change pkg name from "torch-ccl" to ""oneccl_binding_for_pytorch";

Author: Kim

.gitignore

@@ -15,15 +15,15 @@
 */**/**/**/*.pyc
 */**/**/**/**/*.pyc
 
-torch_ccl/include/
-torch_ccl/lib/
-torch_ccl/bin/
-torch_ccl/etc/
-torch_ccl/env/
-torch_ccl/examples/
-torch_ccl/licensing/
-torch_ccl/modulefiles/
-torch_ccl/version.py
+oneccl_bindings_for_pytorch/include/
+oneccl_bindings_for_pytorch/lib/
+oneccl_bindings_for_pytorch/bin/
+oneccl_bindings_for_pytorch/etc/
+oneccl_bindings_for_pytorch/env/
+oneccl_bindings_for_pytorch/examples/
+oneccl_bindings_for_pytorch/licensing/
+oneccl_bindings_for_pytorch/modulefiles/
+oneccl_bindings_for_pytorch/version.py
 
 ## General
 
@@ -107,7 +107,7 @@ gen
 
 # setup.py intermediates
 .eggs
-torch_ccl.egg-info
+oneccl_bindings_for_pytorch.egg-info
 oneccl_bind_pt.egg-info
 
 # Files generated by ctags

CMakeLists.txt

@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 3.13 FATAL_ERROR)
 set(CMAKE_CXX_STANDARD 17)
 
-project(torch_ccl C CXX)
+project(oneccl_bindings_for_pytorch C CXX)
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wformat")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror=cpp")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wformat-security")
@@ -32,7 +32,7 @@ set(DEPENDS_LIB)
 find_package(oneCCL REQUIRED)
 list(APPEND DEPENDS_LIB oneCCL torch)
 
-SET(LIB_NAME "torch_ccl")
+SET(LIB_NAME "oneccl_bindings_for_pytorch")
 
 add_subdirectory(./src)
 
@@ -55,9 +55,9 @@ function (print_configuration_summary)
     message(STATUS "  Compile flags         : ${IPEX_COMPILE_FLAGS}")
     message(STATUS "  Compile definitions   : ${CMAKE_COMPILE_DEFINITIONS}")
     message(STATUS "  Linker options        : ${CMAKE_SHARED_LINKER_FLAGS}")
-    get_target_property(LINK_LIBRARIES torch_ccl LINK_LIBRARIES)
+    get_target_property(LINK_LIBRARIES oneccl_bindings_for_pytorch LINK_LIBRARIES)
     message(STATUS "  Linker libraries        : ${LINK_LIBRARIES}")
-    get_target_property(LINK_DIRECTORS torch_ccl LINK_DIRECTORIES)
+    get_target_property(LINK_DIRECTORS oneccl_bindings_for_pytorch LINK_DIRECTORIES)
     message(STATUS "  Linker directors        : ${LINK_DIRECTORS}")
 
     message(STATUS "")

README.md

@@ -1,76 +1,80 @@
-# torch-ccl
+# Intel® oneCCL Bindings for PyTorch (formerly known as torch_ccl)
 
 This repository holds PyTorch bindings maintained by Intel for the Intel® oneAPI Collective Communications Library (oneCCL).
 
-
-# Introduction
+## Introduction
 
 [PyTorch](https://github.com/pytorch/pytorch) is an open-source machine learning framework.
 
-[Intel® oneCCL](https://github.com/oneapi-src/oneCCL) (collective commnications library) is a library for efficient distributed deep learning training implementing such collectives like allreduce, allgather, alltoall. For more information on oneCCL, please refer to the [oneCCL documentation](https://spec.oneapi.com/versions/latest/elements/oneCCL/source/index.html) and [oneCCL specification](https://spec.oneapi.com/versions/latest/elements/oneCCL/source/index.html).
+[Intel® oneCCL](https://github.com/oneapi-src/oneCCL) (collective communications library) is a library for efficient distributed deep learning training implementing such collectives like `allreduce`, `allgather`, `alltoall`. For more information on oneCCL, please refer to the [oneCCL documentation](https://spec.oneapi.com/versions/latest/elements/oneCCL/source/index.html) and [oneCCL specification](https://spec.oneapi.com/versions/latest/elements/oneCCL/source/index.html).
 
-`torch-ccl` module implements PyTorch C10D ProcessGroup API and can be dynamically loaded as external ProcessGroup and only works on Linux platform now.
+`oneccl_bindings_for_pytorch` module implements PyTorch C10D ProcessGroup API and can be dynamically loaded as external ProcessGroup and only works on Linux platform now.
 
-# Pytorch API Align
-We recommend Anaconda as Python package management system. The following is the corresponding branchs (tags) of torch-ccl and supported Pytorch.
+## Pytorch API Align
 
-   | ``torch`` | ``torch-ccl`` |  
-   | :-----:| :---: |  
-   |  ``master`` |  ``master``  |
-   | [v1.11.0](https://github.com/pytorch/pytorch/tree/v1.11.0) |  [ccl_torch1.11](https://github.com/intel/torch-ccl/tree/ccl_torch1.11)   |
-   | [v1.10.0](https://github.com/pytorch/pytorch/tree/v1.10.0) |  [ccl_torch1.10](https://github.com/intel/torch-ccl/tree/ccl_torch1.10)   |
-   | [v1.9.0](https://github.com/pytorch/pytorch/tree/v1.9.0) |  [ccl_torch1.9](https://github.com/intel/torch-ccl/tree/ccl_torch1.9)   |
-   | [v1.8.1](https://github.com/pytorch/pytorch/tree/v1.8.1) |  [ccl_torch1.8](https://github.com/intel/torch-ccl/tree/ccl_torch1.8)   | 
-   | [v1.7.1](https://github.com/pytorch/pytorch/tree/v1.7.1) |  [ccl_torch1.7](https://github.com/intel/torch-ccl/tree/ccl_torch1.7)   | 
-   | [v1.6.0](https://github.com/pytorch/pytorch/tree/v1.6.0) |  [ccl_torch1.6](https://github.com/intel/torch-ccl/tree/ccl_torch1.6)   | 
-   | [v1.5-rc3](https://github.com/pytorch/pytorch/tree/v1.5.0-rc3) |   [beta09](https://github.com/intel/torch-ccl/tree/beta09)   |
+We recommend Anaconda as Python package management system. The following is the corresponding branches (tags) of `oneccl_bindings_for_pytorch` and supported Pytorch.
 
-The usage details can be found in the README of corresponding branch. The following part is about the usage of v1.9 tag. if you want to use other version of torch-ccl please checkout to that branch(tag). For pytorch-1.5.0-rc3, the [#PR28068](https://github.com/pytorch/pytorch/pull/28068) and [#PR32361](https://github.com/pytorch/pytorch/pull/32361) are need to dynamicall register external ProcessGroup and enable ``alltoall`` collective communication primitive. The patch file about these two PRs is in ``patches`` directory and you can use it directly. 
+   | `torch`                                                         | `oneccl_bindings_for_pytorch`                                             |
+   | :-------------------------------------------------------------: | :-----------------------------------------------------------------------: |
+   | `master`                                                        |  `master`                                                                 |
+   | [v1.12.0](https://github.com/pytorch/pytorch/tree/v1.12.0)      |  [ccl_torch1.12](https://github.com/intel/torch-ccl/tree/ccl_torch1.12)   |
+   | [v1.11.0](https://github.com/pytorch/pytorch/tree/v1.11.0)      |  [ccl_torch1.11](https://github.com/intel/torch-ccl/tree/ccl_torch1.11)   |
+   | [v1.10.0](https://github.com/pytorch/pytorch/tree/v1.10.0)      |  [ccl_torch1.10](https://github.com/intel/torch-ccl/tree/ccl_torch1.10)   |
+   | [v1.9.0](https://github.com/pytorch/pytorch/tree/v1.9.0)        |  [ccl_torch1.9](https://github.com/intel/torch-ccl/tree/ccl_torch1.9)     |
+   | [v1.8.1](https://github.com/pytorch/pytorch/tree/v1.8.1)        |  [ccl_torch1.8](https://github.com/intel/torch-ccl/tree/ccl_torch1.8)     |
+   | [v1.7.1](https://github.com/pytorch/pytorch/tree/v1.7.1)        |  [ccl_torch1.7](https://github.com/intel/torch-ccl/tree/ccl_torch1.7)     |
+   | [v1.6.0](https://github.com/pytorch/pytorch/tree/v1.6.0)        |  [ccl_torch1.6](https://github.com/intel/torch-ccl/tree/ccl_torch1.6)     |
+   | [v1.5-rc3](https://github.com/pytorch/pytorch/tree/v1.5.0-rc3)  |   [beta09](https://github.com/intel/torch-ccl/tree/beta09)                |
 
-# Requirements
+The usage details can be found in the README of corresponding branch. The following part is about the usage of v1.9 tag. if you want to use other version of torch-ccl please checkout to that branch(tag). For pytorch-1.5.0-rc3, the [#PR28068](https://github.com/pytorch/pytorch/pull/28068) and [#PR32361](https://github.com/pytorch/pytorch/pull/32361) are need to dynamicall register external ProcessGroup and enable `alltoall` collective communication primitive. The patch file about these two PRs is in `patches` directory and you can use it directly.
 
-Python 3.6 or later and a C++17 compiler
+## Requirements
 
-PyTorch v1.11.0
+- Python 3.6 or later and a C++17 compiler
 
-# Installation
-## Source install
+- PyTorch v1.12.0
 
-To install `torch-ccl`:
+## Installation
 
-1. clone the `torch-ccl`.
+### Install from Source
 
-```bash
-   git clone https://github.com/intel/torch-ccl.git && cd torch-ccl 
-   git submodule sync 
+1. clone the `oneccl_bindings_for_pytorch`.
+
+   ```bash
+   git clone https://github.com/intel/torch-ccl.git && cd torch-ccl
+   git submodule sync
    git submodule update --init --recursive
-```
-2. Install torch-ccl
+   ```
 
-```bash
+2. Install `oneccl_bindings_for_pytorch`
+
+   ```bash
    python setup.py install
-```
-## Wheel install 
+   ```
+
+### Install PreBuilt Wheel
+
 Wheel files are avaiable for the following Python versions.
 
 | Extension Version | Python 3.6 | Python 3.7 | Python 3.8 | Python 3.9 | Python 3.10 |
-| :--: | :--: | :--: | :--: | :--: | :--: |
-| 1.11.0 | | ✔️ | ✔️ | ✔️ |✔️ |
-| 1.10.0 | ✔️ | ✔️ | ✔️ |✔️ |  |
+| :---------------: | :--------: | :--------: | :--------: | :--------: | :---------: |
+| 1.12.0            |            | √          | √          | √          | √           |
+| 1.11.0            |            | √          | √          | √          | √           |
+| 1.10.0            | √          | √          | √          | √          |             |
 
-```python
-python -m pip install oneccl_bind_pt==1.11.0 -f https://software.intel.com/ipex-whl-stable
+```bash
+python -m pip install  oneccl_bindings_for_pytorch==1.12.0 -f https://software.intel.com/ipex-whl-stable
 ```
 
-# Usage
+## Usage
 
 example.py
 
 ```python
 
 import torch.nn.parallel
 import torch.distributed as dist
-import torch_ccl
+import oneccl_bindings_for_pytorch
 
 ...
 
@@ -91,20 +95,21 @@ model = torch.nn.parallel.DistributedDataParallel(model, ...)
 
 ...
 ```
-(torch_ccl is installed along with the MPI toolset.)
-```
 
-$ source <torch_ccl_path>/env/setvars.sh
+(oneccl_bindings_for_pytorch is installed along with the MPI tool set.)
 
-eg:
-  $ torch_ccl_path=$(python -c "import torch; import torch_ccl; import os;  print(os.path.abspath(os.path.dirname(torch_ccl.__file__)))")
-  $ source $torch_ccl_path/env/setvars.sh
+```bash
 
-$ mpirun -n <N> -ppn <PPN> -f <hostfile> python example.py
-```
+source <oneccl_bindings_for_pytorch_path>/env/setvars.sh
+
+# eg:
+#   $ oneccl_bindings_for_pytorch_path=$(python -c "from oneccl_bindings_for_pytorch import cwd; print(cwd)")
+#   $ source $oneccl_bindings_for_pytorch_path/env/setvars.sh
 
+mpirun -n <N> -ppn <PPN> -f <hostfile> python example.py
+```
 
-# Performance Debugging
+## Performance Debugging
 
 For debugging performance of communication primitives PyTorch's [Autograd profiler](https://pytorch.org/docs/stable/autograd.html#profiler)
 can be used to inspect time spent inside oneCCL calls.
@@ -117,7 +122,7 @@ profiling.py
 
 import torch.nn.parallel
 import torch.distributed as dist
-import torch_ccl
+import oneccl_bindings_for_pytorch
 import os
 
 os.environ['MASTER_ADDR'] = '127.0.0.1'
@@ -142,36 +147,36 @@ print(prof.key_averages(group_by_input_shape=True).table(sort_by="self_cpu_time_
 
 ```
 
-```
-$ mpirun -n 2 -l python profiling.py
+```bash
+mpirun -n 2 -l python profiling.py
 ```
 
-```
+```bash
 [0] my rank = 0  my size = 2
-[0] -----------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------  
-[0]                                Name    Self CPU %      Self CPU   CPU total %     CPU total  CPU time avg    # of Calls          Input Shapes  
-[0] -----------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------  
-[0]                torch_ccl::allreduce        91.41%     297.900ms        91.41%     297.900ms      29.790ms            10              [[2, 2]]  
-[0]     torch_ccl::wait::cpu::allreduce         8.24%      26.845ms         8.24%      26.845ms       2.684ms            10      [[2, 2], [2, 2]]  
-[0]     torch_ccl::wait::cpu::allreduce         0.30%     973.651us         0.30%     973.651us      97.365us            10      [[4, 4], [4, 4]]  
-[0]                torch_ccl::allreduce         0.06%     190.254us         0.06%     190.254us      19.025us            10              [[4, 4]]  
-[0] -----------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------  
+[0] -----------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------
+[0]                                                  Name    Self CPU %      Self CPU   CPU total %     CPU total  CPU time avg    # of Calls          Input Shapes
+[0] -----------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------
+[0]                oneccl_bindings_for_pytorch::allreduce        91.41%     297.900ms        91.41%     297.900ms      29.790ms            10              [[2, 2]]
+[0]     oneccl_bindings_for_pytorch::wait::cpu::allreduce         8.24%      26.845ms         8.24%      26.845ms       2.684ms            10      [[2, 2], [2, 2]]
+[0]     oneccl_bindings_for_pytorch::wait::cpu::allreduce         0.30%     973.651us         0.30%     973.651us      97.365us            10      [[4, 4], [4, 4]]
+[0]                oneccl_bindings_for_pytorch::allreduce         0.06%     190.254us         0.06%     190.254us      19.025us            10              [[4, 4]]
+[0] -----------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------
 [0] Self CPU time total: 325.909ms
-[0] 
+[0]
 [1] my rank = 1  my size = 2
-[1] -----------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------  
-[1]                                Name    Self CPU %      Self CPU   CPU total %     CPU total  CPU time avg    # of Calls          Input Shapes  
-[1] -----------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------  
-[1]                torch_ccl::allreduce        96.03%     318.551ms        96.03%     318.551ms      31.855ms            10              [[2, 2]]  
-[1]     torch_ccl::wait::cpu::allreduce         3.62%      12.019ms         3.62%      12.019ms       1.202ms            10      [[2, 2], [2, 2]]  
-[1]                torch_ccl::allreduce         0.33%       1.082ms         0.33%       1.082ms     108.157us            10              [[4, 4]]  
-[1]     torch_ccl::wait::cpu::allreduce         0.02%      56.505us         0.02%      56.505us       5.651us            10      [[4, 4], [4, 4]]  
-[1] -----------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------  
+[1] -----------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------
+[1]                                                  Name    Self CPU %      Self CPU   CPU total %     CPU total  CPU time avg    # of Calls          Input Shapes
+[1] -----------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------
+[1]                oneccl_bindings_for_pytorch::allreduce        96.03%     318.551ms        96.03%     318.551ms      31.855ms            10              [[2, 2]]
+[1]     oneccl_bindings_for_pytorch::wait::cpu::allreduce         3.62%      12.019ms         3.62%      12.019ms       1.202ms            10      [[2, 2], [2, 2]]
+[1]                oneccl_bindings_for_pytorch::allreduce         0.33%       1.082ms         0.33%       1.082ms     108.157us            10              [[4, 4]]
+[1]     oneccl_bindings_for_pytorch::wait::cpu::allreduce         0.02%      56.505us         0.02%      56.505us       5.651us            10      [[4, 4], [4, 4]]
+[1] -----------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  --------------------
 [1] Self CPU time total: 331.708ms
-[1] 
+[1]
 
 ```
 
+## License
 
-# License
 [BSD License](https://github.com/intel/torch-ccl/blob/master/LICENSE)

demo/demo.py

@@ -8,7 +8,7 @@
 except:
    print("cant't import ipex")
 
-import torch_ccl
+import oneccl_bindings_for_pytorch
 
 
 class Model(nn.Module):

torch_ccl/__init__.py renamed to oneccl_bindings_for_pytorch/__init__.py

@@ -8,7 +8,7 @@
 FI_PROVIDER_PATH = os.path.join(cwd, "lib/prov")
 os.environ['FI_PROVIDER_PATH'] = FI_PROVIDER_PATH
 if not os.path.exists(os.path.join(cwd, "version.py")):
-    raise RuntimeError("torch_ccl is not installed!")
+    raise RuntimeError("oneccl_bindings_for_pytorch is not installed!")
 
 from .version import __version__, git_version
 from . import _C as ccl_lib
@@ -18,7 +18,7 @@
         try:
             # load the CCL/XPU library
             import ctypes
-            my_c_library = ctypes.cdll.LoadLibrary(os.path.join(cwd, "lib/libtorch_ccl_xpu.so"))
+            my_c_library = ctypes.cdll.LoadLibrary(os.path.join(cwd, "lib/liboneccl_bindings_for_pytorch_xpu.so"))
         except OSError:
             print("Warning: Cannot load xpu CCL. CCL doesn't work for XPU device")