OpenCL-Docs

OpenCL-Docs copied to clipboard

filter CL_VERSION_X_Y macros from reference page see also

3

fixes #981 Instead of trying to make the CL_VERSION_X_Y macros link to the right page, just filter them from the "see also" list, because they aren't adding much value. I...

bashbaug

The specification does not seem to make any statement about the dynamic uniformity requirements of images and samplers. See https://gitlab.khronos.org/spirv/SPIR-V/-/merge_requests/286 for relevant SPIR-V discussion.

kpet

[RFC] cl_exp_tensor

5

This draft introduces an extension for tensor data type (`cl_tensor`) for storing N-dimensional data in implementation-defined memory layout. This extension is designed for the [defined built-in kernel extension](https://github.com/KhronosGroup/OpenCL-Docs/pull/1007) for starters....

linehill

Math optimization defaults

1

The API registry says at [5.8.6.3. Optimization Options](https://registry.khronos.org/OpenCL/specs/3.0-unified/html/OpenCL_API.html#optimization-options) that > These options are not turned on by default since it can result in incorrect output... Specifically referring to fmad optimization...

RainerMtb

fixes #890 Adds an error condition (`CL_INVALID_VALUE`) when attempting to unmap a SVM region that was not mapped.

bashbaug

The OpenCL C spec currenlty has a single section "6.15.11. Async Copies from Global to Local Memory, Local to Global Memory, and Prefetch". It makes sense to put the async...

bashbaug

The SPIR-V environment spec currently says: https://registry.khronos.org/OpenCL/specs/3.0-unified/html/OpenCL_Env.html#_rounding_modes > For half precision operations, devices may support either Round to nearest even or Round toward zero as the default rounding mode for...

bashbaug

edge case behavior for fract

1

From a discussion in the SPIR call on November 15th (see also internal SPIR-V issue 768): OpenCL defines the following edge case behavior for the `fract` built-in function - see...

bashbaug

specification doesn't clarify -cl-denorms-are-zero for half precision floating points

2

`-cl-denorms-are-zero` can be used to flush denorms for single and double precision floating point operations, but neither the OpenCL API spec nor the OpenCL Extension spec for `cl_khr_fp16` state anything...

karolherbst

Extension Dependencies on Specific OpenCL Versions

5

The current OpenCL extension spec requires specific OpenCL versions for several extensions: 1. The base-level `cl_khr_semaphore` extension and related external semaphore extensions requires OpenCL 1.2. 2. The `cl_khr_command_buffer` extension requires...

bashbaug