diff --git a/README.md b/README.md index 2408eb9..9c0c242 100644 --- a/README.md +++ b/README.md @@ -25,6 +25,7 @@ To run DeepView.Profile, you need: The current PyTorch install supports CUDA capabilities sm_37 sm_50 sm_60 sm_70. If you want to use the NVIDIA GeForce RTX 3060 Ti GPU with PyTorch, please check the instructions at https://pytorch.org/get-started/locally/ ``` +- **For new RTX 4000 GPUs you need to install pytorch with cuda11.8 [pip install torch==2.0.1+cu118 torchvision==0.15.2+cu118 torchaudio==2.0.2 --index-url https://download.pytorch.org/whl/cu118] otherwise CUPTI will not initialize correctly** ### Installation from PyPi diff --git a/deepview_profile/profiler/utilization.py b/deepview_profile/profiler/utilization.py index 5aa9137..83a4624 100644 --- a/deepview_profile/profiler/utilization.py +++ b/deepview_profile/profiler/utilization.py @@ -158,7 +158,7 @@ def _populate_backward_data(self, node): for ch in node.children: self._populate_backward_data(ch) - @functools.cache + @functools.lru_cache(maxsize=None) def _can_match(self, f, b): if "aten" in f and "Backward0" in b: raw_f = f[len("aten::"):].lower().replace("_", "") diff --git a/pyproject.toml b/pyproject.toml index 8daa98d..e4e972d 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -28,7 +28,7 @@ python = "^3.7" pyyaml = "*" protobuf = "3.19.6" numpy = "^1.15.2" -torch = "*" +torch = "<=2.0.1" nvidia-ml-py3 = "*" toml = "^0.10.2" pyRAPL = "^0.2.3"