Open ceciliapeng2011 opened 10 months ago
.take
Thank you for looking into this issue! Please let us know if you have any questions or require any help.
.take
Thanks for being interested in this issue. It looks like this ticket is already assigned to a contributor. Please communicate with the assigned contributor to confirm the status of the issue.
Hi @siddhant-0707, are you still working on that issue or can I reopen it for other contributors?
PR #22085
hey @siddhant-0707 PR has been closed due lack of activity. Do you have time to pick up a topic and finish changes, or should we move it back to open pool?
Thanks for being interested in this issue. It looks like this ticket is already assigned to a contributor. Please communicate with the assigned contributor to confirm the status of the issue.
The PR has been reopened, I'm moving the ticket to in review.
Context
OpenVINO CPU Plugin internally uses a utility function cpu_convert. In the present moment this function uses AVX2 instruction to load/store fp16 data. Probably it is more efficient to use AVX512 fp16 instruction for platforms with AVX512 ISA.
https://github.com/openvinotoolkit/openvino/blob/master/src/plugins/intel_cpu/src/nodes/common/cpu_convert.cpp#L29-L142
What needs to be done?
Example Pull Requests
No response
Resources
HW Preparation
Before you make decision to take the contribution, please make sure you have the AVX512 CPU platform in hand for development and performance debugging and profiling. https://www.intel.com/content/www/us/en/architecture-and-technology/avx-512-overview.html
Contact points
@ceciliapeng2011 @yuxu42 @luo-cheng2021
Ticket
No response