Docs: File headers

ashvardanian · Jan 16, 2025 · e1ac216 · e1ac216
1 parent 37c0581
commit e1ac216
Show file tree

Hide file tree

Showing 7 changed files with 41 additions and 8 deletions.
diff --git a/.vscode/settings.json b/.vscode/settings.json
@@ -99,6 +99,7 @@
         "blas",
         "CCCL",
         "constexpr",
+        "cublas",
         "CUDA",
         "Kahan",
         "openmp",

diff --git a/reduce_bench.cpp b/reduce_bench.cpp
@@ -1,3 +1,9 @@
+/**
+ *  @date 04/09/2019
+ *  @file reduce_bench.cpp
+ *  @brief Benchmarking parallel reductions
+ *  @author Ash Vardanian
+ */
 #include <cstdlib> // Accessing environment variables
 #include <new>     // `std::launder`
 
@@ -128,17 +134,18 @@ int main(int argc, char **argv) {
         ->UseRealTime();
 #endif // defined(__AVX512F__)
 
-// CUDA
+    // CUDA
 #if defined(__CUDACC__)
     if (cuda_device_count()) {
         bm::RegisterBenchmark("cub@cuda", &make<cuda_cub_t>)->MinTime(10)->UseRealTime();
         bm::RegisterBenchmark("warps@cuda", &make<cuda_warps_t>)->MinTime(10)->UseRealTime();
         bm::RegisterBenchmark("thrust@cuda", &make<cuda_thrust_t>)->MinTime(10)->UseRealTime();
-    } else
+    } else {
         fmt::print("No CUDA capable devices found!\n");
+    }
 #endif
 
-        // OpenCL
+    // OpenCL
 #if defined(__OPENCL__)
     for (auto tgt : ocl_targets) {
         for (auto kernel_name : opencl_t::kernels_k) {

diff --git a/reduce_cpu.hpp b/reduce_cpu.hpp
@@ -1,3 +1,9 @@
+/**
+ *  @date 04/09/2019
+ *  @file reduce_cpu.hpp
+ *  @brief Parallel reduction with SIMD and multicore acceleration
+ *  @author Ash Vardanian
+ */
 #pragma once
 #include <cstring>   // `std::memcpy`
 #include <execution> // `std::execution::par_unseq`

diff --git a/reduce_cublas.cuh b/reduce_cublas.cuh
@@ -1,3 +1,9 @@
+/**
+ *  @date 04/09/2019
+ *  @file reduce_cublas.cuh
+ *  @brief cuBLAS-based reductions
+ *  @author Ash Vardanian
+ */
 #pragma once
 #include <cublas_api.h>
 #include <cuda_runtime_api.h>

diff --git a/reduce_cuda.cuh b/reduce_cuda.cuh
@@ -1,3 +1,9 @@
+/**
+ *  @date 04/09/2019
+ *  @file reduce_cuda.cuh
+ *  @brief Pure CUDA, CUB, and Thrust-based reductions
+ *  @author Ash Vardanian
+ */
 #pragma once
 #include <cuda_runtime_api.h>
 #include <mma.h> // `wmma::`

diff --git a/reduce_opencl.cl b/reduce_opencl.cl
@@ -1,8 +1,9 @@
-// Project: SandboxGPUs.
-// Author: Ash Vardanian.
-// Created: 04/09/2019.
-// Copyright: Check "License" file.
-//
+/**
+ *  @date 04/09/2019
+ *  @file reduce_opencl.cl
+ *  @brief OpenCL kernels for reduction algorithms
+ *  @author Ash Vardanian
+ */
 
 /**
  *  Most of the algorithms here have following properties:

diff --git a/reduce_opencl.hpp b/reduce_opencl.hpp
@@ -1,3 +1,9 @@
+/**
+ *  @date 04/09/2019
+ *  @file reduce_opencl.hpp
+ *  @brief OpenCL host code for parallel reductions
+ *  @author Ash Vardanian
+ */
 #pragma once
 #include <fstream>
 #include <sstream>