Skip to content

Commit c9b57d9

Browse files
authored
Updates torch to 2.7.0 with cuda 12.8 blackwell support (isaac-sim#2998)
# Description As required by isaac-sim#2962 and vulnerabilities in torch 2.5.1, this change updates torch to 2.7.0 during the installation of Isaac Lab. Although inconsistent with Isaac Sim 4.5 (still on torch 2.5.1), Isaac Lab should work fine with 2.7.0+cu128. This update will also allow us to have better support for blackwell GPUs. ## Type of change <!-- As you go through the list, delete the ones that are not applicable. --> - Bug fix (non-breaking change which fixes an issue) - New feature (non-breaking change which adds functionality) - This change requires a documentation update ## Checklist - [x] I have run the [`pre-commit` checks](https://pre-commit.com/) with `./isaaclab.sh --format` - [x] I have made corresponding changes to the documentation - [x] My changes generate no new warnings - [ ] I have added tests that prove my fix is effective or that my feature works - [ ] I have updated the changelog and the corresponding version in the extension's `config/extension.toml` file - [ ] I have added my name to the `CONTRIBUTORS.md` or my name already exists there <!-- As you go through the checklist above, you can mark something as done by putting an x character in it For example, - [x] I have done this task - [ ] I have not done this task -->
1 parent a42793f commit c9b57d9

File tree

12 files changed

+89
-76
lines changed

12 files changed

+89
-76
lines changed

docs/source/setup/installation/binaries_installation.rst

Lines changed: 0 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -383,26 +383,6 @@ Installation
383383
384384
The valid options are ``rl_games``, ``rsl_rl``, ``sb3``, ``skrl``, ``robomimic``, ``none``.
385385

386-
.. attention::
387-
388-
For 50 series GPUs, please use the latest PyTorch nightly build instead of PyTorch 2.5.1, which comes with Isaac Sim:
389-
390-
.. tab-set::
391-
:sync-group: os
392-
393-
.. tab-item:: :icon:`fa-brands fa-linux` Linux
394-
:sync: linux
395-
396-
.. code:: bash
397-
398-
./isaaclab.sh -p -m pip install --upgrade --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/cu128
399-
400-
.. tab-item:: :icon:`fa-brands fa-windows` Windows
401-
:sync: windows
402-
403-
.. code:: batch
404-
405-
isaaclab.bat -p -m pip install --upgrade --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/cu128
406386

407387
Verifying the Isaac Lab installation
408388
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

docs/source/setup/installation/isaaclab_pip_installation.rst

Lines changed: 3 additions & 23 deletions
Original file line numberDiff line numberDiff line change
@@ -51,22 +51,11 @@ To learn about how to set up your own project on top of Isaac Lab, see :ref:`tem
5151
env_isaaclab\Scripts\activate
5252
5353
54-
- Next, install a CUDA-enabled PyTorch 2.5.1 build based on the CUDA version available on your system. This step is optional for Linux, but required for Windows to ensure a CUDA-compatible version of PyTorch is installed.
54+
- Next, install a CUDA-enabled PyTorch 2.7.0 build. This step is optional for Linux, but required for Windows to ensure a CUDA-compatible version of PyTorch is installed.
5555

56-
.. tab-set::
57-
58-
.. tab-item:: CUDA 11
59-
60-
.. code-block:: bash
61-
62-
pip install torch==2.5.1 torchvision==0.20.1 --index-url https://download.pytorch.org/whl/cu118
63-
64-
.. tab-item:: CUDA 12
65-
66-
.. code-block:: bash
67-
68-
pip install torch==2.5.1 torchvision==0.20.1 --index-url https://download.pytorch.org/whl/cu121
56+
.. code-block:: bash
6957
58+
pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
7059
7160
- Before installing Isaac Lab, ensure the latest pip version is installed. To update pip, run
7261

@@ -94,15 +83,6 @@ To learn about how to set up your own project on top of Isaac Lab, see :ref:`tem
9483
pip install isaaclab[isaacsim,all]==2.1.0 --extra-index-url https://pypi.nvidia.com
9584
9685
97-
.. attention::
98-
99-
For 50 series GPUs, please use the latest PyTorch nightly build instead of PyTorch 2.5.1, which comes with Isaac Sim:
100-
101-
.. code:: bash
102-
103-
pip install --upgrade --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/cu128
104-
105-
10686
Verifying the Isaac Sim installation
10787
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
10888

docs/source/setup/installation/pip_installation.rst

Lines changed: 3 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -76,21 +76,11 @@ If you encounter any issues, please report them to the
7676
env_isaaclab\Scripts\activate
7777
7878
79-
- Next, install a CUDA-enabled PyTorch 2.5.1 build based on the CUDA version available on your system. This step is optional for Linux, but required for Windows to ensure a CUDA-compatible version of PyTorch is installed.
79+
- Next, install a CUDA-enabled PyTorch 2.7.0 build. This step is optional for Linux, but required for Windows to ensure a CUDA-compatible version of PyTorch is installed.
8080

81-
.. tab-set::
82-
83-
.. tab-item:: CUDA 11
84-
85-
.. code-block:: bash
81+
.. code-block:: bash
8682
87-
pip install torch==2.5.1 torchvision==0.20.1 --index-url https://download.pytorch.org/whl/cu118
88-
89-
.. tab-item:: CUDA 12
90-
91-
.. code-block:: bash
92-
93-
pip install torch==2.5.1 torchvision==0.20.1 --index-url https://download.pytorch.org/whl/cu121
83+
pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
9484
9585
- Before installing Isaac Sim, ensure the latest pip version is installed. To update pip, run
9686

@@ -300,13 +290,6 @@ Installation
300290
301291
The valid options are ``rl_games``, ``rsl_rl``, ``sb3``, ``skrl``, ``robomimic``, ``none``.
302292

303-
.. attention::
304-
305-
For 50 series GPUs, please use the latest PyTorch nightly build instead of PyTorch 2.5.1, which comes with Isaac Sim:
306-
307-
.. code:: bash
308-
309-
pip install --upgrade --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/cu128
310293

311294
Verifying the Isaac Lab installation
312295
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

docs/source/setup/quickstart.rst

Lines changed: 3 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -43,22 +43,20 @@ To begin, we first define our virtual environment. We recommend using `miniconda
4343
# activate the virtual environment
4444
conda activate env_isaaclab
4545
46-
Next, we need to install the CUDA-enabled version of PyTorch 2.5.1. This step is optional for Linux, but required for Windows to ensure a CUDA-compatible version of PyTorch is installed. If in doubt on which
47-
version to use, use 11.8.
48-
46+
Next, we need to install the CUDA-enabled version of PyTorch 2.7.0. This step is optional for Linux, but required for Windows to ensure a CUDA-compatible version of PyTorch is installed.
4947
.. tab-set::
5048

5149
.. tab-item:: CUDA 11
5250

5351
.. code-block:: bash
5452
55-
pip install torch==2.5.1 torchvision==0.20.1 --index-url https://download.pytorch.org/whl/cu118
53+
pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
5654
5755
.. tab-item:: CUDA 12
5856

5957
.. code-block:: bash
6058
61-
pip install torch==2.5.1 torchvision==0.20.1 --index-url https://download.pytorch.org/whl/cu121
59+
pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
6260
6361
Before we can install Isaac Sim, we need to make sure pip is updated. To update pip, run
6462

docs/source/tutorials/03_envs/policy_inference_in_usd.rst

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -58,7 +58,8 @@ Note that not all learning libraries support exporting the policy to a jit or on
5858
For libraries that don't currently support this functionality, please refer to the corresponding ``play.py`` script for the library
5959
to learn about how to initialize the policy.
6060

61-
We can then load the warehouse asset and run inference on the H1 robot using the exported jit policy.
61+
We can then load the warehouse asset and run inference on the H1 robot using the exported jit policy
62+
(``policy.pt`` file in the ``exported/`` directory).
6263

6364
.. code-block:: bash
6465

isaaclab.bat

Lines changed: 42 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -270,6 +270,27 @@ if "%arg%"=="-i" (
270270
rem install the python packages in isaaclab/source directory
271271
echo [INFO] Installing extensions inside the Isaac Lab repository...
272272
call :extract_python_exe
273+
274+
rem check if pytorch is installed and its version
275+
rem install pytorch with cuda 12.8 for blackwell support
276+
call !python_exe! -m pip list | findstr /C:"torch" >nul
277+
if %errorlevel% equ 0 (
278+
for /f "tokens=2" %%i in ('!python_exe! -m pip show torch ^| findstr /C:"Version:"') do (
279+
set torch_version=%%i
280+
)
281+
if not "!torch_version!"=="2.7.0+cu128" (
282+
echo [INFO] Uninstalling PyTorch version !torch_version!...
283+
call !python_exe! -m pip uninstall -y torch torchvision torchaudio
284+
echo [INFO] Installing PyTorch 2.7.0 with CUDA 12.8 support...
285+
call !python_exe! -m pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
286+
) else (
287+
echo [INFO] PyTorch 2.7.0 is already installed.
288+
)
289+
) else (
290+
echo [INFO] Installing PyTorch 2.7.0 with CUDA 12.8 support...
291+
call !python_exe! -m pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
292+
)
293+
273294
for /d %%d in ("%ISAACLAB_PATH%\source\*") do (
274295
set ext_folder="%%d"
275296
call :install_isaaclab_extension
@@ -295,6 +316,27 @@ if "%arg%"=="-i" (
295316
rem install the python packages in source directory
296317
echo [INFO] Installing extensions inside the Isaac Lab repository...
297318
call :extract_python_exe
319+
320+
rem check if pytorch is installed and its version
321+
rem install pytorch with cuda 12.8 for blackwell support
322+
call !python_exe! -m pip list | findstr /C:"torch" >nul
323+
if %errorlevel% equ 0 (
324+
for /f "tokens=2" %%i in ('!python_exe! -m pip show torch ^| findstr /C:"Version:"') do (
325+
set torch_version=%%i
326+
)
327+
if not "!torch_version!"=="2.7.0+cu128" (
328+
echo [INFO] Uninstalling PyTorch version !torch_version!...
329+
call !python_exe! -m pip uninstall -y torch torchvision torchaudio
330+
echo [INFO] Installing PyTorch 2.7.0 with CUDA 12.8 support...
331+
call !python_exe! -m pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
332+
) else (
333+
echo [INFO] PyTorch 2.7.0 is already installed.
334+
)
335+
) else (
336+
echo [INFO] Installing PyTorch 2.7.0 with CUDA 12.8 support...
337+
call !python_exe! -m pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
338+
)
339+
298340
for /d %%d in ("%ISAACLAB_PATH%\source\*") do (
299341
set ext_folder="%%d"
300342
call :install_isaaclab_extension

isaaclab.sh

Lines changed: 17 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -276,6 +276,23 @@ while [[ $# -gt 0 ]]; do
276276
# install the python packages in IsaacLab/source directory
277277
echo "[INFO] Installing extensions inside the Isaac Lab repository..."
278278
python_exe=$(extract_python_exe)
279+
# check if pytorch is installed and its version
280+
# install pytorch with cuda 12.8 for blackwell support
281+
if ${python_exe} -m pip list 2>/dev/null | grep -q "torch"; then
282+
torch_version=$(${python_exe} -m pip show torch 2>/dev/null | grep "Version:" | awk '{print $2}')
283+
echo "[INFO] Found PyTorch version ${torch_version} installed."
284+
if [[ "${torch_version}" != "2.7.0+cu128" ]]; then
285+
echo "[INFO] Uninstalling PyTorch version ${torch_version}..."
286+
${python_exe} -m pip uninstall -y torch torchvision torchaudio
287+
echo "[INFO] Installing PyTorch 2.7.0 with CUDA 12.8 support..."
288+
${python_exe} -m pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
289+
else
290+
echo "[INFO] PyTorch 2.7.0 is already installed."
291+
fi
292+
else
293+
echo "[INFO] Installing PyTorch 2.7.0 with CUDA 12.8 support..."
294+
${python_exe} -m pip install torch==2.7.0 torchvision==0.22.0 --index-url https://download.pytorch.org/whl/cu128
295+
fi
279296
# recursively look into directories and install them
280297
# this does not check dependencies between extensions
281298
export -f extract_python_exe

source/isaaclab/config/extension.toml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
[package]
22

33
# Note: Semantic Versioning is used: https://semver.org/
4-
version = "0.40.23"
4+
version = "0.41.0"
55

66
# Description
77
title = "Isaac Lab framework for Robot Learning"

source/isaaclab/docs/CHANGELOG.rst

Lines changed: 12 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,18 @@
11
Changelog
22
---------
33

4+
0.41.0 (2025-07-21)
5+
~~~~~~~~~~~~~~~~~~~
6+
7+
Changed
8+
^^^^^^^
9+
10+
* Updates torch version to 2.7.0 and torchvision to 0.22.0.
11+
Some dependencies now require torch>=2.6, and given the vulnerabilities in Torch 2.5.1,
12+
we are updating the torch version to 2.7.0 to also include Blackwell support. Since Isaac Sim 4.5 has not updated the
13+
torch version, we are now overwriting the torch installation step in isaaclab.sh when running ``./isaaclab.sh -i``.
14+
15+
416
0.40.23 (2025-06-29)
517
~~~~~~~~~~~~~~~~~~~~
618

source/isaaclab/setup.py

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -20,7 +20,7 @@
2020
INSTALL_REQUIRES = [
2121
# generic
2222
"numpy<2",
23-
"torch==2.5.1",
23+
"torch>=2.5.1",
2424
"onnx==1.16.1", # 1.16.2 throws access violation on Windows
2525
"prettytable==3.3.0",
2626
"toml",
@@ -53,7 +53,7 @@
5353
"dex-retargeting==0.4.6", # required by isaaclab.devices.openxr.retargeters.humanoid.fourier.gr1_t2_dex_retargeting_utils
5454
]
5555

56-
PYTORCH_INDEX_URL = ["https://download.pytorch.org/whl/cu118"]
56+
PYTORCH_INDEX_URL = ["https://download.pytorch.org/whl/cu128"]
5757

5858
# Installation operation
5959
setup(

0 commit comments

Comments
 (0)