mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-07 12:21:27 +01:00
a1ada050f8
Remove redundant to_dtype like `load_bf16 + to_fp32 + to_bf16 + store_bf16` => `load_bf16 + store_bf16` Pull Request resolved: https://github.com/pytorch/pytorch/pull/97147 Approved by: https://github.com/jgong5, https://github.com/EikanWang, https://github.com/jansel
28 lines
687 B
Python
28 lines
687 B
Python
# counter for tracking how many kernels have been generated
|
|
generated_kernel_count = 0
|
|
generated_cpp_vec_kernel_count = 0
|
|
num_bytes_accessed = 0
|
|
nodes_num_elem = []
|
|
|
|
# counters for tracking fusions
|
|
ir_nodes_pre_fusion = 0
|
|
|
|
# counters for tracking to_dtype inserted
|
|
cpp_to_dtype_count = 0
|
|
|
|
|
|
# reset all counters
|
|
def reset():
|
|
global generated_kernel_count
|
|
global generated_cpp_vec_kernel_count
|
|
global num_bytes_accessed, nodes_num_elem
|
|
global ir_nodes_pre_fusion
|
|
global cpp_to_dtype_count
|
|
|
|
generated_kernel_count = 0
|
|
generated_cpp_vec_kernel_count = 0
|
|
num_bytes_accessed = 0
|
|
nodes_num_elem.clear()
|
|
ir_nodes_pre_fusion = 0
|
|
cpp_to_dtype_count = 0
|