Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Flops profiler support einops.einsum #6755

Open
wants to merge 11 commits into
base: master
Choose a base branch
from
34 changes: 34 additions & 0 deletions deepspeed/profiling/flops_profiler/profiler.py
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@
from deepspeed.moe.layer import MoE
from deepspeed.utils.timer import FORWARD_GLOBAL_TIMER, BACKWARD_GLOBAL_TIMER, STEP_GLOBAL_TIMER
from deepspeed.utils.torch import required_torch_version
import einops
loadams marked this conversation as resolved.
Show resolved Hide resolved

Tensor = torch.Tensor

Expand Down Expand Up @@ -82,6 +83,7 @@ def start_profile(self, ignore_list=None):
self.reset_profile()
_patch_functionals()
_patch_tensor_methods()
_patch_miscellaneous_operations()

def register_module_hooks(module, ignore_list):
if ignore_list and type(module) in ignore_list:
Expand Down Expand Up @@ -137,6 +139,7 @@ def stop_profile(self):
if self.started and self.func_patched:
_reload_functionals()
_reload_tensor_methods()
_reload_miscellaneous_operations()
self.func_patched = False

def remove_profile_attrs(module):
Expand Down Expand Up @@ -787,6 +790,29 @@ def _einsum_flops_compute(equation, *operands):
raise NotImplementedError("Unsupported einsum operation.")


def _einops_einsum_flops_compute(*args):
"""
Count flops for the einops.einsum operation.
"""
*operands, equation = args
input_shapes = [o.shape for o in operands]

# Re-map equation so that same equation with different alphabet
# representations will look the same.
letter_order = OrderedDict((k, 0) for k in equation if k.isalpha()).keys()
mapping = {ord(x): 97 + i for i, x in enumerate(letter_order)}
equation = equation.translate(mapping)

np_arrs = [np.zeros(s) for s in input_shapes]
optim = np.einsum_path(equation, *np_arrs, optimize="optimal")[1]
for line in optim.split("\n"):
if "optimized flop" in line.lower():
flop = int(float(line.split(":")[-1]))
return flop, 0

raise NotImplementedError("Unsupported einops.einsum operation.")


def _tensor_addmm_flops_compute(self, mat1, mat2, *, beta=1, alpha=1, out=None):
"""
Count flops for the tensor addmm operation.
Expand Down Expand Up @@ -937,6 +963,10 @@ def _patch_tensor_methods():
torch.baddbmm = wrapFunc(torch.baddbmm, _tensor_addmm_flops_compute)


def _patch_miscellaneous_operations():
einops.einsum = wrapFunc(einops.einsum, _einops_einsum_flops_compute)


def _reload_functionals():
# torch.nn.functional does not support importlib.reload()
F.linear = old_functions[F.linear.__str__]
Expand Down Expand Up @@ -995,6 +1025,10 @@ def _reload_tensor_methods():
torch.baddbmm = old_functions[torch.baddbmm.__str__]


def _reload_miscellaneous_operations():
einops.einsum = old_functions[einops.einsum.__str__]


def _rnn_flops(flops, rnn_module, w_ih, w_hh, input_size):
gates_size = w_ih.shape[0]
# matrix matrix mult ih state and internal state
Expand Down
1 change: 1 addition & 0 deletions requirements/requirements-dev.txt
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@ clang-format==18.1.3
comet_ml>=3.41.0
deepspeed-kernels ; sys_platform == 'linux'
docutils<0.18
einops
future
importlib-metadata>=4
mup
Expand Down