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.

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

Use cupy to measure memory leak #777

Merged
merged 3 commits into from
Sep 6, 2024
Merged

Use cupy to measure memory leak #777

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
1b83eb1
Use cuda-python to measure memory leak
bdice Sep 6, 2024
8075ba3
Clarify units are MiB.
bdice Sep 6, 2024
a509f2d
Use cupy instead of cuda-python.
bdice Sep 6, 2024
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
2 changes: 1 addition & 1 deletion conda/environments/all_cuda-118_arch-x86_64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,10 +6,10 @@ channels:
- conda-forge
- nvidia
dependencies:
- GPUtil>=1.4.0
- c-compiler
- click
- cmake>=3.26.4,!=3.30.0
- cuda-python>=11.7.1,<12.0a0
- cuda-version=11.8
- cudatoolkit
- cupy>=12.0.0
Expand Down
2 changes: 1 addition & 1 deletion conda/environments/all_cuda-125_arch-x86_64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,12 +6,12 @@ channels:
- conda-forge
- nvidia
dependencies:
- GPUtil>=1.4.0
- c-compiler
- click
- cmake>=3.26.4,!=3.30.0
- cuda-cudart-dev
- cuda-nvcc
- cuda-python>=12.0,<13.0a0
- cuda-version=12.5
- cupy>=12.0.0
- cxx-compiler
Expand Down
11 changes: 10 additions & 1 deletion dependencies.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -295,7 +295,6 @@ dependencies:
common:
- output_types: [conda, requirements, pyproject]
packages:
- GPUtil>=1.4.0
- psutil>=5.8.0
- pytest>=6.2.4,<8.0.0a0
- pytest-cov>=2.12.1
Expand Down Expand Up @@ -323,3 +322,13 @@ dependencies:
packages:
# Already added to requirements via docs. This is for tests.
- numpydoc>=1.5
specific:
- output_types: [conda, requirements, pyproject]
matrices:
- matrix: {cuda: "12.*"}
packages:
- cuda-python>=12.0,<13.0a0
- matrix: {cuda: "11.*"}
packages: &test_cuda_python_cu11
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

out of curiosity, what does this &test_cuda_python_cu11 notation do?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is a YAML anchor. We use this to define a list that we re-use in *test_cuda_python_cu11. We need a "fallback" list when the cuda selector is not defined.

- cuda-python>=11.7.1,<12.0a0
- {matrix: null, packages: *test_cuda_python_cu11}
2 changes: 1 addition & 1 deletion python/cucim/pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ Tracker = "https://github.com/rapidsai/cucim/issues"

[project.optional-dependencies]
test = [
"GPUtil>=1.4.0",
"cuda-python>=11.7.1,<12.0a0",
"imagecodecs>=2021.6.8; platform_machine=='x86_64'",
"matplotlib",
"numpydoc>=1.5",
Expand Down
23 changes: 13 additions & 10 deletions python/cucim/tests/performance/clara/test_read_region_memory_usage.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@
# limitations under the License.
#

import cuda.cudart
import pytest

from ...util.io import open_image_cucim
Expand All @@ -22,23 +23,25 @@


def test_read_region_cuda_memleak(testimg_tiff_stripe_4096x4096_256_jpeg):
import GPUtil

gpus = GPUtil.getGPUs()

if len(gpus) == 0:
def get_used_gpu_memory_mib():
"""Get the used GPU memory in MiB."""
status, free, total = cuda.cudart.cudaMemGetInfo()
if status != cuda.cudart.cudaError_t.cudaSuccess:
raise RuntimeError("Failed to get GPU memory info.")
memory_used = (total - free) / (2**20)
return memory_used

status, num_gpus = cuda.cudart.cudaGetDeviceCount()
if status != cuda.cudart.cudaError_t.cudaSuccess or num_gpus == 0:
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We can also potentially just use CuPy for this and avoid having to add the cuda-python dependency:

add up top:

import cupy as cp

then can use here:

    def get_used_gpu_memory_mib():
        """Get the used GPU memory in MiB."""
        dev = cp.cuda.Device()
        free, total = dev.mem_info
        memory_used = (total - free) / (2**20)
        return memory_used

    num_gpus = cp.cuda.runtime.getDeviceCount()
    if num_gpus == 0:

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Great. I'll implement that and remove the cuda-python dependency.

pytest.skip("No gpu available")

img = open_image_cucim(testimg_tiff_stripe_4096x4096_256_jpeg)

gpu = gpus[0]
mem_usage_history = [gpu.memoryUsed]
mem_usage_history = [get_used_gpu_memory_mib()]

for i in range(10):
_ = img.read_region(device="cuda")
gpus = GPUtil.getGPUs()
gpu = gpus[0]
mem_usage_history.append(gpu.memoryUsed)
mem_usage_history.append(get_used_gpu_memory_mib())

print(mem_usage_history)

Expand Down