Skip to content

Bump CANN version to 9.0.0 #64

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

Merged
zouzias merged 17 commits into from
May 12, 2026
Merged

Bump CANN version to 9.0.0 #64

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
17 commits
Select commit Hold shift + click to select a range
3a1e27a
(ci) bump CANN version
zouzias Mar 21, 2026
83ff20a
(cmake) bump pto-isa tag ver
zouzias Mar 21, 2026
d4dda45
Update Docker image tag in build workflow
zouzias Mar 21, 2026
f48e8d1
Update build.yml
zouzias Apr 14, 2026
4307a97
Update Docker image version in build workflow
zouzias Apr 14, 2026
c8075ea
Merge branch 'main' into bump_cann_9_0_0
Apr 17, 2026
fb4a17b
Update Docker image tag in build workflow
zouzias May 7, 2026
f1a5f50
Update GIT_TAG for libpto_isa_headers to v9.0.0
zouzias May 7, 2026
e9f7f5c
Update build.yml to remove Python 3.12 entries
zouzias May 7, 2026
e983104
Remove duplicate architecture entry in build.yml
zouzias May 7, 2026
426cf85
Update npu_device from 'a3' to '950' in build.yml
zouzias May 10, 2026
21526ec
Merge branch 'main' into bump_cann_9_0_0
May 10, 2026
f6a9bdf
fix TGATHER
May 10, 2026
13e3fef
fix aicpu examples
May 10, 2026
bb1bdee
update internal ci
May 12, 2026
b041d64
fix dup envs
May 12, 2026
9eadcc6
bump torch-npu version
May 12, 2026
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
10 changes: 2 additions & 8 deletions .github/workflows/build-examples-aicpu.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,18 +21,12 @@ jobs:
- arch: x86_64
python_version: 3.11
npu_device: 910b
- arch: x86_64
python_version: 3.12
npu_device: 910b
- arch: x86_64
python_version: 3.11
npu_device: a3
- arch: x86_64
python_version: 3.12
npu_device: a3
npu_device: 950

container:
image: quay.io/ascend/cann:8.5.1-${{ matrix.npu_device }}-ubuntu22.04-py${{ matrix.python_version }}
image: quay.io/ascend/cann:9.0.0-${{ matrix.npu_device }}-ubuntu22.04-py${{ matrix.python_version }}
steps:
- name: Show OS
run: cat /etc/os-release
Expand Down
24 changes: 6 additions & 18 deletions .github/workflows/build.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,18 +67,12 @@ jobs:
- arch: x86_64
python_version: 3.11
npu_device: 910b
- arch: x86_64
python_version: 3.12
npu_device: 910b
- arch: x86_64
python_version: 3.11
npu_device: a3
- arch: x86_64
python_version: 3.12
npu_device: a3
npu_device: 950

container:
image: quay.io/ascend/cann:8.5.1-${{ matrix.npu_device }}-ubuntu22.04-py${{ matrix.python_version }}
image: quay.io/ascend/cann:9.0.0-${{ matrix.npu_device }}-ubuntu22.04-py${{ matrix.python_version }}
steps:
- name: Show OS
run: cat /etc/os-release
Expand All @@ -96,7 +90,7 @@ jobs:
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib # to avoid libascend_hal.so issue
pip3 install pyyaml setuptools pytest packaging pybind11[global]
pip3 install -r requirements.txt
pip3 install torch-npu==2.8.0.post2 --extra-index-url https://download.pytorch.org/whl/cpu
pip3 install torch-npu==2.8.0.post4 --extra-index-url https://download.pytorch.org/whl/cpu
make clean build_cmake

build_wheel_docker:
Expand All @@ -110,17 +104,11 @@ jobs:
- arch: x86_64
python_version: 3.11
npu_device: 910b
- arch: x86_64
python_version: 3.12
npu_device: 910b
- arch: x86_64
python_version: 3.11
npu_device: a3
- arch: x86_64
python_version: 3.12
npu_device: a3
npu_device: 950
container:
image: quay.io/ascend/cann:8.5.1-${{ matrix.npu_device }}-ubuntu22.04-py${{ matrix.python_version }}
image: quay.io/ascend/cann:9.0.0-${{ matrix.npu_device }}-ubuntu22.04-py${{ matrix.python_version }}
steps:
- name: Show OS
run: cat /etc/os-release
Expand All @@ -138,5 +126,5 @@ jobs:
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib # to avoid libascend_hal.so issue
pip3 install pyyaml setuptools pytest packaging pybind11[global]
pip3 install -r requirements.txt
pip3 install torch-npu==2.8.0.post2 --extra-index-url https://download.pytorch.org/whl/cpu
pip3 install torch-npu==2.8.0.post4 --extra-index-url https://download.pytorch.org/whl/cpu
make clean build_wheel
30 changes: 8 additions & 22 deletions .gitlab-ci.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,6 @@ stages:
- build_in_docker
- test
- test_aarch64_910B2_py3.11
- test_aarch64_910B2_py3.12 # Serialize the py3.11 and py3.12 tests otherwise CI fails!


# Define the workflow rules manually to make sure only a single pipeline
Expand All @@ -23,11 +22,11 @@ workflow:


.build_config: &build_config
image: quay.io/ascend/cann:8.5.1-910b-ubuntu22.04-${PYTHON_VERSION}
image: quay.io/ascend/cann:9.0.0-${NPU}-ubuntu22.04-py3.11
parallel:
matrix:
- {PYTHON_VERSION: py3.11}
- {PYTHON_VERSION: py3.12}
- {NPU: 910b}
- {NPU: 950}
before_script:
- source /usr/local/Ascend/ascend-toolkit/set_env.sh
- pip3 install pyyaml setuptools pytest packaging
Expand All @@ -42,20 +41,9 @@ build_docker:
- make build_wheel

build_and_test_x86_910B4:
image: quay.io/ascend/cann:9.0.0-910b-ubuntu22.04-py3.11
tags:
- docker_npu # Build and test on-device NPU (x86-64 / 910B4)
<<: *build_config
script:
- make setup_once
- make build_wheel
- pip install pto_kernels*.whl
- python3 -m pytest -v tests/

build_and_test_aarch64_910B2_py3.11:
image: quay.io/ascend/cann:8.5.1-910b-ubuntu22.04-py3.11
stage: test_aarch64_910B2_py3.11
tags:
- docker_npu_910B2 # Build and test on-device (aarch64 / 910B2)
before_script:
- source /usr/local/Ascend/ascend-toolkit/set_env.sh
- pip3 install pyyaml setuptools pytest packaging
Expand All @@ -65,12 +53,10 @@ build_and_test_aarch64_910B2_py3.11:
- make build_wheel
- pip install pto_kernels*.whl
- python3 -m pytest -v tests/
when: manual
allow_failure: true

build_and_test_aarch64_910B2_py3.12:
image: quay.io/ascend/cann:8.5.1-910b-ubuntu22.04-py3.12
stage: test_aarch64_910B2_py3.12
build_and_test_aarch64_910B2_py3.11:
image: quay.io/ascend/cann:9.0.0-910b-ubuntu22.04-py3.11
stage: test_aarch64_910B2_py3.11
tags:
- docker_npu_910B2 # Build and test on-device (aarch64 / 910B2)
before_script:
Expand All @@ -89,7 +75,7 @@ build_and_test_aarch64_910B2_py3.12:
# On-device test of examples/jit_cpp #
######################################
test_examples_on_x86_910B4:
image: quay.io/ascend/cann:8.5.1-910b-ubuntu22.04-py3.11
image: quay.io/ascend/cann:9.0.0-910b-ubuntu22.04-py3.11
tags:
- docker_npu # Build and test on-device NPU (x86-64 / 910B4)
parallel:
Expand Down
2 changes: 1 addition & 1 deletion CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,7 +96,7 @@ include(FetchContent)
FetchContent_Declare(
libpto_isa_headers
GIT_REPOSITORY https://gitcode.com/cann/pto-isa.git
GIT_TAG "313817be696792a4e16a7ea5994ec98e34391613")
GIT_TAG v9.0.0)

FetchContent_Populate(libpto_isa_headers)

Expand Down
2 changes: 1 addition & 1 deletion Makefile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@ clean:

setup_once:
pip3 install -r requirements.txt
pip3 install torch-npu==2.8.0.post2 --extra-index-url https://download.pytorch.org/whl/cpu
pip3 install torch-npu==2.8.0.post4 --extra-index-url https://download.pytorch.org/whl/cpu

build_cmake: clean
bash scripts/build.sh
Expand Down
5 changes: 4 additions & 1 deletion csrc/kernel/kernel_csr_gather.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -116,10 +116,13 @@ AICORE void runTCsrGather(__gm__ T* values, __gm__ int32_t* indices,
TileDataVal wTiles(remaining_elements);
TileDataVal zTiles(remaining_elements);
TileDataIdx idxTiles(remaining_elements);
TileDataIdx tmpTiles(remaining_elements);

// Assign the UB address for each tile
TASSIGN(valTiles, V_T_ADDR + stage * TILE_SIZE_IN_BYTES);
TASSIGN(wTiles, W_T_ADDR + stage * 2 * TILE_SIZE_IDX_IN_BYTES);
TASSIGN(tmpTiles, W_T_ADDR + stage * 2 * TILE_SIZE_IDX_IN_BYTES +
TILE_SIZE_IDX_IN_BYTES);
TASSIGN(zTiles, Z_T_ADDR + stage * TILE_SIZE_IN_BYTES);
TASSIGN(idxTiles, IDX_T_ADDR + stage * TILE_SIZE_IDX_IN_BYTES);

Expand Down Expand Up @@ -147,7 +150,7 @@ AICORE void runTCsrGather(__gm__ T* values, __gm__ int32_t* indices,
pipe_barrier(PIPE_V);

// Gather
TGATHER(wTiles, xTiles, idxTiles);
TGATHER(wTiles, xTiles, idxTiles, tmpTiles);
Comment thread
zouzias marked this conversation as resolved.

// Signal end of gather to MTE2 (next load)
set_flag(PIPE_V, PIPE_MTE2, ev1);
Expand Down
4 changes: 2 additions & 2 deletions pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ requires = [
"psutil==6.1.0",
"PyYAML==6.0.2",
"attrs==24.2.0",
"torch-npu==2.8.0.post2",
"torch-npu==2.8.0.post4",
]
build-backend = "scikit_build_core.build"

Expand Down Expand Up @@ -51,7 +51,7 @@ dependencies = [

[project.optional-dependencies]
torch = [
"torch-npu==2.8.0.post2",
"torch-npu==2.8.0.post4",
]

[tool.scikit-build]
Expand Down
2 changes: 1 addition & 1 deletion requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ setuptools==80.9.0
packaging
wheel
scikit-build-core==0.11.6
pip==26.0.1
pip==26.1.1
pytest==8.3.4
numpy==1.26.4
scipy==1.13.1
Expand Down
Loading