.gitignore

@@ -4,6 +4,7 @@ build-*/
 __pycache__/
 .pytest_cache/
 .mypy_cache/
+.vscode/
 .ruff_cache/
 *.pyc
 *.so
@@ -12,3 +13,5 @@ __pycache__/
 *.log
 /generated/
 /third_party/
+.git-credentials
+.vscode/ 

cmd.md

@@ -0,0 +1,33 @@
+
+```bash
+# git保存用户名密码
+# 本地仓库配置，不加 --global
+git config credential.helper store
+# 指定凭证文件存在项目目录（默认存在～/.git-credentials，改到本地）
+git config credential.helper "store --file ./.git-credentials"
+echo ".git-credentials" >> .gitignore
+# 查看本仓库git配置
+git config --list --local
+# 查看所有远程仓库
+git remote -v
+
+# 将本地 master 分支重命名为 main
+git branch -m master main
+# 拉取远程 main 分支，同步远端代码
+git pull origin main --allow-unrelated-histories
+# 绑定本地 main 与远程 main，推送
+git push -u origin main
+# 查看本地 + 远程所有分支
+git branch -a
+
+# 默认3.7.0版本不行
+rm -rf third_party/cutlass
+mkdir -p third_party
+cd third_party/cutlass
+git clone --depth 1 --branch v4.4.0 https://gitcode.com/NVIDIA/cutlass.git third_party/cutlass
+
+cd build-nvidia
+cmake .. -DCMAKE_CUDA_ARCHITECTURES=80 -DCUTLASS_PATH=../third_party/cutlass
+
+PYTHONPATH=python:. CAMP_BUILD_DIR=build-nvidia pytest tests/op_tests/test_copy.py -v
+```

tests/op_tests/.ipynb_checkpoints/test_copy-checkpoint.py

@@ -0,0 +1,40 @@
+from __future__ import annotations
+
+import pytest
+import torch
+
+from operator_runtime import copy, copy_
+from operator_runtime_testing import assert_close, require_cuda
+from tests.cases import copy as copy_cases
+
+
+@pytest.mark.parametrize("case", copy_cases.correctness_cases(), ids=lambda c: c["name"])
+def test_copy_correctness(case, backend):
+    require_cuda()
+    src = torch.randn(case["shape"], dtype=case["dtype"], device="cuda")
+    out = copy(src, backend=backend)
+    assert_close(out, src, atol=case["atol"], rtol=case["rtol"])
+
+
+@pytest.mark.parametrize("case", copy_cases.api_error_cases(), ids=lambda c: c["name"])
+def test_copy_api_contract(case, backend):
+    require_cuda()
+    if case["name"] == "shape_mismatch":
+        src = torch.randn(case["shape"], device="cuda", dtype=case["dtype"])
+        out = torch.empty(case["out_shape"], device="cuda", dtype=case["dtype"])
+        with pytest.raises(ValueError, match="matching shapes"):
+            copy_(out, src, backend=backend)
+        return
+    if case["name"] == "dtype_mismatch":
+        src = torch.randn(case["shape"], device="cuda", dtype=case["dtype"])
+        out = torch.empty(case["shape"], device="cuda", dtype=case["out_dtype"])
+        with pytest.raises(TypeError, match="matching dtypes"):
+            copy_(out, src, backend=backend)
+        return
+    if case["name"] == "non_contiguous":
+        src = torch.randn(case["shape"], device="cuda", dtype=case["dtype"]).t()
+        out = torch.empty(case["shape"], device="cuda", dtype=case["dtype"]).t()
+        with pytest.raises(ValueError, match="contiguous"):
+            copy_(out, src, backend=backend)
+        return
+    raise AssertionError(f"unhandled case: {case['name']}")