Omar Ahmed

@kayhayen there is something that i don't understand here will we use a preallocated big space in the heap that we will distribute them on our lists in a smart...

@kayhayen i mean that we want to avoid malloc and we want to assign a space for our lists so will we use malloc with a big size one time...

This is a suggested alternative way: * Add an optional flag to clspv that says it's safe to add the Linkage capability to the SPIR-V and then add function declarations...

I had an idea to standardize some of the transitioning work for the passes: 1. Modify InferType so that when it recurse down through function definitions to infer the type...

@intel/dpcpp-nativecpu-pi-reviewers @intel/llvm-reviewers-cuda @intel/unified-runtime-reviewers @intel/llvm-reviewers-runtime @sergey-semenov Could you review this when possible, Thanks!

@oneapi-src/unified-runtime-cuda-write Could I have a review on this when available. Thanks!

@RossBrunton Please rebase this when possible, I removed the ready to merge label and please readd it when rebased.

@winstonzhang-intel Please make sure that https://github.com/intel/llvm/pull/14536 is updated and passing CI, so we could merge this. Thanks!

I removed the ready to merge label temporarily until the intel/llvm side is updated and passing tests. Please readd it when intel/llvm PR is updated.

Reading a float number from an image with non-normalized coord, `UR_SAMPLER_ADDRESSING_MODE_REPEAT/MIRRORED_REPEAT` and `UR_SAMPLER_FILTER_MODE_NEAREST` sampler, give an error when executing the kernel on opencl cpu platform. So, avoided reading with a...