Adding Documentation for ROCm (facebookresearch#3856)

Summary:
Adding documentation for ROCm in README.md and INSTALL.md

Pull Request resolved: facebookresearch#3856

Reviewed By: ramilbakhshyiev

Differential Revision: D62774430

Pulled By: mnorris11

fbshipit-source-id: c5692be5bb6d5a6876ad168edf1855ed8ccfd26b

Author: ItsPitt

INSTALL.md

@@ -19,6 +19,8 @@ $ conda install -c pytorch -c nvidia faiss-gpu=1.8.0
 
 # GPU(+CPU) version with NVIDIA RAFT
 $ conda install -c pytorch -c nvidia -c rapidsai -c conda-forge faiss-gpu-raft=1.8.0
+
+# GPU(+CPU) version using AMD ROCm not yet available
 ```
 
 For faiss-gpu, the nvidia channel is required for CUDA, which is not
@@ -37,6 +39,8 @@ $ conda install -c pytorch/label/nightly -c nvidia faiss-gpu=1.8.0
 
 # GPU(+CPU) version with NVIDIA RAFT
 conda install -c pytorch -c nvidia -c rapidsai -c conda-forge faiss-gpu-raft=1.8.0 pytorch pytorch-cuda numpy
+
+# GPU(+CPU) version using AMD ROCm not yet available
 ```
 In the above commands, pytorch-cuda=11 or pytorch-cuda=12 would select a specific CUDA version, if it’s required.
 
@@ -63,6 +67,8 @@ $ conda install -c conda-forge faiss-cpu
 
 # GPU version
 $ conda install -c conda-forge faiss-gpu
+
+# AMD ROCm version not yet available
 ```
 
 You can tell which channel your conda packages come from by using `conda list`.
@@ -87,6 +93,8 @@ The optional requirements are:
 - for GPU indices:
   - nvcc,
   - the CUDA toolkit,
+- for AMD GPUs:
+  - AMD ROCm,
 - for the python bindings:
   - python 3,
   - numpy,
@@ -137,6 +145,10 @@ Several options can be passed to CMake, among which:
   - `-DCMAKE_CUDA_ARCHITECTURES="75;72"` for specifying which GPU architectures
   to build against (see [CUDA docs](https://developer.nvidia.com/cuda-gpus) to
   determine which architecture(s) you should pick),
+  - `-DFAISS_ENABLE_ROCM=ON` in order to enable building GPU indices for AMD GPUs.
+  The hipify script must be executed before using this option. 
+  Invoke `./faiss/gpu/hipify.sh` to execute. `-DFAISS_ENABLE_GPU` must be `ON`
+  when using this option. (possible values are `ON` and `OFF`),
 - python-related options:
   - `-DPython_EXECUTABLE=/path/to/python3.7` in order to build a python
   interface for a different python than the default one (see

README.md

@@ -16,7 +16,7 @@ The GPU implementation can accept input from either CPU or GPU memory. On a serv
 
 ## Installing
 
-Faiss comes with precompiled libraries for Anaconda in Python, see [faiss-cpu](https://anaconda.org/pytorch/faiss-cpu) and [faiss-gpu](https://anaconda.org/pytorch/faiss-gpu). The library is mostly implemented in C++, the only dependency is a [BLAS](https://en.wikipedia.org/wiki/Basic_Linear_Algebra_Subprograms) implementation. Optional GPU support is provided via CUDA, and the Python interface is also optional. It compiles with cmake. See [INSTALL.md](INSTALL.md) for details.
+Faiss comes with precompiled libraries for Anaconda in Python, see [faiss-cpu](https://anaconda.org/pytorch/faiss-cpu) and [faiss-gpu](https://anaconda.org/pytorch/faiss-gpu). The library is mostly implemented in C++, the only dependency is a [BLAS](https://en.wikipedia.org/wiki/Basic_Linear_Algebra_Subprograms) implementation. Optional GPU support is provided via CUDA or AMD ROCm, and the Python interface is also optional. It compiles with cmake. See [INSTALL.md](INSTALL.md) for details.
 
 ## How Faiss works