1
1
How to use TorchInductor on Windows CPU
2
2
=======================================
3
3
4
- **Author **: `Xu, Han <https://github.com/xuhancn >`_, `Zhaoqiong Zheng <https://github.com/ZhaoqiongZ >`_
4
+ **Author **: `Zhaoqiong Zheng <https://github.com/ZhaoqiongZ >`_, `Xu, Han <https://github.com/xuhancn >`_
5
5
6
6
7
7
Introduction
@@ -29,7 +29,8 @@ During Installation, chosse `Desktop Development with C++` in the `Desktop & Mob
29
29
30
30
.. note ::
31
31
32
- You can also choose other compiler like Clang & Intel Compiler.
32
+ We recommend C++ compiler `Clang <https://github.com/llvm/llvm-project/releases >`_ and `Intel Compiler<https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit-download.html> `_.
33
+ Intel compiler setup guide(TODO: Intel will provide URL for this guide recently)
33
34
34
35
35
36
Install Miniforge
@@ -41,21 +42,26 @@ Set Up Environment
41
42
^^^^^^^^^^^^^^^^^^
42
43
43
44
1. Open a command line environment via cmd.exe.
45
+
44
46
2. Activate `MSVC ` via below command:
45
47
::
46
48
47
49
"C:/Program Files/Microsoft Visual Studio/2022/Community/VC/Auxiliary/Build/vcvars64.bat"
48
- 3. Activate `conda ` via below command:
50
+
51
+ 1. Activate `conda ` via below command:
49
52
::
50
53
51
54
"C:/ProgramData/miniforge3/Scripts/activate.bat"
52
- 4. Create and activate customer conda environment:
55
+
56
+ 1. Create and activate customer conda environment:
53
57
::
54
58
55
59
conda create -n inductor_cpu_windows python=3.10 -y
56
60
conda activate inductor_cpu_windows
57
- 5. Install `PyTorch 2.5 <https://pytorch.org/get-started/locally/ >`_ or later.
58
- 6. Try `torchinductor ` on Windows CPU:
61
+
62
+ 1. Install `PyTorch 2.5 <https://pytorch.org/get-started/locally/ >`_ or later.
63
+
64
+ 2. Try `torchinductor ` on Windows CPU:
59
65
::
60
66
61
67
import torch
@@ -66,6 +72,30 @@ Set Up Environment
66
72
opt_foo1 = torch.compile(foo)
67
73
print(opt_foo1(torch.randn(10, 10), torch.randn(10, 10)))
68
74
75
+ output of the above example.
76
+ ::
77
+
78
+ tensor([[-3.9074e-02, 1.3994e+00, 1.3894e+00, 3.2630e-01, 8.3060e-01,
79
+ 1.1833e+00, 1.4016e+00, 7.1905e-01, 9.0637e-01, -1.3648e+00],
80
+ [ 1.3728e+00, 7.2863e-01, 8.6888e-01, -6.5442e-01, 5.6790e-01,
81
+ 5.2025e-01, -1.2647e+00, 1.2684e+00, -1.2483e+00, -7.2845e-01],
82
+ [-6.7747e-01, 1.2028e+00, 1.1431e+00, 2.7196e-02, 5.5304e-01,
83
+ 6.1945e-01, 4.6654e-01, -3.7376e-01, 9.3644e-01, 1.3600e+00],
84
+ [-1.0157e-01, 7.7200e-02, 1.0146e+00, 8.8175e-02, -1.4057e+00,
85
+ 8.8119e-01, 6.2853e-01, 3.2773e-01, 8.5082e-01, 8.4615e-01],
86
+ [ 1.4140e+00, 1.2130e+00, -2.0762e-01, 3.3914e-01, 4.1122e-01,
87
+ 8.6895e-01, 5.8852e-01, 9.3310e-01, 1.4101e+00, 9.8318e-01],
88
+ [ 1.2355e+00, 7.9290e-02, 1.3707e+00, 1.3754e+00, 1.3768e+00,
89
+ 9.8970e-01, 1.1171e+00, -5.9944e-01, 1.2553e+00, 1.3394e+00],
90
+ [-1.3428e+00, 1.8400e-01, 1.1756e+00, -3.0654e-01, 9.7973e-01,
91
+ 1.4019e+00, 1.1886e+00, -1.9194e-01, 1.3632e+00, 1.1811e+00],
92
+ [-7.1615e-01, 4.6622e-01, 1.2089e+00, 9.2011e-01, 1.0659e+00,
93
+ 9.0892e-01, 1.1932e+00, 1.3888e+00, 1.3898e+00, 1.3218e+00],
94
+ [ 1.4139e+00, -1.4000e-01, 9.1192e-01, 3.0175e-01, -9.6432e-01,
95
+ -1.0498e+00, 1.4115e+00, -9.3212e-01, -9.0964e-01, 1.0127e+00],
96
+ [ 5.7244e-04, 1.2799e+00, 1.3595e+00, 1.0907e+00, 3.7191e-01,
97
+ 1.4062e+00, 1.3672e+00, 6.8502e-02, 8.5216e-01, 8.6046e-01]])
98
+
69
99
Conclusion
70
100
----------
71
101
0 commit comments