Build uploaded using `kernels`.

.gitattributes

@@ -99,3 +99,4 @@ build/torch210-cxx11-cu130-aarch64-linux/_mra_cuda_8d73b81.abi3.so filter=lfs di
 build/torch29-cxx11-cu126-aarch64-linux/_mra_cuda_8d73b81.abi3.so filter=lfs diff=lfs merge=lfs -text
 build/torch29-cxx11-cu128-aarch64-linux/_mra_cuda_8d73b81.abi3.so filter=lfs diff=lfs merge=lfs -text
 build/torch29-cxx11-cu130-aarch64-linux/_mra_cuda_8d73b81.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch210-cu128-x86_64-windows/_mra_cuda_6ec000c.pyd filter=lfs diff=lfs merge=lfs -text

build/torch210-cu128-x86_64-windows/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/torch210-cu128-x86_64-windows/_mra_cuda_6ec000c.pyd

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:768bbfeb18791e29510399a249bd7aee37fef0349bf44d1ff443bbe006c3d9af
+size 795648

build/torch210-cu128-x86_64-windows/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _mra_cuda_6ec000c
+ops = torch.ops._mra_cuda_6ec000c
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_mra_cuda_6ec000c::{op_name}"

build/torch210-cu128-x86_64-windows/metadata.json

@@ -0,0 +1,20 @@
+{
+  "version": 1,
+  "python-depends": [],
+  "backend": {
+    "type": "cuda",
+    "archs": [
+      "10.0",
+      "10.1",
+      "12.0+PTX",
+      "7.0",
+      "7.2",
+      "7.5",
+      "8.0",
+      "8.6",
+      "8.7",
+      "8.9",
+      "9.0"
+    ]
+  }
+}

build/torch210-cu128-x86_64-windows/mra/__init__.py

@@ -0,0 +1,26 @@
+import ctypes
+import sys
+
+import importlib
+from pathlib import Path
+from types import ModuleType
+
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
+
+
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))