measurement files added

ECRv2/.vscode/c_cpp_properties.json

@@ -4,7 +4,9 @@
             "name": "Linux",
             "includePath": [
                 "${workspaceFolder}/**",
-                "/usr/local/cuda-12.1/targets/x86_64-linux/include"
+                "/usr/local/cuda-12.1/targets/x86_64-linux/include/**",
+                "/home/gokay/llvm-project/llvm/include/**",
+                "/usr/include/llvm-16"
             ],
             "defines": [],
             "compilerPath": "/home/gokay/llvm-project/build/bin/clang",

ECRv2/README.md

@@ -1,6 +1,8 @@
 
 To use LLVM library together with CUDA CMake can be used:
-## Requirements 
+
+## Requirements
+
 - ninja
 - cmake
 - clang and lld
@@ -23,24 +25,27 @@ cmake -G Ninja -S .. -B . \
 
 ninja
 ```
-# singleECR 
+
+# singleECR
+
 singleECR driver allows us to run single convolution operations per feature and kernel. When the code is built its binary is under build directory. Time measurement is printed to console.
 
 ## Usage
+
 Feature and kernel paths can be provided via command line arguments. Optionally, output directory can be provided for the resulting convolution.
 
 ```shell
 ./singleECR --kernel <path-to-kernel>  --feature <path-to-feature>
 ```
+
 For example,
 
 ```shell
 ./singleECR --kernel ../../dataset/resnet/kernel/layer3.2.conv2.weight  --feature ../../dataset/resnet/feature/feature38 --output singleECR_result.txt
 ```
+
 ## Testing
 
-(TODO_GOKAY A python script that takes in the kernel, feature and optionally the output of singlECR)
 ```shell
-python3 verify_convolution.py --kernel ../../dataset/resnet/kernel/layer3.2.conv2.weight  --feature ../../dataset/resnet/feature/feature38 --test_output singleECR_result.txt
+python3 conv_test.py --kernel ../../dataset/resnet/kernel/layer3.2.conv2.weight  --feature ../../dataset/resnet/feature/feature38 --test_output singleECR_result.txt
 ```
-