Enables testing for pull requests

.github/workflows/build-and-test.yml

@@ -0,0 +1,40 @@
+name: Build And Test
+
+on:
+    pull_request:
+      branches:
+        - main
+      types: [synchronize, opened, reopened, ready_for_review]
+
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    container:
+      image: nvcr.io/nvidia/tritonserver:24.10-py3
+      volumes:
+        - ${{ github.workspace }}:/core
+
+    steps:
+    - uses: actions/checkout@v3
+
+    - name: Install dependencies
+      run: |
+        apt update
+        apt install -y --no-install-recommends clang-format-15 cmake libb64-dev rapidjson-dev libre2-dev
+        wget -O /tmp/boost.tar.gz https://archives.boost.io/release/1.80.0/source/boost_1_80_0.tar.gz && (cd /tmp && tar xzf boost.tar.gz) && mv /tmp/boost_1_80_0/boost /usr/include/boost
+        pip install build pytest
+
+    - name: Build
+      run: |
+        mkdir -p /core/build
+        cd /core/build
+        cmake -DCMAKE_INSTALL_PREFIX:PATH=`pwd`/install -DTRITON_CORE_HEADERS_ONLY=OFF ..
+        export TRITON_PYBIND="_c/triton_bindings.cpython-310-x86_64-linux-gnu.so"
+        make -j8
+
+    - name: Run tests with pytest
+      run: |
+        cd /core
+        python3 -m pip install --force-reinstall build/python/generic/wheel/dist/tritonserver-*.whl
+        pytest python/test -v

python/test/test_api.py

@@ -357,11 +357,7 @@ def test_stop(self):
                     {
                         "backend": "python",
                         "parameters": {"decoupled": {"string_value": "False"}},
-                        # Keep instance count low for fast startup/cleanup.
-                        # Alternatively can use KIND_CPU here, but keeping gpus/count explicit.
-                        "instance_group": [
-                            {"kind": "KIND_GPU", "gpus": [0], "count": 1}
-                        ],
+                        "instance_group": [{"kind": "KIND_CPU"}],
                     }
                 )
             },

src/test/input_byte_size_test.cc

@@ -378,7 +378,7 @@ TEST_F(InputByteSizeTest, InputByteSizeLarge)
       "setting request release callback");
 
   // Define input shape and data
-  size_t element_cnt = (1LL << 31) / sizeof(float);
+  int64_t element_cnt = (1LL << 31) / sizeof(float);
   std::vector<int64_t> shape{1, element_cnt};
   std::vector<float> input_data(element_cnt, 1);
   const auto input0_byte_size = sizeof(input_data[0]) * input_data.size();