Update release versioning to 1.3 (pytorch#4408)

Summary:
X-link: facebookresearch/FBGEMM#1479

- Update release versioning to 1.3 for upcoming release

Pull Request resolved: pytorch#4408

Reviewed By: spcyppt

Differential Revision: D77394007

Pulled By: q10

fbshipit-source-id: 2faa1d4d3962516863b0f5094068a22c061abc42

Author: q10

.github/workflows/fbgemm_gpu_pip.yml

@@ -26,12 +26,12 @@ on:
   workflow_dispatch:
     inputs:
       pytorch_version:
-        description: PyTorch Version (e.g. '2.6.0', 'nightly', 'test')
+        description: PyTorch Version (e.g. '2.8.0', 'nightly', 'test')
         type: string
         required: true
         default: "nightly"
       fbgemm_gpu_channel_version:
-        description: FBGEMM-GPU Channel + Version (e.g. '1.2.0', 'nightly', 'test/1.2.0')
+        description: FBGEMM-GPU Channel + Version (e.g. '1.3.0', 'nightly', 'test/1.3.0')
         type: string
         required: true
         default: "nightly"

fbgemm_gpu/CMakeLists.txt

@@ -103,7 +103,7 @@ endif()
 # Declare CMake project
 project(
   fbgemm_gpu
-  VERSION 1.2.0
+  VERSION 1.3.0
   LANGUAGES ${project_languages})
 
 # AVX Flags Setup - must be set AFTER project declaration

fbgemm_gpu/docs/src/conf.py

@@ -28,10 +28,10 @@
 author = "FBGEMM Team"
 
 # The short X.Y version.
-version = "1.2"
+version = "1.3"
 
 # The full version, including alpha/beta/rc tags
-release = "1.2.0"
+release = "1.3.0"
 
 
 # -- Path setup --------------------------------------------------------------

fbgemm_gpu/docs/src/fbgemm_genai/development/BuildInstructions.rst

@@ -48,7 +48,7 @@ Clone the repo along with its submodules, and install ``requirements_genai.txt``
   # !! Run inside the Conda environment !!
 
   # Select a version tag
-  FBGEMM_VERSION=v1.2.0
+  FBGEMM_VERSION=v1.3.0
 
   # Clone the repo along with its submodules
   git clone --recursive -b ${FBGEMM_VERSION} https://github.com/pytorch/FBGEMM.git fbgemm_${FBGEMM_VERSION}

fbgemm_gpu/docs/src/fbgemm_gpu/development/BuildInstructions.rst

@@ -475,7 +475,7 @@ Clone the repo along with its submodules, and install ``requirements.txt``:
   # !! Run inside the Conda environment !!
 
   # Select a version tag
-  FBGEMM_VERSION=v1.2.0
+  FBGEMM_VERSION=v1.3.0
 
   # Clone the repo along with its submodules
   git clone --recursive -b ${FBGEMM_VERSION} https://github.com/pytorch/FBGEMM.git fbgemm_${FBGEMM_VERSION}

fbgemm_gpu/docs/src/general/Releases.rst

@@ -14,6 +14,9 @@ older than the one that the FBGEMM release corresponds to.
 | FBGEMM Release  | Corresponding    | Supported        | Supported      | Supported CUDA     | (Experimental) Supported  | (Experimental) Supported  |
 |                 | PyTorch Release  | Python Versions  | CUDA Versions  | Architectures      | ROCm Versions             | ROCm Architectures        |
 +=================+==================+==================+================+====================+===========================+===========================+
+| 1.3.0           | 2.8.x            | 3.9, 3.10, 3.11, | 12.6, 12.8,    | 7.0, 8.0, 9.0a,    | 6.3, 6.4                  | gfx908, gfx90a, gfx942    |
+|                 |                  | 3.12, 3.13       | 12.9           | 10.0a, 12.0a       |                           |                           |
++-----------------+------------------+------------------+----------------+--------------------+---------------------------+---------------------------+
 | 1.2.0           | 2.7.x            | 3.9, 3.10, 3.11, | 11.8, 12.6,    | 7.0, 8.0, 9.0,     | 6.2.4, 6.3                | gfx908, gfx90a, gfx942    |
 |                 |                  | 3.12, 3.13       | 12.8           | 9.0a, 10.0a, 12.0a |                           |                           |
 +-----------------+------------------+------------------+----------------+--------------------+---------------------------+---------------------------+

fbgemm_gpu/experimental/gen_ai/README.md

@@ -44,8 +44,8 @@ y = torch.ops.fbgemm.f8f8bf16_rowwise(
 
 ```bash
 # Full FBGEMM library
-pip install fbgemm-gpu==1.2.0
-pip install fbgemm-gpu==1.2.0 --index-url https://download.pytorch.org/whl/cu126
+pip install fbgemm-gpu==1.3.0
+pip install fbgemm-gpu==1.3.0 --index-url https://download.pytorch.org/whl/cu126
 
 # FBGEMM library with GenAI operator only
 pip install fbgemm-gpu-genai

fbgemm_gpu/test/release/stable_ops_v1.json

@@ -52,6 +52,23 @@
         "torch.ops.fbgemm.segment_sum_csr":"fbgemm::segment_sum_csr(SymInt batch_size, Tensor csr_seg, Tensor values) -> Tensor",
         "torch.ops.fbgemm.keyed_jagged_index_select_dim1":"fbgemm::keyed_jagged_index_select_dim1(Tensor values, Tensor lengths, Tensor offsets, Tensor indices, SymInt batch_size, Tensor? weights=None, SymInt? selected_lengths_sum=None) -> Tensor[]"
       }
+    },
+    {
+      "version":"1.3.0",
+      "api":{
+        "torch.ops.fbgemm.jagged_to_padded_dense":"fbgemm::jagged_to_padded_dense(Tensor values, Tensor[] offsets, SymInt[] max_lengths, float padding_value = 0) -> Tensor",
+        "torch.ops.fbgemm.merge_pooled_embeddings":"fbgemm::merge_pooled_embeddings(Tensor[] pooled_embeddings, SymInt uncat_dim_size, Device target_device, SymInt cat_dim=1) -> Tensor",
+        "torch.ops.fbgemm.permute_pooled_embs_auto_grad":"fbgemm::permute_pooled_embs_auto_grad(Tensor pooled_embs, Tensor offset_dim_list, Tensor permute_list, Tensor inv_offset_dim_list, Tensor inv_permute_list) -> Tensor",
+        "torch.ops.fbgemm.FloatOrHalfToFusedNBitRowwiseQuantizedSBHalf":"fbgemm::FloatOrHalfToFusedNBitRowwiseQuantizedSBHalf(Tensor input, int bit_rate) -> Tensor",
+        "torch.ops.fbgemm.permute_2D_sparse_data":"fbgemm::permute_2D_sparse_data(Tensor permute, Tensor lengths, Tensor values, Tensor? weights=None, SymInt? permuted_lengths_sum=None) -> (Tensor, Tensor, Tensor?)",
+        "torch.ops.fbgemm.permute_1D_sparse_data":"fbgemm::permute_1D_sparse_data(Tensor permute, Tensor lengths, Tensor values, Tensor? weights=None, SymInt? permuted_lengths_sum=None) -> (Tensor, Tensor, Tensor?)",
+        "torch.ops.fbgemm.expand_into_jagged_permute":"fbgemm::expand_into_jagged_permute(Tensor permute, Tensor input_offset, Tensor output_offset, SymInt output_size) -> Tensor",
+        "torch.ops.fbgemm.block_bucketize_sparse_features":"fbgemm::block_bucketize_sparse_features(Tensor lengths, Tensor indices, bool bucketize_pos, bool sequence, Tensor block_sizes, SymInt my_size, Tensor? weights=None, Tensor? batch_size_per_feature=None, SymInt max_B= -1, Tensor[]? block_bucketize_pos=None, bool keep_orig_idx=False, Tensor? total_num_blocks=None, Tensor? keep_orig_idx_per_feature=None) -> (Tensor, Tensor, Tensor?, Tensor?, Tensor?)",
+        "torch.ops.fbgemm.asynchronous_complete_cumsum":"fbgemm::asynchronous_complete_cumsum(Tensor t_in) -> Tensor",
+        "torch.ops.fbgemm.offsets_range":"fbgemm::offsets_range(Tensor offsets, SymInt range_size) -> Tensor",
+        "torch.ops.fbgemm.segment_sum_csr":"fbgemm::segment_sum_csr(SymInt batch_size, Tensor csr_seg, Tensor values) -> Tensor",
+        "torch.ops.fbgemm.keyed_jagged_index_select_dim1":"fbgemm::keyed_jagged_index_select_dim1(Tensor values, Tensor lengths, Tensor offsets, Tensor indices, SymInt batch_size, Tensor? weights=None, SymInt? selected_lengths_sum=None) -> Tensor[]"
+      }
     }
   ]
 }

fbgemm_gpu/test/release/stable_release_test.py

@@ -155,7 +155,7 @@ def test_backwards_compatibility(self) -> None:
         """
         Test the schema compatibility of the operators against previous versions of the API.
         """
-        for version in ["1.0.0", "1.1.0", "1.2.0"]:
+        for version in ["1.0.0", "1.1.0", "1.2.0", "1.3.0"]:
             try:
                 self._test_stable_schema(version)
             except Exception as e: