unified-runtime

unified-runtime copied to clipboard

Add ability to log from the tracing layer directly rather than use XPTI

1

This adds the ability to use the tracing layer with a logger rather than relying on XPTI. The motivation for this is to provide an equivalent to `SYCL_PI_TRACE` when we...

callumfare

fix leaks on level-zero interop program and kernel handles

5

pbalcer

Add two tests that check that the sampler filter and addressing modes are correct and implemented correctly on the adapter.

RossBrunton

This is a follow-up of https://github.com/oneapi-src/unified-runtime/pull/1412 which added the `isPowerOf2` condition to the OpenCL fill function. This is correct since `clEnqueueMemFillINTEL_fn` only accepts such patterns. What was not correct was...

keyradical

[HIP][CUDA] Refactor using profiling events

1

HIP changes: - To match with the current behaviour of CUDA adapter, `EvBase` in HIP was moved to `device` and `getElapsedTime` function now handles the profiling events' synchronization. Also, we...

keyradical

The HIP adapter doesn't support UR spec constants, just like the CUDA adapter.

npmiller

Add more output validation for urEventGetProfilingInfo test

2

Validate the output timings for `urEventGetProfilingInfo` entry-point. [intel/llvm testing](https://github.com/intel/llvm/pull/13778)

omarahmed1111

kbenzie

Handle the `UR_DEVICE_INFO_KERNEL_SET_SPECIALIZATION_CONSTANTS` device query to report the feature is not supported. This also causes the `urKernelSetSpecializationConstants` tests to be skipped. https://github.com/intel/llvm/pull/13809

kbenzie

Add initial API spec for experimental native enqueue.

hdelan