numba-cuda 0.9.0__py3-none-any.whl → 0.10.1__py3-none-any.whl

numba_cuda/numba/cuda/tests/cudapy/test_extending.py

@@ -35,15 +35,17 @@ if not config.ENABLE_CUDASIM:
     from numba.core import cgutils
     from numba.core.extending import (
         lower_builtin,
-        make_attribute_wrapper,
         models,
-        register_model,
         type_callable,
         typeof_impl,
     )
     from numba.core.typing.templates import AttributeTemplate
     from numba.cuda.cudadecl import registry as cuda_registry
     from numba.cuda.cudaimpl import lower_attr as cuda_lower_attr
+    from numba.cuda.extending import (
+        register_model,
+        make_attribute_wrapper,
+    )
 
     class IntervalType(types.Type):
         def __init__(self):

numba_cuda/numba/cuda/tests/cudapy/test_inline.py

@@ -0,0 +1,156 @@
+import re
+import numpy as np
+from numba import cuda, types
+from numba.cuda.testing import (
+    unittest,
+    CUDATestCase,
+    skip_on_cudasim,
+)
+
+
+@skip_on_cudasim("Cudasim does not support inline and forceinline")
+class TestCudaInline(CUDATestCase):
+    def _test_call_inline(self, inline):
+        """Test @cuda.jit(inline=...)"""
+        a = np.ones(2, dtype=np.int32)
+
+        sig = (types.int32[::1],)
+
+        @cuda.jit(inline=inline)
+        def set_zero(a):
+            a[0] = 0
+
+        @cuda.jit(sig)
+        def call_set_zero(a):
+            set_zero(a)
+
+        call_set_zero[1, 2](a)
+
+        expected = np.arange(2, dtype=np.int32)
+        self.assertTrue(np.all(a == expected))
+
+        llvm_ir = call_set_zero.inspect_llvm(sig)
+        pat = r"call [a-zA-Z0-9]* @"
+        match = re.compile(pat).search(llvm_ir)
+
+        if inline == "always" or inline is True:
+            # check that call was inlined
+            self.assertIsNone(match, msg=llvm_ir)
+        else:
+            assert inline == "never" or inline is False
+
+            # check that call was not inlined
+            self.assertIsNotNone(match, msg=llvm_ir)
+
+        # alwaysinline should not be in the IR when the inline kwarg is used
+        self.assertNotIn("alwaysinline", llvm_ir)
+
+    def test_call_inline_always(self):
+        self._test_call_inline("always")
+
+    def test_call_inline_never(self):
+        self._test_call_inline("never")
+
+    def test_call_inline_true(self):
+        self._test_call_inline(True)
+
+    def test_call_inline_false(self):
+        self._test_call_inline(False)
+
+    def _test_call_forceinline(self, forceinline):
+        """Test @cuda.jit(forceinline=...)"""
+        a = np.ones(2, dtype=np.int32)
+
+        sig = (types.int32[::1],)
+
+        @cuda.jit(forceinline=forceinline)
+        def set_zero(a):
+            a[0] = 0
+
+        @cuda.jit(sig)
+        def call_set_zero(a):
+            set_zero(a)
+
+        call_set_zero[1, 2](a)
+
+        expected = np.arange(2, dtype=np.int32)
+        self.assertTrue(np.all(a == expected))
+
+        llvm_ir = call_set_zero.inspect_llvm(sig)
+        pat = r"call [a-zA-Z0-9]* @"
+        match = re.compile(pat).search(llvm_ir)
+
+        # Check that call was not inlined at the Numba IR level - the call
+        # should still be present in the IR
+        self.assertIsNotNone(match)
+
+        # Check the definition of set_zero - it is a definition where the
+        # name does not include an underscore just before "set_zero", because
+        # that would match the "call_set_zero" definition
+        pat = r"define.*[^_]set_zero.*"
+        match = re.compile(pat).search(llvm_ir)
+        self.assertIsNotNone(match)
+        if forceinline:
+            self.assertIn("alwaysinline", match.group())
+        else:
+            self.assertNotIn("alwaysinline", match.group())
+
+        # The kernel, "call_set_zero", should never have "alwaysinline" set
+        pat = r"define.*call_set_zero.*"
+        match = re.compile(pat).search(llvm_ir)
+        self.assertIsNotNone(match)
+        self.assertNotIn("alwaysinline", match.group())
+
+    def test_call_forceinline_true(self):
+        self._test_call_forceinline(True)
+
+    def test_call_forceinline_false(self):
+        self._test_call_forceinline(False)
+
+    def test_compile_forceinline_ltoir_only(self):
+        def set_zero(a):
+            a[0] = 0
+
+        args = (types.float32[::1],)
+        msg = r"Can only designate forced inlining in LTO-IR"
+        with self.assertRaisesRegex(ValueError, msg):
+            cuda.compile(
+                set_zero,
+                args,
+                device=True,
+                forceinline=True,
+            )
+
+    def _compile_set_zero(self, forceinline):
+        def set_zero(a):
+            a[0] = 0
+
+        args = (types.float32[::1],)
+        ltoir, resty = cuda.compile(
+            set_zero,
+            args,
+            device=True,
+            output="ltoir",
+            forceinline=forceinline,
+        )
+
+        # Sanity check
+        self.assertEqual(resty, types.none)
+
+        return ltoir
+
+    def test_compile_forceinline(self):
+        ltoir_noinline = self._compile_set_zero(False)
+        ltoir_forceinline = self._compile_set_zero(True)
+
+        # As LTO-IR is opaque, the best we can do is check that changing the
+        # flag resulted in a change in the generated LTO-IR in some way.
+        self.assertNotEqual(
+            ltoir_noinline,
+            ltoir_forceinline,
+            "forceinline flag appeared to have no effect on LTO-IR",
+        )
+
+
+if __name__ == "__main__":
+    unittest.main()

numba_cuda/numba/cuda/tests/cudapy/test_laplace.py

@@ -14,7 +14,7 @@ SM_SIZE = tpb, tpb
 
 class TestCudaLaplace(CUDATestCase):
     def test_laplace_small(self):
-        @cuda.jit(float64(float64, float64), device=True, inline=True)
+        @cuda.jit(float64(float64, float64), device=True, inline="always")
         def get_max(a, b):
             if a > b:
                 return a

numba_cuda/numba/cuda/tests/cudapy/test_warp_ops.py

@@ -1,6 +1,9 @@
+import re
+
 import numpy as np
 from numba import cuda, int32, int64, float32, float64
 from numba.cuda.testing import unittest, CUDATestCase, skip_on_cudasim
+from numba.cuda.compiler import compile_ptx
 from numba.core import config
 
 
@@ -144,6 +147,47 @@ class TestCudaWarpOperations(CUDATestCase):
         compiled[1, nelem](ary, xor)
         self.assertTrue(np.all(ary == exp))
 
+    def test_shfl_sync_const_mode_val(self):
+        # Test `mode` argument is constant in shfl_sync calls.
+        # Related to https://github.com/NVIDIA/numba-cuda/pull/231
+        subtest = [
+            (use_shfl_sync_idx, 4),
+            (use_shfl_sync_up, 4),
+            (use_shfl_sync_down, 4),
+            (use_shfl_sync_xor, 16),
+        ]
+
+        args_re = r"\((.*)\)"
+        m = re.compile(args_re)
+
+        for func, value in subtest:
+            with self.subTest(func=func.__name__):
+                compiled = cuda.jit("void(int32[:], int32)")(func)
+                nelem = 32
+                ary = np.empty(nelem, dtype=np.int32)
+                compiled[1, nelem](ary, value)
+                irs = next(iter(compiled.inspect_llvm().values()))
+
+                for ir in irs.split("\n"):
+                    if "call" in ir and "llvm.nvvm.shfl.sync.i32" in ir:
+                        args = m.search(ir).group(0)
+                        arglist = args.split(",")
+                        mode_arg = arglist[1]
+                        self.assertNotIn("%", mode_arg)
+
+    def test_shfl_sync_const_mode_val_sm100(self):
+        # Test shfl_sync compiles with cc=(10, 0)
+        subtest = [
+            use_shfl_sync_idx,
+            use_shfl_sync_up,
+            use_shfl_sync_down,
+            use_shfl_sync_xor,
+        ]
+
+        for func in subtest:
+            with self.subTest(func=func.__name__):
+                compile_ptx(func, (int32[:], int32), cc=(10, 0))
+
     def test_shfl_sync_types(self):
         types = int32, int64, float32, float64
         values = (
@@ -153,11 +197,12 @@ class TestCudaWarpOperations(CUDATestCase):
             np.float64(np.pi),
         )
         for typ, val in zip(types, values):
-            compiled = cuda.jit((typ[:], typ))(use_shfl_sync_with_val)
-            nelem = 32
-            ary = np.empty(nelem, dtype=val.dtype)
-            compiled[1, nelem](ary, val)
-            self.assertTrue(np.all(ary == val))
+            with self.subTest(typ=typ):
+                compiled = cuda.jit((typ[:], typ))(use_shfl_sync_with_val)
+                nelem = 32
+                ary = np.empty(nelem, dtype=val.dtype)
+                compiled[1, nelem](ary, val)
+                self.assertTrue(np.all(ary == val))
 
     def test_vote_sync_all(self):
         compiled = cuda.jit("void(int32[:], int32[:])")(use_vote_sync_all)

numba_cuda/numba/cuda/vector_types.py

@@ -5,13 +5,15 @@ from typing import List, Tuple, Dict
 
 from numba import types
 from numba.core import cgutils
-from numba.core.extending import make_attribute_wrapper, models, register_model
+from numba.core.extending import models
 from numba.core.imputils import Registry as ImplRegistry
 from numba.core.typing.templates import ConcreteTemplate
 from numba.core.typing.templates import Registry as TypingRegistry
 from numba.core.typing.templates import signature
 from numba.cuda import stubs
 from numba.cuda.errors import CudaLoweringError
+from numba.cuda.extending import make_attribute_wrapper, register_model
+
 
 typing_registry = TypingRegistry()
 impl_registry = ImplRegistry()

numba_cuda/numba/cuda/vectorizers.py

@@ -206,7 +206,7 @@ def __vectorized_{name}({args}, __out__):
 
 class CUDAVectorize(deviceufunc.DeviceVectorize):
     def _compile_core(self, sig):
-        cudevfn = cuda.jit(sig, device=True, inline=True)(self.pyfunc)
+        cudevfn = cuda.jit(sig, device=True, inline="always")(self.pyfunc)
         return cudevfn, cudevfn.overloads[sig.args].signature.return_type
 
     def _get_globals(self, corefn):

{numba_cuda-0.9.0.dist-info → numba_cuda-0.10.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: numba-cuda
-Version: 0.9.0
+Version: 0.10.1
 Summary: CUDA target for Numba
 Author: Anaconda Inc., NVIDIA Corporation
 License: BSD 2-clause

{numba_cuda-0.9.0.dist-info → numba_cuda-0.10.1.dist-info}/RECORD

@@ -1,6 +1,6 @@
 _numba_cuda_redirector.pth,sha256=cmfMMmV0JPh3yEpl4bGeM9AuXiVVMSo6Z_b7RaQL3XE,30
 _numba_cuda_redirector.py,sha256=n_r8MYbu5-vcXMnLJW147k8DnFXXvgb7nPIXnlXwTyQ,2659
-numba_cuda/VERSION,sha256=nYyU8a0-qWseKsSRT9pMuTx2tKPg2Mxt2JdtbAsifRU,6
+numba_cuda/VERSION,sha256=9NQ54LUjIIoJ0ThiwWggzDAo_ZRBcxDOHVOjHRTWosQ,7
 numba_cuda/__init__.py,sha256=atXeUvJKR3JHcAiCFbXCVOJQUHgB1TulmsqSL_9RT3Q,114
 numba_cuda/_version.py,sha256=nzrrJXi85d18m6SPdsPsetJNClDETkmF1MrEhGLYDBs,734
 numba_cuda/numba/cuda/__init__.py,sha256=3siqMXEKqa9ezQ8RxPC3KMdebUjgJt-EKxxV4CX9818,607
@@ -9,30 +9,30 @@ numba_cuda/numba/cuda/api_util.py,sha256=jK8oUD3zf_D5IX7vbjc3uY_5kmOxwgEqO2m_lDH
 numba_cuda/numba/cuda/args.py,sha256=UlTHTJpwPeCtnW0Bb-Wetm5UO9TPR-PCgIt5ys8b8tQ,1894
 numba_cuda/numba/cuda/cg.py,sha256=azz1sIT_jXQfJEZfDjBeqboJc6Pu_NtrZxfE7D1eQLQ,1484
 numba_cuda/numba/cuda/codegen.py,sha256=4hAdztvCcpwVbWcl9b5zK9xu04f7mVMNAgekpfc-8uw,14049
-numba_cuda/numba/cuda/compiler.py,sha256=I4fviK5cuIr8zqXaJHTC0jDuH96E2IBRs58BqRQf0SU,24130
+numba_cuda/numba/cuda/compiler.py,sha256=sFreZM07D8zp4QyUBL2IKoBtDjzdxj80wN4KUgEQOS8,25283
 numba_cuda/numba/cuda/cpp_function_wrappers.cu,sha256=8lUPmU6FURxphzEqkPLZRPYBCEK_wmDtHq2voPkckfs,950
-numba_cuda/numba/cuda/cuda_fp16.h,sha256=8Ss_QAi5Ij9Dv_o08ur-PovyLgH7VyDhdjsEqBf7xcI,126180
-numba_cuda/numba/cuda/cuda_fp16.hpp,sha256=ojrzEMG6WM7X3_jglFCKZ8-cORfHDJ53nGVCPpcJsZI,89127
-numba_cuda/numba/cuda/cuda_paths.py,sha256=RbYexjtLbhsywDc_eR1KGayBvZ3cBqwb4As0QKgtIAI,15812
-numba_cuda/numba/cuda/cudadecl.py,sha256=9R7T-d_8o67auSuXNQi4pI_sf64C9Ax4x6XKAMQgaw8,23313
-numba_cuda/numba/cuda/cudaimpl.py,sha256=eiF9KPhPAJUuWA_yB5ZXWYUweYLhTZ77TG5X84jtS88,38273
+numba_cuda/numba/cuda/cuda_bf16.py,sha256=RfnWMV2_zSAW9FLN4JqfW6GfmWR8ZVO16e9Bw3jZnto,152203
+numba_cuda/numba/cuda/cuda_paths.py,sha256=kMIJ_1yV2qtcKEM5rCgSDJ3Gz7bgxbfAWh54E5cDndg,15872
+numba_cuda/numba/cuda/cudadecl.py,sha256=4DhYDnKg95AKsmDHetJvL1rfdvhnuz9PKS1Ncf4nO20,22343
+numba_cuda/numba/cuda/cudaimpl.py,sha256=-a5dvGHORH4RypGliHqXvwG3Rc0CAJVntYGxoYHmbpc,35656
 numba_cuda/numba/cuda/cudamath.py,sha256=wbGjlyGVwcUAoQjgXIaAaasLdVuDSKHkf6KyID5IYBw,3979
-numba_cuda/numba/cuda/debuginfo.py,sha256=JCdtmIKGD8pob2lu2trwDavYovwsDqz6yQ6G0kuCGU8,1495
-numba_cuda/numba/cuda/decorators.py,sha256=W6S5G87Eo5c-y2nW8DbhNPhpSNNpd6bEGjRPbtHYyP8,8245
+numba_cuda/numba/cuda/debuginfo.py,sha256=tWlRAC1-AsSQp0pG9kXQY9tlVdZPA-nDUJsrvru4eaM,4504
+numba_cuda/numba/cuda/decorators.py,sha256=kqzbv7eEQSyQg2G_XtIyKIfvmm354jw2vZDlOmK-t9s,9454
 numba_cuda/numba/cuda/descriptor.py,sha256=t1rSVJSCAlVACC5_Un3FQ7iubdTTBe-euqz88cvs2tI,985
-numba_cuda/numba/cuda/device_init.py,sha256=lPh7zssW8q88B3oISb1muRq9unBY458u4VJeY3DveTM,3474
+numba_cuda/numba/cuda/device_init.py,sha256=Rtwd6hQMHMLMkj6MXtndbWYFJfkIaRe0MwOIJF2nzhU,3449
 numba_cuda/numba/cuda/deviceufunc.py,sha256=zj9BbLiZD-dPttHew4olw8ANgR2nXnXEE9qjCeGLrQI,30731
-numba_cuda/numba/cuda/dispatcher.py,sha256=_lEKvUcystUwgMvEyT3lCuvi41OULn0VE3H36HQ21o8,44369
+numba_cuda/numba/cuda/dispatcher.py,sha256=uX6ltCDQq9mIBqSHV6Ci-2mJtuAmeZXBb3yWp8gXZ2U,44426
 numba_cuda/numba/cuda/errors.py,sha256=WRso1Q_jCoWP5yrDBMhihRhhVtVo1-7KdN8QVE9j46o,1712
-numba_cuda/numba/cuda/extending.py,sha256=2g_YgNqTSnoe08s24XOnj5xNhLUsnS8JM96OpSJuj84,142
+numba_cuda/numba/cuda/extending.py,sha256=VwuU5F0AQFlJsqaiwoWk-6Itihew1FsjVT_BVjhY8Us,2278
 numba_cuda/numba/cuda/initialize.py,sha256=0SnpjccQEYiWITIyfAJx833H1yhYFFDY42EpnwYyMn8,487
-numba_cuda/numba/cuda/intrinsic_wrapper.py,sha256=B7oUiPYxglegicKawNgYCEv7ddcC2gU7J9xK8HO1T0s,2239
-numba_cuda/numba/cuda/intrinsics.py,sha256=tmy4PqDuMkcUSxabMar-WGWaaiSWaoyyhY56wybQ2E8,6106
+numba_cuda/numba/cuda/intrinsic_wrapper.py,sha256=-b7w5ywqW-upPG3WCKjRS9AWKzAmHKpie15BHnJd5vY,1075
+numba_cuda/numba/cuda/intrinsics.py,sha256=G57JZ-DM-wgvJm8FeIpXErX1sQkkafP95Y1OG3CksR0,11896
 numba_cuda/numba/cuda/libdevice.py,sha256=jOeNrjs6K08lv7P6eENLIVUMP-WJRR86a7Hco1y1B94,61333
 numba_cuda/numba/cuda/libdevicedecl.py,sha256=xdZbb_rCaftMf8Pbw63g_Lr230N-1QoaYzBxq8udKTg,532
 numba_cuda/numba/cuda/libdevicefuncs.py,sha256=c80lGpGoFIYkAdgr4fzbxzdNCyJYrLdss64bwa0Mc6w,37471
 numba_cuda/numba/cuda/libdeviceimpl.py,sha256=m4Fog_OPPEg2RkOk7LEeqF26MK4aEFlKxITlSCZKMAo,2798
 numba_cuda/numba/cuda/locks.py,sha256=yF6WcwMyzauJ9H7JuCRq2Ynx7kFVAnlkkvmWp7UdZ5w,388
+numba_cuda/numba/cuda/lowering.py,sha256=6XXpTRfTBTVHPh1M4jVAL9APvKk1UWSb-A5WJTEMsqQ,1602
 numba_cuda/numba/cuda/mathimpl.py,sha256=-8IOkhorbMg8iPBMIdgjk3qJZSyRWYJDwPAWrTMkODI,14356
 numba_cuda/numba/cuda/models.py,sha256=jbvmbL51mt0Z1nZTSiniBJTFhnOfPzzcVD6xCEpXDMA,1282
 numba_cuda/numba/cuda/nvvmutils.py,sha256=x-0nCqwkoB8DzX7bSrvTH0h-aKSDx0rVWKR7Eqx4ldA,7993
@@ -40,14 +40,14 @@ numba_cuda/numba/cuda/printimpl.py,sha256=AO_KvkKhlJacjoq8IV1nDm7YBNKnqN7SBkvTG1
 numba_cuda/numba/cuda/random.py,sha256=V30KaFdkuDyjxoP14awz-KkY3lRIXqIZuuH27UotINE,10451
 numba_cuda/numba/cuda/reshape_funcs.cu,sha256=frw1uoeMSYlkPC38LiKE8Tz2P70X2e4UZGyLKkaPzho,4326
 numba_cuda/numba/cuda/simulator_init.py,sha256=Hvzty6NJp1SeKspyb-b887xpeNLMMI0x9aPmV--X77E,450
-numba_cuda/numba/cuda/stubs.py,sha256=0IH4puoOizBShZV_bvYvXVCO0aSFdxckhwTdWm4zQuk,22389
-numba_cuda/numba/cuda/target.py,sha256=dtORdiwGyippoTOPvda_QTDd0YKk7-oLXXZvpt_c_HI,11285
+numba_cuda/numba/cuda/stubs.py,sha256=mCS65wc4MuaDnL_XYYxkKbDOH991o6a0JsN9KrLNMGQ,22104
+numba_cuda/numba/cuda/target.py,sha256=mSMnS-bSsC8_4KqkAsa1Byi2mO8jPJdKW3m31qxsxUE,12520
 numba_cuda/numba/cuda/testing.py,sha256=OR37AuDdzg7vLG4G_4s2uRAkNTScZc-BzHmTMJYuxhQ,6827
 numba_cuda/numba/cuda/types.py,sha256=hC1MUvgUwy-SLgbzFzXwssJzPR8BxQwqUcjwGJFzVac,1317
 numba_cuda/numba/cuda/ufuncs.py,sha256=AJifQgapyv62fdJeMm939R1I5TvIRmaA8dJ83Jy8DCw,23559
 numba_cuda/numba/cuda/utils.py,sha256=Bk9TZZerYrnAaeKjjAAYkbm6YoP0ptxcPrCysRi_nRI,631
-numba_cuda/numba/cuda/vector_types.py,sha256=8JYxlrR3EJTDiFYRcElopwvzVXZQslAPHLW9ZvWRJu0,6750
-numba_cuda/numba/cuda/vectorizers.py,sha256=4YFwbcJggu96raPhyHGUCeIWZi3VYfloZh7xlHUPakc,8383
+numba_cuda/numba/cuda/vector_types.py,sha256=FlzOKufhvBnZ-VC-liA7y9is8BV-uj0fD-En_vP6zl0,6783
+numba_cuda/numba/cuda/vectorizers.py,sha256=nEfQxjSA4oCX8ZzvoqjDRygDfwzxFVDXtnjx-K1aPqA,8387
 numba_cuda/numba/cuda/cudadrv/__init__.py,sha256=inat2K8K1OVrgDe64FK7CyRmyFyNKcNO4p2_L79yRZ0,201
 numba_cuda/numba/cuda/cudadrv/devicearray.py,sha256=6tF2TYnmjMbKk2fho1ONoD_QsRD9QVTT2kHP7x1u1J0,31556
 numba_cuda/numba/cuda/cudadrv/devices.py,sha256=k87EDIRhj1ncM9PxJCjZGPFfEks99vzmHlTc55GK5X0,8062
@@ -57,13 +57,21 @@ numba_cuda/numba/cuda/cudadrv/dummyarray.py,sha256=2jycZhniMy3ncoVWQG9D8dBehTEeo
 numba_cuda/numba/cuda/cudadrv/enums.py,sha256=raWKryxamWQZ5A8ivMpyYVhhwbSpaD9lu7l1_wl2W9M,23742
 numba_cuda/numba/cuda/cudadrv/error.py,sha256=C2tTPT5h3BGgzjaFTCqbY7hOk2PgkVh0iuM1EiRp1eI,583
 numba_cuda/numba/cuda/cudadrv/libs.py,sha256=qjknQxYXd2ucwDLQqzhWC_srNg6FnwvcVHIpKyPxJ9A,7287
-numba_cuda/numba/cuda/cudadrv/linkable_code.py,sha256=ltsRRGFvuJ2nU5axf1rzVKR_EPRiImiz1q5-lYmZxJA,2256
+numba_cuda/numba/cuda/cudadrv/linkable_code.py,sha256=bgXfXIVLx-R5BGr6aiORJ8uWakMl_2dh1SxDn9fH8EI,2582
 numba_cuda/numba/cuda/cudadrv/mappings.py,sha256=9uEs1KepeVGRbEpVhLjtxSsvZpZsbrHnPywmx--y88A,804
 numba_cuda/numba/cuda/cudadrv/ndarray.py,sha256=HtULWWFyDlgqvrH5459yyPTvU4UbUo2DSdtcNfvbH00,473
-numba_cuda/numba/cuda/cudadrv/nvrtc.py,sha256=kkc4rElfZcvi3-lDoxd6DlYPDjQjKJGtgvQ1kS4_JnU,14096
+numba_cuda/numba/cuda/cudadrv/nvrtc.py,sha256=6xtAR1af5BsBkDMJcQsTIUFFO02wwpfLClNIsh5L33Y,14324
 numba_cuda/numba/cuda/cudadrv/nvvm.py,sha256=7tTy6-VEbMBpDUmuSMnUwqPFfBndTh3aPq_n7nxhEA0,26344
 numba_cuda/numba/cuda/cudadrv/rtapi.py,sha256=J6PRGGK07XSLRzgCw5xs8VU5xVoqavvhojk1mxiQsi4,226
 numba_cuda/numba/cuda/cudadrv/runtime.py,sha256=CFumwg4iblWap_E7l7GM_hMYz1PsbH81-N0tZwFFooA,4372
+numba_cuda/numba/cuda/include/11/cuda_bf16.h,sha256=Z7HGJEOhMjQzD0Gs0eq0qdzD-Wr8Zbty-FeeLtahN-s,138713
+numba_cuda/numba/cuda/include/11/cuda_bf16.hpp,sha256=NQQkLp3doxSllrkNOjbl_fKigtGsKccNd1g-NJ7G-2k,101350
+numba_cuda/numba/cuda/include/11/cuda_fp16.h,sha256=s77PdhSlN_b2KqymCX_gUwIc4JWcqOeAkvsgJDVQRhg,131916
+numba_cuda/numba/cuda/include/11/cuda_fp16.hpp,sha256=evyhsAqX9BFZoLi4AtTjQuPebuXsmihSrm9fXVc1pKE,93578
+numba_cuda/numba/cuda/include/12/cuda_bf16.h,sha256=B3m8XisQyXlu6dkJrgmWaiQYf9xWAJDNZK1qlCTIcBc,203796
+numba_cuda/numba/cuda/include/12/cuda_bf16.hpp,sha256=mAeEsTy3N5ZrAM-r-7FTjP90sOLzFfKXY3CrFbaxceg,136542
+numba_cuda/numba/cuda/include/12/cuda_fp16.h,sha256=5AlJ5h6KS8G-ecZWFzyJJhtLtSaFN3ahsevB7HwTnQ0,206162
+numba_cuda/numba/cuda/include/12/cuda_fp16.hpp,sha256=o1ITDmuN67N8YUGUcvTpV3IdpS-6wwlm65M_H-8LYKs,120927
 numba_cuda/numba/cuda/kernels/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 numba_cuda/numba/cuda/kernels/reduction.py,sha256=RsVubg8uNumxNxo9HBlFVCDicA-KZKsksKId0ktgQyY,9101
 numba_cuda/numba/cuda/kernels/transpose.py,sha256=FbtFmOqaj_e7ARR_kkiTpSvj4BJyqBta5ci1CWtJ690,2033
@@ -125,15 +133,16 @@ numba_cuda/numba/cuda/tests/cudadrv/test_streams.py,sha256=rrQEA8iawR6UyKnK2MdI5
 numba_cuda/numba/cuda/tests/cudapy/__init__.py,sha256=43EXdiXXRBd6yIcVGMrU9F_EJCD9Uw3mzOP3SB53AEE,260
 numba_cuda/numba/cuda/tests/cudapy/cache_usecases.py,sha256=FnvjeqTZ-YBmroHctPrHgMHxnJ-HiT9KI79aHTej5G8,5840
 numba_cuda/numba/cuda/tests/cudapy/cache_with_cpu_usecases.py,sha256=9CbjosLNPN5IzrD-15sD_4B0BMmjo02Y7faZiS82cyk,1143
-numba_cuda/numba/cuda/tests/cudapy/extensions_usecases.py,sha256=UATKkU16Ki5To99XFQ7tAT2DoMbB0ECDQuX3HtGwBh4,1606
+numba_cuda/numba/cuda/tests/cudapy/extensions_usecases.py,sha256=2i_xq4B1t1tctr6ZrWA29ZHkmQlD_vCSewhr-AT9tMc,1651
 numba_cuda/numba/cuda/tests/cudapy/jitlink.ptx,sha256=PKVafUhDH1SKRWXkt4N3v8SDMh4RyDFiJM-CMksa5uc,519
 numba_cuda/numba/cuda/tests/cudapy/recursion_usecases.py,sha256=wrWx8AeRhBHM74iYPKKrZqiyWrYCtQU3J-g3Zv7JmoY,1782
 numba_cuda/numba/cuda/tests/cudapy/test_alignment.py,sha256=RkhAcVkGtze8JpZTlYYvqTesDYE7xfKQZd1izgxDQpU,1219
 numba_cuda/numba/cuda/tests/cudapy/test_array.py,sha256=lT7XWXl0_lqtXyyXN-w0cd0wH7EBklRElYDnHUM5G1I,13215
-numba_cuda/numba/cuda/tests/cudapy/test_array_args.py,sha256=5gw1bW782Xjk06oJ2eOggCNJc9qLC7noE7fnpbf8AnM,4978
+numba_cuda/numba/cuda/tests/cudapy/test_array_args.py,sha256=iiFrt5Yn7gfheAGOYG2VBeWeuW3JlBhRLXNfSz4cHAA,4982
 numba_cuda/numba/cuda/tests/cudapy/test_array_methods.py,sha256=SWa1MvpwG07yBkrFIUeM9pm3BIwUbhttMNBdUW-CpSM,969
 numba_cuda/numba/cuda/tests/cudapy/test_atomics.py,sha256=agsfUN3WOoh6ICAECtuMuxZNcKq5ivK30Ew3h_m76m0,57689
-numba_cuda/numba/cuda/tests/cudapy/test_blackscholes.py,sha256=fvmmlRg93F4_I1E2_ALe2laZsP3O7ncTzxMW-mDyMkM,4413
+numba_cuda/numba/cuda/tests/cudapy/test_bfloat16_bindings.py,sha256=NYLa_e60NYc63X7japCAsjUS84lXn92k4_S_E6-sEX4,6779
+numba_cuda/numba/cuda/tests/cudapy/test_blackscholes.py,sha256=0_wr6MSeHh0QVzPeH8SB7j0Nv_RrPAK01hNoQ_dGT5I,4417
 numba_cuda/numba/cuda/tests/cudapy/test_boolean.py,sha256=j4mIOv4rJTLjJzpKk1O9UFLT41_iOQRtwsmteXdKZ-M,547
 numba_cuda/numba/cuda/tests/cudapy/test_caching.py,sha256=qbNisdxvoErKlDkD5dw7IkdJhfcQUpIdfHX11UzGBOo,18990
 numba_cuda/numba/cuda/tests/cudapy/test_casting.py,sha256=3LaN3ZsSuOZXAZXCV85wYyhh0ih7JqABnjGTa7Y2YBE,8748
@@ -148,13 +157,13 @@ numba_cuda/numba/cuda/tests/cudapy/test_cuda_array_interface.py,sha256=RXCNHAZM3
 numba_cuda/numba/cuda/tests/cudapy/test_cuda_jit_no_types.py,sha256=8prL2FTiaajW-UHSL9al-nBniygOfpdAOT_Dkej4PWI,2138
 numba_cuda/numba/cuda/tests/cudapy/test_datetime.py,sha256=MnOeDWMz-rL3-07FsswM06Laxmm0KjTmTwhrP3rmchQ,3526
 numba_cuda/numba/cuda/tests/cudapy/test_debug.py,sha256=1P369s02AvGu7fSIEe_YxSgh3c6S72Aw1gRgmepDbQY,3383
-numba_cuda/numba/cuda/tests/cudapy/test_debuginfo.py,sha256=FV3rsPrmbAJakNPXKYrVe-T2m3njPMxn8k9l7vLcZG4,11070
+numba_cuda/numba/cuda/tests/cudapy/test_debuginfo.py,sha256=796d8Oa1ZV2mZ9LTcwR3g6_j5sjSBk7kZEHYMOXPBfU,12606
 numba_cuda/numba/cuda/tests/cudapy/test_device_func.py,sha256=LNGBZfqFGUtVVQeC6FcHo8T3DbG-j6AjeBwJmwp9HH4,13157
 numba_cuda/numba/cuda/tests/cudapy/test_dispatcher.py,sha256=Oc6CdI1j9Ad_wklHdIYSMytrzUpzK6oXD0BGe45sTwg,26636
-numba_cuda/numba/cuda/tests/cudapy/test_enums.py,sha256=2P5hRmI77UVRQXfovefN98VkAyH8t9n8CTKPKCIQt5A,3562
+numba_cuda/numba/cuda/tests/cudapy/test_enums.py,sha256=Yxac6S5P6C8GN0kMwieL3dQb1uogOVZQEx969B0AMpM,4533
 numba_cuda/numba/cuda/tests/cudapy/test_errors.py,sha256=w6ipW9UIvUD_ZIt_6fQ-uJsHyKLyHVqv2bym-9vyGyY,2757
 numba_cuda/numba/cuda/tests/cudapy/test_exception.py,sha256=W5NF022DOOTaEjFmhfr8BnfhRXvYyXHiGwznQrm_9T4,5507
-numba_cuda/numba/cuda/tests/cudapy/test_extending.py,sha256=UcQVHvT-cS3Fx1oPClWMJfnpXcU_UfyfnZ3IQ-O17Zk,4099
+numba_cuda/numba/cuda/tests/cudapy/test_extending.py,sha256=2QWcl8yJvp0A22V8qItJuzVvmt5Ng1JdhqR7hn5XX0E,4144
 numba_cuda/numba/cuda/tests/cudapy/test_fastmath.py,sha256=fiUoOiwWjctZNFN-DGw1A8eGfHLqNulo2OQ7v1DFS9o,8552
 numba_cuda/numba/cuda/tests/cudapy/test_forall.py,sha256=Ory5s-_9MauSCP2RuWUEmcGFvP0kS7ytV-3iYPFYR6o,1470
 numba_cuda/numba/cuda/tests/cudapy/test_freevar.py,sha256=JvWn7Lw137HI61mouKnPvDxZIqLppiCF_351osxQQYE,753
@@ -164,12 +173,13 @@ numba_cuda/numba/cuda/tests/cudapy/test_gufunc.py,sha256=Rl35HQdN6J3ZPjSLIz2mFJx
 numba_cuda/numba/cuda/tests/cudapy/test_gufunc_scalar.py,sha256=vAP2ggp2arBqJS8kNbGeC5jrZuYzLtFstgvxX0PI-I0,5322
 numba_cuda/numba/cuda/tests/cudapy/test_gufunc_scheduling.py,sha256=1USofSlavYFaVhP8oep5oJ-CLzXxYwkI3EtOkY6jrVw,2610
 numba_cuda/numba/cuda/tests/cudapy/test_idiv.py,sha256=tTy7hN2LJ4897UzO3EUxjuUzbBcs9QITHJu3s_eknq0,1054
+numba_cuda/numba/cuda/tests/cudapy/test_inline.py,sha256=T7DHquV_4HuX5fFQQS3kcZzgifTzwYbMFiY7SgQzoLA,4584
 numba_cuda/numba/cuda/tests/cudapy/test_inspect.py,sha256=L9-62nPmiWC90PST5EZrnGdAcrsbhMS_mbEkwdDkFQ0,4901
 numba_cuda/numba/cuda/tests/cudapy/test_intrinsics.py,sha256=uQ0S_XXds-F9Z5GhuFYzRVXu5XYD1ULa-y55Wi92i5I,36726
 numba_cuda/numba/cuda/tests/cudapy/test_ipc.py,sha256=bNT6UZgsgeVWyzBrlKXucQW6IKcD6NEmbwV5cFhf-7I,10553
 numba_cuda/numba/cuda/tests/cudapy/test_iterators.py,sha256=WCRkQfkEnB0d9aj55dVvyQzD4QxrOLubnlKO0xTiNto,2343
 numba_cuda/numba/cuda/tests/cudapy/test_lang.py,sha256=TP1spLeJfmBKKrU7G3bvkhNPvVm-oQX134taQsZeNbE,1693
-numba_cuda/numba/cuda/tests/cudapy/test_laplace.py,sha256=_S3AiFK9ws-3nB8jUy6rrtvlcB7eUD5Ylx7RS3uApu8,3199
+numba_cuda/numba/cuda/tests/cudapy/test_laplace.py,sha256=ZE5dOhI3NUZUyvygM480DfFwP8dlV-s1lhfXq6Lm2ro,3203
 numba_cuda/numba/cuda/tests/cudapy/test_libdevice.py,sha256=YVGdHBh2FOYxICTVugN14VGldvJyzOsdAnbH8TCZqMI,6531
 numba_cuda/numba/cuda/tests/cudapy/test_lineinfo.py,sha256=HuJWaeRDzQV-91PwpILxsM1HjvlFJp_w9qVXhJTxYgw,6693
 numba_cuda/numba/cuda/tests/cudapy/test_localmem.py,sha256=1i2ECv31bFnSBO_pVgmwtSz-pxyww1qs1MLj3FtY_FA,5368
@@ -209,7 +219,7 @@ numba_cuda/numba/cuda/tests/cudapy/test_vectorize_decor.py,sha256=ouA7bMO8L87mzQ
 numba_cuda/numba/cuda/tests/cudapy/test_vectorize_device.py,sha256=gNbVk8-Uv3jm795-zWyig2JXb4Jo4iB5hyiHH6YLCiE,987
 numba_cuda/numba/cuda/tests/cudapy/test_vectorize_scalar_arg.py,sha256=VqBNr8SdEhldCZLP8MdxMrE2CXJUc6yqhGNB_Aj20wE,933
 numba_cuda/numba/cuda/tests/cudapy/test_warning.py,sha256=7xHFLmVvYlQUoOdush4lDBWwrBDe8Z9vN0NYciNP5QA,5716
-numba_cuda/numba/cuda/tests/cudapy/test_warp_ops.py,sha256=LeNFBU3BYKysQgcJQ8R_LvXqAX9PbZgDHx5OzcuPhlk,8999
+numba_cuda/numba/cuda/tests/cudapy/test_warp_ops.py,sha256=sDMvdudkePl10Vq7rco67vGL7IDpCBWdOGQuL0UPeG4,10602
 numba_cuda/numba/cuda/tests/cudasim/__init__.py,sha256=GdfSq6pRVSOQwmgNi7ZFQ5l0yg4-2gNar_0Rz0buUpM,157
 numba_cuda/numba/cuda/tests/cudasim/support.py,sha256=JjRrfrrLKS0V5p6GX6ibs6QTuFb1NanKfBQSgbLeiHs,114
 numba_cuda/numba/cuda/tests/cudasim/test_cudasim_issues.py,sha256=-GJCl2c063Ig6EUB8w5L_0GcmXzTLatGe_ddEzdnbgc,3177
@@ -246,8 +256,8 @@ numba_cuda/numba/cuda/tests/test_binary_generation/Makefile,sha256=P2WzCc5d64JGq
 numba_cuda/numba/cuda/tests/test_binary_generation/generate_raw_ltoir.py,sha256=SE5FrbZdkVrnzS0R62YPPyH25r6Jevd2nuB6HRJ3PZ0,5011
 numba_cuda/numba/cuda/tests/test_binary_generation/test_device_functions.cu,sha256=cUf-t6ZM9MK_x7X_aKwsrKW1LdR97XcpR-qnYr5faOE,453
 numba_cuda/numba/cuda/tests/test_binary_generation/undefined_extern.cu,sha256=q3oxZziT8KDodeNcEBiWULH6vMrHCWucmJmtrg8C0d0,128
-numba_cuda-0.9.0.dist-info/licenses/LICENSE,sha256=eHeYE-XjASmwbxfsP5AImgfzRwZurZGqH1f6OFwJ4io,1326
-numba_cuda-0.9.0.dist-info/METADATA,sha256=-r7qoK8WPBoXcxCVrYFLo5pdjRaycv2UZ8cchnR00rA,1858
-numba_cuda-0.9.0.dist-info/WHEEL,sha256=pxyMxgL8-pra_rKaQ4drOZAegBVuX-G_4nRHjjgWbmo,91
-numba_cuda-0.9.0.dist-info/top_level.txt,sha256=C50SsH-8tXDmt7I0Y3nlJYhS5s6pqWflCPdobe9vx2M,11
-numba_cuda-0.9.0.dist-info/RECORD,,
+numba_cuda-0.10.1.dist-info/licenses/LICENSE,sha256=eHeYE-XjASmwbxfsP5AImgfzRwZurZGqH1f6OFwJ4io,1326
+numba_cuda-0.10.1.dist-info/METADATA,sha256=nP_9oLjsU48Y-dOmumPuN2JsiapA9t5ViCU_paTk7Uw,1859
+numba_cuda-0.10.1.dist-info/WHEEL,sha256=0CuiUZ_p9E4cD6NyLD6UG80LBXYyiSYZOKDm5lp32xk,91
+numba_cuda-0.10.1.dist-info/top_level.txt,sha256=C50SsH-8tXDmt7I0Y3nlJYhS5s6pqWflCPdobe9vx2M,11
+numba_cuda-0.10.1.dist-info/RECORD,,

{numba_cuda-0.9.0.dist-info → numba_cuda-0.10.1.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (79.0.0)
+Generator: setuptools (80.3.1)
 Root-Is-Purelib: true
 Tag: py3-none-any