Exposes a SPIR-V loop control hint MultipleWaitQueuesQCOM that instructs the compiler to use multiple hardware wait queues to deepen the data fetch pipeline for memory-latency-bound workloads. This is particularly beneficial for cooperative matrix loops where register pressure and monolithic instruction granularity limit the effectiveness of traditional compiler optimizations such as loop unrolling and instruction scheduling.
The task list for the VK_QCOM_shader_multiple_wait_queues release is:
As each component is made public, the task will be checked off. When all tasks have been completed this issue will be closed and the extension will be fully released.
Exposes a SPIR-V loop control hint
MultipleWaitQueuesQCOMthat instructs the compiler to use multiple hardware wait queues to deepen the data fetch pipeline for memory-latency-bound workloads. This is particularly beneficial for cooperative matrix loops where register pressure and monolithic instruction granularity limit the effectiveness of traditional compiler optimizations such as loop unrolling and instruction scheduling.The task list for the VK_QCOM_shader_multiple_wait_queues release is:
As each component is made public, the task will be checked off. When all tasks have been completed this issue will be closed and the extension will be fully released.