Rebuild for & Support NumPy 2 #16300

jakirkham · 2024-07-17T22:41:26Z

Description

Part of issue: rapidsai/build-planning#38

Start building cudf with numpy version 2.0. This remains compatible with numpy version 1.x and 2.x. Allows us to test building with numpy version 2.0 (and make sure we catch any issues that show up). Also relaxes the numpy 1.x pin. Pulls in the RDFG changes that are rolling out for broader RAPIDS NumPy 2 support.

Checklist

I am familiar with the Contributing Guidelines.
New or existing tests cover these changes.
The documentation is up to date with these changes.

jakirkham · 2024-07-17T23:21:35Z

Need NumPy 2 compatible RMM wheels to proceed. Marking as draft

seberg · 2024-08-13T06:06:50Z

conda/recipes/cudf/meta.yaml

@@ -64,8 +64,7 @@ requirements:
    - rapids-build-backend >=0.3.0,<0.4.0.dev0
    - scikit-build-core >=0.10.0
    - dlpack >=0.8,<1.0
-    # TODO: Change to `2.0` for NumPy 2
-    - numpy 1.23
+    - numpy 2.0


Thanks, good to get this done. But I suspect we can just remove NumPy as a host requirement for cudf? (at least as a direct requirement)

Unfortunately as Arrow is currently needed during the build (and Arrow uses NumPy headers in some cases), there is a NumPy build dependency. Please see this comment: #15165 (comment)

Once Arrow is dropped as a build dependency, we can remove NumPy from build dependencies too: #15193 (comment)

Ah right, sorry I forgot. For that it should be OK to build with 1.x, but I agree it is better to just change it!

Ah sorry I should have mentioned the second issue, which is we need to relax the NumPy pinning in RMM

For example we see this error on CI:

ERROR: Cannot install numpy==2.0.*, rmm-cu11==24.10.0a21 and rmm-cu11==24.10.0a22 because these package versions have conflicting dependencies. The conflict is caused by: The user requested numpy==2.0.* rmm-cu11 24.10.0a22 depends on numpy<2.0a0 and >=1.23 The user requested numpy==2.0.* rmm-cu11 24.10.0a21 depends on numpy<2.0a0 and >=1.23

If we do that, we can build with NumPy 2 here (even if we are still using NumPy 1 at runtime)

Edit: Should add we can do this whenever we like

Thanks to Sebastian we are now building RMM to allow NumPy 2: rapidsai/rmm#1650

Will update this PR once there are RMM packages available

Once Arrow is dropped as a build dependency, we can remove NumPy from build dependencies too: #15193 (comment)

Just noting that #16640 is up for review and all its CI is passing, I suspect it'll be merged soon.

This PR and that one will have some conflicts... which is good, because I think one should be tested again after the other is merged.

This changes the builds to start using NumPy 2 when building cuDF and friends. However this maintains the same runtime dependency to restrict cuDF and friends to NumPy 1.23+ (pre-2). As building libraries against NumPy 2 provides NumPy 2 compatibility and maintains NumPy 1 compatibility (back to NumPy 1.19), this provides equivalent NumPy compatibility to what we had before and extends it further. This also helps us ensure that our builds are (and remain) compatible with NumPy 2. Finally this will help us prep recipes for the NumPy 2 migration with RDFG still to come.

jakirkham · 2024-08-21T20:52:36Z

It looks like cudf has build requirements of pylibcudf and numpy

cudf/conda/recipes/cudf/meta.yaml

Lines 67 to 71 in 6a2f323

    
           # TODO: Change to `2.0` for NumPy 2 
        
           - numpy 1.23 
        
           - pyarrow ==16.1.0.* 
        
           - libcudf ={{ version }} 
        
           - pylibcudf ={{ version }}

pylibcudf in turn builds with numpy and constrains it

cudf/conda/recipes/pylibcudf/meta.yaml

Lines 67 to 68 in 6a2f323

    
           # TODO: Change to `2.0` for NumPy 2 
        
           - numpy 1.23

cudf/conda/recipes/pylibcudf/meta.yaml

Lines 84 to 85 in 6a2f323

    
           # TODO: Update `numpy` in `host` when dropping `<2.0a0` 
        
           - numpy >=1.23,<2.0a0

So in order to rebuild with NumPy 2, we also need to soften the NumPy constraints at runtime

Given this we can either...

Consolidate this and PR ( Remove NumPy <2 pin #16594 ) into a single PR
Looks at trimming the NumPy requirement: Drop NumPy build dep from cudf (keep in pylibcudf) #16631
Wait for the Arrow + NumPy cleanup work: [FEA] Reduce arrow library dependencies in cudf #15193

Personally would prefer to decouple this from the Arrow upgrade. So would lean towards 1 or 2

jakirkham · 2024-08-21T21:17:54Z

Went ahead and cherry-picked/squashed the changes from PR: #16594

That should fix the CI issues previously seen here

vyasr · 2024-08-21T21:51:02Z

Personally would prefer to decouple this from the Arrow upgrade. So would lean towards 1 or 2

Depends on your timeline. After #16590 is merged, my next PR will remove arrow as a build dependency, so it should be done by EOW 🤞

jakirkham · 2024-08-21T22:48:06Z

Am hoping we have this done before EOW. Sebastian is at EuroSciPy next week

jakirkham · 2024-08-23T01:09:26Z

Removing ci-codeowners as this is not touching relevant files

jakirkham · 2024-08-23T01:12:56Z

There are some tests here that use Dask-CUDA. However Dask-CUDA also was pinned to NumPy 1. So it wasn't possible to test with NumPy 2 and Dask-CUDA in the same environment

Now that we have relaxed Dask-CUDA constraints to allow NumPy 2 in PR ( rapidsai/dask-cuda#1375 ) and those packages are now available, will merge in upstream changes to this PR to retest with Dask-CUDA and NumPy 2

jakirkham · 2024-08-23T04:25:33Z

Ok can now see the tests install NumPy 2 and Dask-CUDA in the same environment

This and all other tests are passing! 🥳

jameslamb

The changes here make sense to me. Glad we were able to relax this constraint.

jakirkham · 2024-08-23T18:38:31Z

/merge

jakirkham · 2024-08-23T18:51:29Z

/merge

jakirkham · 2024-08-23T22:35:12Z

Hmm...when resolving conflicts all jobs passed except one CI job

That one CI job for some reason is downgrading Pandas from 2.2.2 (which is NumPy 2 compatible) to 2.0.0 (which is not):

Downloading pandas-2.0.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (12.3 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 12.3/12.3 MB 261.9 MB/s eta 0:00:00
Installing collected packages: pandas
  Attempting uninstall: pandas
    Found existing installation: pandas 2.2.2
    Uninstalling pandas-2.2.2:
      Successfully uninstalled pandas-2.2.2
Successfully installed pandas-2.0.0

The exception that follows later is an expected consequence of combining an incompatible Pandas & NumPy

Not sure why that downgrading is happening. For now have tried restarting CI

It is worth noting before conflicts that same job was passing and nothing in the logic here has changed after those conflicts

If this issue persists, may need help from others to identify what may have changed outside this PR

jakirkham · 2024-08-23T22:41:06Z

Hmm...that job has now failed with a more perplexing error, which is definitely unrelated. Something about having difficulty identifying the CUDA Driver and Runtime versions

+ python -m pytest -p cudf.pandas --cov-config=./python/cudf/.coveragerc --cov=cudf --cov-report=xml:/__w/cudf/cudf/coverage-results/cudf-pandas-coverage.xml --cov-report=term ./python/cudf/cudf_pandas_tests/
/pyenv/versions/3.10.14/lib/python3.10/site-packages/cudf/utils/_ptxcompiler.py:64: UserWarning: Error getting driver and runtime versions:

stdout:



stderr:

Traceback (most recent call last):
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/numba/cuda/cudadrv/driver.py", line 254, in ensure_initialized
    self.cuInit(0)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/numba/cuda/cudadrv/driver.py", line 327, in safe_cuda_api_call
    self._check_ctypes_error(fname, retcode)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/numba/cuda/cudadrv/driver.py", line 395, in _check_ctypes_error
    raise CudaAPIError(retcode, msg)
numba.cuda.cudadrv.driver.CudaAPIError: [999] Call to cuInit results in CUDA_ERROR_UNKNOWN

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<string>", line 4, in <module>
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/numba/cuda/cudadrv/driver.py", line 292, in __getattr__
    self.ensure_initialized()
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/numba/cuda/cudadrv/driver.py", line 258, in ensure_initialized
    raise CudaSupportError(f"Error at driver init: {description}")
numba.cuda.cudadrv.error.CudaSupportError: Error at driver init: Call to cuInit results in CUDA_ERROR_UNKNOWN (999)


Not patching Numba
  warnings.warn(msg, UserWarning)
/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py:331: PluggyTeardownRaisedWarning: A plugin raised an exception during an old-style hookwrapper teardown.
Plugin: helpconfig, Hook: pytest_cmdline_parse
CUDARuntimeError: cudaErrorUnknown: unknown error
For more information see https://pluggy.readthedocs.io/en/stable/api_reference.html#pluggy.PluggyTeardownRaisedWarning
  config = pluginmanager.hook.pytest_cmdline_parse(
Traceback (most recent call last):
  File "/pyenv/versions/3.10.14/lib/python3.10/runpy.py", line 196, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/pyenv/versions/3.10.14/lib/python3.10/runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/pytest/__main__.py", line 5, in <module>
    raise SystemExit(pytest.console_main())
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 192, in console_main
    code = main()
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 150, in main
    config = _prepareconfig(args, plugins)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 331, in _prepareconfig
    config = pluginmanager.hook.pytest_cmdline_parse(
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/pluggy/_hooks.py", line 513, in __call__
    return self._hookexec(self.name, self._hookimpls.copy(), kwargs, firstresult)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/pluggy/_manager.py", line 120, in _hookexec
    return self._inner_hookexec(hook_name, methods, kwargs, firstresult)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/pluggy/_callers.py", line 156, in _multicall
    teardown[0].send(outcome)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/helpconfig.py", line 104, in pytest_cmdline_parse
    config: Config = outcome.get_result()
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/pluggy/_result.py", line 100, in get_result
    raise exc.with_traceback(exc.__traceback__)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/pluggy/_callers.py", line 103, in _multicall
    res = hook_impl.function(*args)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 1075, in pytest_cmdline_parse
    self.parse(args)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 1425, in parse
    self._preparse(args, addopts=addopts)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 1301, in _preparse
    self.pluginmanager.consider_preparse(args, exclude_only=False)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 709, in consider_preparse
    self.consider_pluginarg(parg)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 735, in consider_pluginarg
    self.import_plugin(arg, consider_entry_points=True)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/_pytest/config/__init__.py", line 781, in import_plugin
    __import__(importspec)
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/cudf/__init__.py", line 20, in <module>
    validate_setup()
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/cudf/utils/gpu_utils.py", line 55, in validate_setup
    raise e
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/cudf/utils/gpu_utils.py", line 52, in validate_setup
    gpus_count = getDeviceCount()
  File "/pyenv/versions/3.10.14/lib/python3.10/site-packages/rmm/_cuda/gpu.py", line 102, in getDeviceCount
    raise CUDARuntimeError(status)
rmm._cuda.gpu.CUDARuntimeError: cudaErrorUnknown: unknown error

Restarting again

These Pandas test now try all minor versions of Pandas 2 before Pandas latest. Though only Pandas 2.2.2+ is NumPy 2. However when we downgrade Pandas, NumPy is not downgraded to NumPy 1. As a result, we get the expected errors when trying to mix incompatible NumPy & Pandas versions. This is in part due to Pandas not having an upper bound on its NumPy constraint. Though it is possible `pip` would ignore this anyways. So to fix this issue, just use NumPy 1 for these older versions of Pandas, which should work consistently for them. We have plenty of NumPy 2 coverage outside of this particular test. Especially consider we do run tests with the latest Pandas and NumPy first, which will use NumPy 2 and Pandas 2.2.2+. https://pandas.pydata.org/pandas-docs/version/2.2.2/whatsnew/v2.2.2.html#pandas-2-2-2-is-now-compatible-with-numpy-2-0 https://github.com/pandas-dev/pandas/blob/v2.0.3/pyproject.toml#L26-L28 https://github.com/pandas-dev/pandas/blob/v2.1.4/pyproject.toml#L32-L34

jakirkham · 2024-08-23T23:24:59Z

ci/cudf_pandas_scripts/run_tests.sh

@@ -75,7 +75,7 @@ IFS=',' read -r -a versions <<< "$output"

 for version in "${versions[@]}"; do
    echo "Installing pandas version: ${version}"
-    python -m pip install "pandas==${version}"
+    python -m pip install "numpy>=1.23,<2.0a0" "pandas==${version}"


In the rerun got the original failure with incompatible Pandas & NumPy versions as explained above ( #16300 (comment) ). That said, now see this is due to PR ( #16595 ), which added legacy Pandas testing

Wrote much more details about this in the commit message ( 3bf2311 ), the gist is only Pandas 2.2.2+ is NumPy 2 compatible. However when pip downgrades Pandas here NumPy is not downgraded from NumPy 2 to NumPy 1. As these older Pandas versions don't have upper bounds (restricting to NumPy 1), pip won't downgrade them (though Idk whether that would have been enough for pip anyways)

To fix this issue, am just pinning to NumPy 1 in this install command. Have confirmed this works locally and expect it to work here as well

Since we already test with NumPy 2 above (not to mention in various other tests here), think those tests are sufficient and it is fine to stick to NumPy 1 for these older Pandas versions (even as the long tail includes more NumPy 2 support over time)

cudf/ci/cudf_pandas_scripts/run_tests.sh

Lines 64 to 69 in 3bf2311

python -m pytest -p cudf.pandas \

--cov-config=./python/cudf/.coveragerc \

--cov=cudf \

--cov-report=xml:"${RAPIDS_COVERAGE_DIR}/cudf-pandas-coverage.xml" \

--cov-report=term \

./python/cudf/cudf_pandas_tests/

Confirmed this fixed that job

jakirkham · 2024-08-24T00:58:38Z

One job had an unrelated nvjitlink error:

=================================== FAILURES ===================================
___________ test_arith_masked_vs_constant_reflected[data1-True-mod] ____________
[gw6] linux -- Python 3.10.14 /opt/conda/envs/test/bin/python3.10

self = <pynvjitlink.api.NvJitLinker object at 0xfffdf2679e70>
input_type = <InputType.PTX: 2>
data = b'//\n// Generated by NVIDIA NVVM Compiler\n//\n// Compiler Build ID: CL-31968024\n// Cuda compilation tools, release ...bal.u64 \t%rd62, %rd12;\n\tadd.s64 \t%rd63, %rd62, %rd61;\n\tst.global.u8 \t[%rd63], %rs4;\n\n$L__BB0_10:\n\tret;\n\n}'
name = '<cudapy-ptx>'

    def add_data(self, input_type, data, name):
        if self._complete:
            raise NvJitLinkError("Cannot add data to already-completeted link")
    
        try:
>           _nvjitlinklib.add_data(self.handle, input_type.value, data, name)
E           RuntimeError: NVJITLINK_ERROR_INTERNAL error when calling nvJitLinkAddData

/opt/conda/envs/test/lib/python3.10/site-packages/pynvjitlink/api.py:53: RuntimeError

During handling of the above exception, another exception occurred:

self =    data
0     1
1  <NA>
2     1
func = <function test_arith_masked_vs_constant_reflected.<locals>.func at 0xfffdf277c0d0>
kernel_getter = <function _get_row_kernel at 0xfffe61b3cc10>, args = ()
kwargs = {}

    @acquire_spill_lock()
    @_performance_tracking
    def _apply(self, func, kernel_getter, *args, **kwargs):
        """Apply `func` across the rows of the frame."""
        if kwargs:
            raise ValueError("UDFs using **kwargs are not yet supported.")
        try:
>           kernel, retty = _compile_or_get(
                self, func, args, kernel_getter=kernel_getter
            )

/opt/conda/envs/test/lib/python3.10/site-packages/cudf/core/indexed_frame.py:3472: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
/opt/conda/envs/test/lib/python3.10/site-packages/cudf/utils/performance_tracking.py:51: in wrapper
    return func(*args, **kwargs)
/opt/conda/envs/test/lib/python3.10/site-packages/cudf/core/udf/utils.py:275: in _compile_or_get
    kernel, scalar_return_type = kernel_getter(frame, func, args)
/opt/conda/envs/test/lib/python3.10/site-packages/cudf/core/udf/row_function.py:162: in _get_row_kernel
    kernel = _get_kernel(kernel_string, global_exec_context, sig, func)
            return op(constant, x)
    
        # Just a single column -> result will be all NA
        gdf = cudf.DataFrame({"data": data})
    
        # cudf differs from pandas for 1**NA
        request.applymarker(
            pytest.mark.xfail(
                condition=(constant == 1 and op in {operator.pow, operator.ipow}),
                reason="https://github.com/rapidsai/cudf/issues/7478",
            )
        )
>       run_masked_udf_test(func, gdf, check_dtype=False)

tests/test_udf_masked_ops.py:267: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
tests/test_udf_masked_ops.py:73: in run_masked_udf_test
    obtain = gdf.apply(func, args=args, axis=1)
/opt/conda/envs/test/lib/python3.10/site-packages/cudf/utils/performance_tracking.py:51: in wrapper
    return func(*args, **kwargs)
/opt/conda/envs/test/lib/python3.10/site-packages/cudf/core/dataframe.py:4724: in apply
    return self._apply(func, _get_row_kernel, *args, **kwargs)
/opt/conda/envs/test/lib/python3.10/contextlib.py:79: in inner
    return func(*args, **kwds)
/opt/conda/envs/test/lib/python3.10/site-packages/cudf/utils/performance_tracking.py:51: in wrapper
    return func(*args, **kwargs)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self =    data
0     1
1  <NA>
2     1
func = <function test_arith_masked_vs_constant_reflected.<locals>.func at 0xfffdf277c0d0>
kernel_getter = <function _get_row_kernel at 0xfffe61b3cc10>, args = ()
kwargs = {}

    @acquire_spill_lock()
    @_performance_tracking
    def _apply(self, func, kernel_getter, *args, **kwargs):
        """Apply `func` across the rows of the frame."""
        if kwargs:
            raise ValueError("UDFs using **kwargs are not yet supported.")
        try:
            kernel, retty = _compile_or_get(
                self, func, args, kernel_getter=kernel_getter
            )
        except Exception as e:
>           raise ValueError(
                "user defined function compilation failed."
            ) from e
E           ValueError: user defined function compilation failed.

/opt/conda/envs/test/lib/python3.10/site-packages/cudf/core/indexed_frame.py:3476: ValueError

Merely documenting this here. Will restart that job after the rest complete

Edit: Another job ran into the CUDA Driver error mentioned above ( #16300 (comment) )

jakirkham · 2024-08-24T02:48:23Z

Ah looks like it cleared and the old merge comment took affect

In any event the small tweak to the test requirement above ( #16300 (comment) ) is consistent with what happened before this PR. Just that test condition now test latest Pandas with NumPy 2

If anything comes up or needs discussion, would be happy to chat 🙂

jakirkham · 2024-08-24T06:39:59Z

Updated guidance on dropping NumPy build dependency in comment ( #15193 (comment) )

jakirkham requested a review from a team as a code owner July 17, 2024 22:41

jakirkham requested a review from msarahan July 17, 2024 22:41

github-actions bot added the Python Affects Python cuDF API. label Jul 17, 2024

jakirkham added improvement Improvement / enhancement to an existing function non-breaking Non-breaking change labels Jul 17, 2024

jakirkham changed the title ~~Drop {{ pin_compatible('numpy') }} & build with NumPy 2 (constrain to NumPy 1 though)~~ Drop {{ pin_compatible('numpy') }} & build w/NumPy 2 (restrict to 1) Jul 17, 2024

jakirkham force-pushed the rebld_np_2 branch from 373c34e to af67c0f Compare July 17, 2024 22:52

jakirkham marked this pull request as draft July 17, 2024 23:21

jakirkham force-pushed the rebld_np_2 branch from af67c0f to 3bc1e3d Compare July 18, 2024 19:49

jakirkham changed the title ~~Drop {{ pin_compatible('numpy') }} & build w/NumPy 2 (restrict to 1)~~ Rebuild w/NumPy 2 (restrict to 1) Jul 18, 2024

jakirkham changed the base branch from branch-24.08 to branch-24.10 August 12, 2024 22:08

jakirkham force-pushed the rebld_np_2 branch from 3bc1e3d to ec9e262 Compare August 12, 2024 22:12

seberg reviewed Aug 13, 2024

View reviewed changes

jakirkham force-pushed the rebld_np_2 branch 2 times, most recently from 6b98bce to 44b1a7b Compare August 21, 2024 00:40

jakirkham marked this pull request as ready for review August 21, 2024 00:41

jakirkham mentioned this pull request Aug 21, 2024

Remove NumPy <2 pin #16594

Closed

jakirkham added the 3 - Ready for Review Ready for review by team label Aug 21, 2024

jakirkham force-pushed the rebld_np_2 branch from 44b1a7b to d2693f4 Compare August 21, 2024 01:57

jakirkham requested a review from a team as a code owner August 21, 2024 15:54

jakirkham force-pushed the rebld_np_2 branch from 1b27a6d to d2693f4 Compare August 21, 2024 19:47

jakirkham mentioned this pull request Aug 21, 2024

Drop NumPy build dep from cudf (keep in pylibcudf) #16631

Closed

3 tasks

Remove NumPy <2 pin.

40cba60

jakirkham changed the title ~~Rebuild w/NumPy 2 (restrict to 1)~~ Rebuild for & Support NumPy 2 Aug 21, 2024

jakirkham removed the request for review from a team August 23, 2024 01:08

Merge branch 'branch-24.10' into rebld_np_2

9c86ca4

jakirkham mentioned this pull request Aug 23, 2024

[NO MRG] Test CI with NumPy 2 rapidsai/dask-cuda#1378

Closed

galipremsagar approved these changes Aug 23, 2024

View reviewed changes

raydouglass approved these changes Aug 23, 2024

View reviewed changes

jameslamb self-requested a review August 23, 2024 18:05

jameslamb approved these changes Aug 23, 2024

View reviewed changes

Merge rapidsai/branch-24.10 into jakirkha/rebld_np_2

afe9512

seberg mentioned this pull request Aug 23, 2024

NumPy 2.0 support rapidsai/build-planning#38

Closed

jakirkham added 5 - Ready to Merge Testing and reviews complete, ready to merge and removed 3 - Ready for Review Ready for review by team labels Aug 23, 2024

jakirkham commented Aug 23, 2024

View reviewed changes

Merge branch 'branch-24.10' into rebld_np_2

6d17cf0

rapids-bot bot merged commit 508bdea into rapidsai:branch-24.10 Aug 24, 2024
85 checks passed

jakirkham deleted the rebld_np_2 branch August 24, 2024 02:44

This was referenced Aug 24, 2024

Remove NumPy <2 pin rapidsai/cuml#6031

Merged

Remove NumPy <2 pin rapidsai/cugraph#4615

Merged

Remove NumPy <2 pin, remove proj pin rapidsai/cuspatial#1441

Merged

[FEA] Reduce arrow library dependencies in cudf #15193

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Rebuild for & Support NumPy 2 #16300

Rebuild for & Support NumPy 2 #16300

jakirkham commented Jul 17, 2024 •

edited

Loading

jakirkham commented Jul 17, 2024

seberg Aug 13, 2024

jakirkham Aug 13, 2024

seberg Aug 13, 2024

jakirkham Aug 13, 2024 •

edited

Loading

jakirkham Aug 20, 2024

jameslamb Aug 23, 2024

jakirkham commented Aug 21, 2024

jakirkham commented Aug 21, 2024

vyasr commented Aug 21, 2024

jakirkham commented Aug 21, 2024

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jameslamb left a comment

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jakirkham Aug 23, 2024 •

edited

Loading

jakirkham Aug 24, 2024

jakirkham commented Aug 24, 2024 •

edited

Loading

jakirkham commented Aug 24, 2024

jakirkham commented Aug 24, 2024

	python -m pytest -p cudf.pandas \
	--cov-config=./python/cudf/.coveragerc \
	--cov=cudf \
	--cov-report=xml:"${RAPIDS_COVERAGE_DIR}/cudf-pandas-coverage.xml" \
	--cov-report=term \
	./python/cudf/cudf_pandas_tests/

Rebuild for & Support NumPy 2 #16300

Rebuild for & Support NumPy 2 #16300

Conversation

jakirkham commented Jul 17, 2024 • edited Loading

Description

Checklist

jakirkham commented Jul 17, 2024

seberg Aug 13, 2024

Choose a reason for hiding this comment

jakirkham Aug 13, 2024

Choose a reason for hiding this comment

seberg Aug 13, 2024

Choose a reason for hiding this comment

jakirkham Aug 13, 2024 • edited Loading

Choose a reason for hiding this comment

jakirkham Aug 20, 2024

Choose a reason for hiding this comment

jameslamb Aug 23, 2024

Choose a reason for hiding this comment

jakirkham commented Aug 21, 2024

jakirkham commented Aug 21, 2024

vyasr commented Aug 21, 2024

jakirkham commented Aug 21, 2024

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jameslamb left a comment

Choose a reason for hiding this comment

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jakirkham commented Aug 23, 2024

jakirkham Aug 23, 2024 • edited Loading

Choose a reason for hiding this comment

jakirkham Aug 24, 2024

Choose a reason for hiding this comment

jakirkham commented Aug 24, 2024 • edited Loading

jakirkham commented Aug 24, 2024

jakirkham commented Aug 24, 2024

jakirkham commented Jul 17, 2024 •

edited

Loading

jakirkham Aug 13, 2024 •

edited

Loading

jakirkham Aug 23, 2024 •

edited

Loading

jakirkham commented Aug 24, 2024 •

edited

Loading