What is your question?
The pytorch profiler can aggregate runtimes by the operation and its traceback via key_averages(group_by_stack_n=n): https://docs.pytorch.org/tutorials/beginner/profiler.html#print-profiler-result
Identifying the top operators / kernels and their source user Python code is helpful to find the main bottlenecks and perform targeted optimizations. Does HTA support something equivalent, or could you shed some light on implementing one? Thanks!
I also found get_aten_op_kernels_and_delay(), but seems like it traces down from aten ops only instead of source user training code.
Code
No response
What have you tried?
No response
Environment
No response
What is your question?
The pytorch profiler can aggregate runtimes by the operation and its traceback via
key_averages(group_by_stack_n=n): https://docs.pytorch.org/tutorials/beginner/profiler.html#print-profiler-resultIdentifying the top operators / kernels and their source user Python code is helpful to find the main bottlenecks and perform targeted optimizations. Does HTA support something equivalent, or could you shed some light on implementing one? Thanks!
I also found get_aten_op_kernels_and_delay(), but seems like it traces down from aten ops only instead of source user training code.
Code
No response
What have you tried?
No response
Environment
No response