OpenCL 3.1.1 Released To Address A Possible Performance Regression
OpenCL 3.1.1 has been released to address a potential performance regression identified in OpenCL 3.1. This point release reverts the behavior of clGetEventInfo to its OpenCL 3.0 semantics, which avoids unnecessary host synchronization. The update also includes reserved enum blocks for future Intel and Qualcomm extensions and minor fixes.
- ▪OpenCL 3.1.1 focuses on addressing a performance regression from OpenCL 3.1.
- ▪The release reverts clGetEventInfo behavior to match OpenCL 3.0 to avoid performance issues.
- ▪The update reserves enum blocks for upcoming Intel and Qualcomm extensions.
Opening excerpt (first ~120 words) tap to expand
OpenCL 3.1.1 Released To Address A Possible Performance Regression Written by Michael Larabel in Standards on 22 May 2026 at 12:29 PM EDT. Add A Comment Released earlier this month was the OpenCL 3.1 specification with a focus on enhancing AI and HPC workloads for this long-time Khronos specification. Out today is OpenCL 3.1.1 as a point release with an emphasis on addressing a possible performance regression of OpenCL 3.1. OpenCL 3.1.1 reverts the short-lived OpenCL 3.1 behavior of clGetEventInfo returning CL_COMPLETE as a host synchronization point. Those wanting a host synchronization point should instead call a function waiting on the OpenCL event instead like with clWaitForEvents.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Phoronix.