Skip to content

Add support for apple silicon #113

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.

Sign up for GitHub

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

Open
zakki wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Add support for apple silicon #113

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
45d8bad
Add support for apple silicon
zakki Apr 2, 2025
c0b494d
CUDA以外のバックエンド用にtorch.cuda.ampの代わりにtorch.ampを使う
zakki Apr 8, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions nn/learn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -150,7 +150,7 @@ def train_on_gpu(program_dir: str, board_size: int, batch_size: int, \
weight_decay=WEIGHT_DECAY,
nesterov=True)

scaler = torch.cuda.amp.GradScaler()
scaler = torch.amp.GradScaler(device.type)

current_lr = SL_LEARNING_RATE

Expand All @@ -166,7 +166,7 @@ def train_on_gpu(program_dir: str, board_size: int, batch_size: int, \
dual_net.train()
epoch_time = time.time()
for i in range(0, len(value_data) - batch_size + 1, batch_size):
with torch.cuda.amp.autocast(enabled=True):
with torch.amp.autocast(device.type, enabled=True):
plane = torch.tensor(plane_data[i:i+batch_size]).to(device)
policy = torch.tensor(policy_data[i:i+batch_size]).to(device)
value = torch.tensor(value_data[i:i+batch_size]).to(device)
Expand Down Expand Up @@ -339,7 +339,7 @@ def train_with_gumbel_alphazero_on_gpu(program_dir: str, board_size: int, \
weight_decay=WEIGHT_DECAY,
nesterov=True)

scaler = torch.cuda.amp.GradScaler()
scaler = torch.amp.GradScaler(device.type)

num_trained_batches = 0

Expand Down Expand Up @@ -368,7 +368,7 @@ def train_with_gumbel_alphazero_on_gpu(program_dir: str, board_size: int, \
dual_net.train()
epoch_time = time.time()
for i in range(0, len(value_data) - batch_size + 1, batch_size):
with torch.cuda.amp.autocast(enabled=True):
with torch.amp.autocast(device.type, enabled=True):
plane = torch.tensor(plane_data[i:i+batch_size]).to(device)
policy = torch.tensor(policy_data[i:i+batch_size]).to(device)
value = torch.tensor(value_data[i:i+batch_size]).to(device)
Expand Down
7 changes: 5 additions & 2 deletions nn/utility.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,8 +19,11 @@ def get_torch_device(use_gpu: bool) -> torch.device:
torch.device: デバイス情報。
"""
if use_gpu:
torch.cuda.set_device(0)
return torch.device("cuda")
if torch.cuda.is_available():
torch.cuda.set_device(0)
return torch.device("cuda")
if torch.backends.mps.is_available():
return torch.device("mps")
return torch.device("cpu")


Expand Down