
Notifications
You must be signed in to change notification settings - Fork 218
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.
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
Consider to switch to SPIR-V Tools format of text representation (assembly) #32
Comments
For this activity we'd need to switch to using the SPIR-V Tools In-memory format for representing the SPIR-V objects, then we can make calls directly into libspirv_tools to perform the toText serialization. We can also then standardise on other parts of the SPIRV-Tools repo, such as SPIR-V parsing. I think it will entail reworking a lot of the classes in libSPIRV, Module, Function, BasicBlock etc to at least include the same implementation objects as the SPIRVTools repo, we could replace the existing classes with an adaptor of the SPIRVTools classes to allow the interface to SPIRVReader et al to be the same. This would at least allow work to continue on supporting Vulkan in parallel. I expect about a month of work. We'd also need to modify the tests to use the new textual format. |
OpenCL TSG call July 31:
|
…> 64bits (KhronosGroup#32) In the LLVM IR if input is an integer with a width more than 64 bits the function will take this input through 'byval' pointer. If we leave the translation as is it will violate the extension spec, which says, that the input must by of an integer type. Add an extra load to fix this. Signed-off-by: Dmitry Sidorov <[email protected]>
It looks beneficial to have unified format.
This should also fix KhronosGroup/SPIRV-LLVM#199
The text was updated successfully, but these errors were encountered: