Merge branch 'lwawrzyniak/release-1.2/v1.2.2' into 'release-1.2'

Release v1.2.2

See merge request omniverse/warp!580

.gitlab-ci.yml

@@ -220,6 +220,9 @@ windows-x86_64 test:
     - .\_venv\Scripts\Activate.ps1
     - python -m pip install --upgrade pip
     - python -m pip install --upgrade usd-core coverage[toml]
+    # Temporary HACK: use NumPy < 2.0 on Windows due to issues with Torch wheels that are not compatible
+    # https://github.com/pytorch/pytorch/issues/128860
+    - python -m pip install "numpy<2"
     - python -m pip install --upgrade torch --extra-index-url https://download.pytorch.org/whl/cu121
     - python -m pip install -e .
     - Write-Output "$([char]27)[0Ksection_end:$(GetTime):install_dependencies$([char]13)$([char]27)[0K"
@@ -285,6 +288,9 @@ windows-x86_64 test mgpu:
     - .\_venv\Scripts\Activate.ps1
     - python -m pip install --upgrade pip
     - python -m pip install --upgrade usd-core
+    # Temporary HACK: use NumPy < 2.0 on Windows due to issues with Torch wheels that are not compatible
+    # https://github.com/pytorch/pytorch/issues/128860
+    - python -m pip install "numpy<2"
     - python -m pip install --upgrade torch --extra-index-url https://download.pytorch.org/whl/cu121
     - python -m pip install -e .
     - Write-Output "$([char]27)[0Ksection_end:$(GetTime):install_dependencies$([char]13)$([char]27)[0K"

CHANGELOG.md

@@ -1,5 +1,9 @@
 # CHANGELOG
 
+## [1.2.2] - 2024-06-19
+
+- Support for NumPy >= 2.0
+
 ## [1.2.1] - 2024-06-14
 
 - Fix generic function caching

VERSION.md

@@ -1 +1 @@
-1.2.1
+1.2.2

exts/omni.warp.core/config/extension.toml

@@ -1,6 +1,6 @@
 [package]
 # Semantic Versioning is used: https://semver.org/
-version = "1.2.1"
+version = "1.2.2"
 authors = ["NVIDIA"]
 title = "Warp Core"
 description="The core Warp Python module"

exts/omni.warp.core/docs/CHANGELOG.md

@@ -1,5 +1,9 @@
 # CHANGELOG
 
+## [1.2.2] - 2024-06-19
+
+- Support for NumPy >= 2.0
+
 ## [1.2.1] - 2024-06-14
 
 - Fix generic function caching

exts/omni.warp/config/extension.toml

@@ -1,6 +1,6 @@
 [package]
 # Semantic Versioning is used: https://semver.org/
-version = "1.2.1"
+version = "1.2.2"
 authors = ["NVIDIA"]
 title = "Warp"
 description="Warp OmniGraph Nodes and Sample Scenes"
@@ -35,7 +35,7 @@ exclude = ["Ogn*Database.py", "*/ogn*"]
 "omni.timeline" = {}
 "omni.ui" = {optional = true}
 "omni.usd" = {}
-"omni.warp.core" = {version = "1.2.1", exact = true}
+"omni.warp.core" = {version = "1.2.2", exact = true}
 
 [[python.module]]
 name = "omni.warp._extension"

exts/omni.warp/docs/CHANGELOG.md

@@ -1,5 +1,9 @@
 # CHANGELOG
 
+## [1.2.2] - 2024-06-19
+
+- Support for NumPy >= 2.0
+
 ## [1.2.1] - 2024-06-14
 
 - Fix generic function caching

warp/config.py

@@ -7,7 +7,7 @@
 
 from typing import Optional
 
-version: str = "1.2.1"
+version: str = "1.2.2"
 
 verify_fp: bool = False  # verify inputs and outputs are finite after each launch
 verify_cuda: bool = False  # if true will check CUDA errors after each kernel launch / memory operation

warp/context.py

@@ -4045,7 +4045,7 @@ def full(
             # a sequence, assume it's a vector or matrix value
             try:
                 # try to convert to a numpy array first
-                na = np.array(value, copy=False)
+                na = np.asarray(value)
             except Exception as e:
                 raise ValueError(f"Failed to interpret the value as a vector or matrix: {e}") from e
 

warp/tests/test_vec_scalar_ops.py

@@ -687,10 +687,22 @@ def test_equality(test, device, dtype, register_kernels=False):
     vec4 = wp.types.vector(length=4, dtype=wptype)
     vec5 = wp.types.vector(length=5, dtype=wptype)
 
-    def check_equality(
+    def check_unsigned_equality(
         v20: wp.array(dtype=vec2),
         v21: wp.array(dtype=vec2),
         v22: wp.array(dtype=vec2),
+        v30: wp.array(dtype=vec3),
+        v40: wp.array(dtype=vec4),
+        v50: wp.array(dtype=vec5),
+    ):
+        wp.expect_eq(v20[0], v20[0])
+        wp.expect_neq(v21[0], v20[0])
+        wp.expect_neq(v22[0], v20[0])
+        wp.expect_eq(v30[0], v30[0])
+        wp.expect_eq(v40[0], v40[0])
+        wp.expect_eq(v50[0], v50[0])
+
+    def check_signed_equality(
         v30: wp.array(dtype=vec3),
         v31: wp.array(dtype=vec3),
         v32: wp.array(dtype=vec3),
@@ -707,29 +719,21 @@ def check_equality(
         v54: wp.array(dtype=vec5),
         v55: wp.array(dtype=vec5),
     ):
-        wp.expect_eq(v20[0], v20[0])
-        wp.expect_neq(v21[0], v20[0])
-        wp.expect_neq(v22[0], v20[0])
-
-        wp.expect_eq(v30[0], v30[0])
         wp.expect_neq(v31[0], v30[0])
         wp.expect_neq(v32[0], v30[0])
         wp.expect_neq(v33[0], v30[0])
-
-        wp.expect_eq(v40[0], v40[0])
         wp.expect_neq(v41[0], v40[0])
         wp.expect_neq(v42[0], v40[0])
         wp.expect_neq(v43[0], v40[0])
         wp.expect_neq(v44[0], v40[0])
-
-        wp.expect_eq(v50[0], v50[0])
         wp.expect_neq(v51[0], v50[0])
         wp.expect_neq(v52[0], v50[0])
         wp.expect_neq(v53[0], v50[0])
         wp.expect_neq(v54[0], v50[0])
         wp.expect_neq(v55[0], v50[0])
 
-    kernel = getkernel(check_equality, suffix=dtype.__name__)
+    unsigned_kernel = getkernel(check_unsigned_equality, suffix=dtype.__name__)
+    signed_kernel = getkernel(check_signed_equality, suffix=dtype.__name__)
 
     if register_kernels:
         return
@@ -739,49 +743,64 @@ def check_equality(
     v22 = wp.array([3.0, 2.0], dtype=vec2, requires_grad=True, device=device)
 
     v30 = wp.array([1.0, 2.0, 3.0], dtype=vec3, requires_grad=True, device=device)
-    v31 = wp.array([-1.0, 2.0, 3.0], dtype=vec3, requires_grad=True, device=device)
-    v32 = wp.array([1.0, -2.0, 3.0], dtype=vec3, requires_grad=True, device=device)
-    v33 = wp.array([1.0, 2.0, -3.0], dtype=vec3, requires_grad=True, device=device)
-
     v40 = wp.array([1.0, 2.0, 3.0, 4.0], dtype=vec4, requires_grad=True, device=device)
-    v41 = wp.array([-1.0, 2.0, 3.0, 4.0], dtype=vec4, requires_grad=True, device=device)
-    v42 = wp.array([1.0, -2.0, 3.0, 4.0], dtype=vec4, requires_grad=True, device=device)
-    v43 = wp.array([1.0, 2.0, -3.0, 4.0], dtype=vec4, requires_grad=True, device=device)
-    v44 = wp.array([1.0, 2.0, 3.0, -4.0], dtype=vec4, requires_grad=True, device=device)
-
     v50 = wp.array([1.0, 2.0, 3.0, 4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
-    v51 = wp.array([-1.0, 2.0, 3.0, 4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
-    v52 = wp.array([1.0, -2.0, 3.0, 4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
-    v53 = wp.array([1.0, 2.0, -3.0, 4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
-    v54 = wp.array([1.0, 2.0, 3.0, -4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
-    v55 = wp.array([1.0, 2.0, 3.0, 4.0, -5.0], dtype=vec5, requires_grad=True, device=device)
+
     wp.launch(
-        kernel,
+        unsigned_kernel,
         dim=1,
         inputs=[
             v20,
             v21,
             v22,
             v30,
-            v31,
-            v32,
-            v33,
             v40,
-            v41,
-            v42,
-            v43,
-            v44,
             v50,
-            v51,
-            v52,
-            v53,
-            v54,
-            v55,
         ],
         outputs=[],
         device=device,
     )
 
+    if dtype not in np_unsigned_int_types:
+        v31 = wp.array([-1.0, 2.0, 3.0], dtype=vec3, requires_grad=True, device=device)
+        v32 = wp.array([1.0, -2.0, 3.0], dtype=vec3, requires_grad=True, device=device)
+        v33 = wp.array([1.0, 2.0, -3.0], dtype=vec3, requires_grad=True, device=device)
+
+        v41 = wp.array([-1.0, 2.0, 3.0, 4.0], dtype=vec4, requires_grad=True, device=device)
+        v42 = wp.array([1.0, -2.0, 3.0, 4.0], dtype=vec4, requires_grad=True, device=device)
+        v43 = wp.array([1.0, 2.0, -3.0, 4.0], dtype=vec4, requires_grad=True, device=device)
+        v44 = wp.array([1.0, 2.0, 3.0, -4.0], dtype=vec4, requires_grad=True, device=device)
+
+        v51 = wp.array([-1.0, 2.0, 3.0, 4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
+        v52 = wp.array([1.0, -2.0, 3.0, 4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
+        v53 = wp.array([1.0, 2.0, -3.0, 4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
+        v54 = wp.array([1.0, 2.0, 3.0, -4.0, 5.0], dtype=vec5, requires_grad=True, device=device)
+        v55 = wp.array([1.0, 2.0, 3.0, 4.0, -5.0], dtype=vec5, requires_grad=True, device=device)
+
+        wp.launch(
+            signed_kernel,
+            dim=1,
+            inputs=[
+                v30,
+                v31,
+                v32,
+                v33,
+                v40,
+                v41,
+                v42,
+                v43,
+                v44,
+                v50,
+                v51,
+                v52,
+                v53,
+                v54,
+                v55,
+            ],
+            outputs=[],
+            device=device,
+        )
+
 
 def test_scalar_multiplication(test, device, dtype, register_kernels=False):
     rng = np.random.default_rng(123)

warp/types.py

@@ -1685,7 +1685,7 @@ def _init_from_data(self, data, dtype, shape, device, copy, pinned):
         if dtype == Any:
             # infer dtype from data
             try:
-                arr = np.array(data, copy=False, ndmin=1)
+                arr = np.asarray(data)
             except Exception as e:
                 raise RuntimeError(f"Failed to convert input data to an array: {e}") from e
             dtype = np_dtype_to_warp_type.get(arr.dtype)
@@ -1724,7 +1724,7 @@ def _init_from_data(self, data, dtype, shape, device, copy, pinned):
                     f"Failed to convert input data to an array with Warp type {warp.context.type_str(dtype)}"
                 )
             try:
-                arr = np.array(data, dtype=npdtype, copy=False, ndmin=1)
+                arr = np.asarray(data, dtype=npdtype)
             except Exception as e:
                 raise RuntimeError(f"Failed to convert input data to an array with type {npdtype}: {e}") from e
 
@@ -2333,7 +2333,7 @@ def numpy(self):
                 a = self.to("cpu", requires_grad=False)
             # convert through __array_interface__
             # Note: this handles arrays of structs using `descr`, so the result will be a structured NumPy array
-            return np.array(a, copy=False)
+            return np.asarray(a)
         else:
             # return an empty numpy array with the correct dtype and shape
             if isinstance(self.dtype, warp.codegen.Struct):