diff options
| -rw-r--r-- | CMakeLists.txt | 7 | ||||
| -rw-r--r-- | src/tuning/routines/xgemm.cpp | 136 | ||||
| -rw-r--r-- | src/utilities/timing.hpp | 52 |
3 files changed, 195 insertions, 0 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt index d3b202c2..73b47637 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -377,6 +377,13 @@ if(TUNERS) target_include_directories(clblast_tuner_${KERNEL} PUBLIC ${CLTUNE_INCLUDE_DIRS}) install(TARGETS clblast_tuner_${KERNEL} DESTINATION bin) endforeach() + set(ROUTINE_TUNERS xgemm) + foreach(ROUTINE_TUNER ${ROUTINE_TUNERS}) + add_executable(clblast_tuner_routine_${ROUTINE_TUNER} ${TUNERS_COMMON} src/tuning/routines/${ROUTINE_TUNER}.cpp) + target_link_libraries(clblast_tuner_routine_${ROUTINE_TUNER} clblast ${CLTUNE_LIBRARIES} ${API_LIBRARIES}) + target_include_directories(clblast_tuner_routine_${ROUTINE_TUNER} PUBLIC ${CLTUNE_INCLUDE_DIRS}) + install(TARGETS clblast_tuner_routine_${ROUTINE_TUNER} DESTINATION bin) + endforeach() # Adds 'alltuners' target: runs all tuners for all precisions set(ALLTUNERS ) diff --git a/src/tuning/routines/xgemm.cpp b/src/tuning/routines/xgemm.cpp new file mode 100644 index 00000000..9590323a --- /dev/null +++ b/src/tuning/routines/xgemm.cpp @@ -0,0 +1,136 @@ + +// ================================================================================================= +// This file is part of the CLBlast project. The project is licensed under Apache Version 2.0. This +// project loosely follows the Google C++ styleguide and uses a tab-size of two spaces and a max- +// width of 100 characters per line. +// +// Author(s): +// Cedric Nugteren <www.cedricnugteren.nl> +// +// This file tunes the Xgemm routine at a high-level: choosing between the direct (single-kernel) +// and the in-direct (kernel plus pre/post-processing) methods. +// +// ================================================================================================= + +#include <exception> +#include <string> +#include <vector> +#include <assert.h> + +#include "utilities/utilities.hpp" +#include "utilities/timing.hpp" + +namespace clblast { +// ================================================================================================= + +template <typename T> +void RunGemmRoutine(const size_t value, const Queue& queue, const std::vector<Buffer<T>>& buffers) { + auto queue_plain = queue(); + auto event = cl_event{}; + auto status = Gemm(Layout::kRowMajor, Transpose::kNo, Transpose::kNo, + value, value, value, ConstantOne<T>(), + buffers[0](), 0, value, + buffers[1](), 0, value, ConstantOne<T>(), + buffers[2](), 0, value, + &queue_plain, &event); + if (status != StatusCode::kSuccess) { + throw RuntimeError("Gemm failed with status " + ToString(status)); + } + clWaitForEvents(1, &event); + clReleaseEvent(event); +} + +template <typename T> +void ForceSelectIndirectFrom(const size_t minimum_size, const Device &device) { + const auto override_status = OverrideParameters(device(), "KernelSelection", PrecisionValue<T>(), + {{"XGEMM_MIN_INDIRECT_SIZE", minimum_size}}); + if (override_status != StatusCode::kSuccess) { + throw RuntimeError("OverrideParameters failed with status " + ToString(override_status)); + } +} + +template <typename T> +void TuneXgemm(int argc, char* argv[]) { + auto command_line_args = RetrieveCommandLineArguments(argc, argv); + auto help = std::string{"* Options given/available:\n"}; + const auto platform_id = GetArgument(command_line_args, help, kArgPlatform, ConvertArgument(std::getenv("CLBLAST_PLATFORM"), size_t{0})); + const auto device_id = GetArgument(command_line_args, help, kArgDevice, ConvertArgument(std::getenv("CLBLAST_DEVICE"), size_t{0})); + const auto precision = GetArgument(command_line_args, help, kArgPrecision, Precision::kSingle); + const auto num_runs = GetArgument(command_line_args, help, kArgNumRuns, size_t{10}); + fprintf(stdout, "%s\n", help.c_str()); + + // Values for m, n, and k + const auto from = size_t{64}; + const auto to = size_t{1024}; + const auto step = size_t{64}; + + // OpenCL initialisation + const auto platform = Platform(platform_id); + const auto device = Device(platform, device_id); + if (!PrecisionSupported<T>(device)) { + printf("* Unsupported precision, skipping this tuning run\n\n"); + return; + } + const auto context = Context(device); + const auto queue = Queue(context, device); + + // Buffers + auto a_mat = Buffer<T>(context, to * to); + auto b_mat = Buffer<T>(context, to * to); + auto c_mat = Buffer<T>(context, to * to); + auto buffers = std::vector<Buffer<T>>{a_mat, b_mat, c_mat}; + + // In-direct version + printf("[----------] Testing the in-direct GEMM routine for m=n=k\n"); + ForceSelectIndirectFrom<T>(0, device); + const auto indirect = TimeRoutine(from, to, step, num_runs, queue, buffers, RunGemmRoutine<T>); + + // Direct version + printf("[----------] Testing the direct GEMM routine for m=n=k\n"); + ForceSelectIndirectFrom<T>(to * to * to + 1, device); + const auto direct = TimeRoutine(from, to, step, num_runs, queue, buffers, RunGemmRoutine<T>); + + // Results + printf("[----------] Collecting results\n"); + assert(indirect.size() == direct.size()); + for (auto i = size_t{0}; i < indirect.size(); ++i) { + assert(indirect[i].first == direct[i].first); + const auto value = indirect[i].first; + const auto gflops_indirect = (2 * value * value * value) / (indirect[i].second * 1.0e6); + const auto gflops_direct = (2 * value * value * value) / (direct[i].second * 1.0e6); + printf("[ -------> ] %7zu %8.2lf %8.2lf\n", value, gflops_indirect, gflops_direct); + } + + // Outputs the results as JSON to disk, including some meta-data + const auto precision_string = std::to_string(static_cast<size_t>(precision)); + auto metadata = std::vector<std::pair<std::string,std::string>>{ + {"kernel_family", "gemm_routine"}, + {"precision", precision_string}, + }; + PrintTimingsToFileAsJSON("clblast_routine_gemm_" + precision_string + ".json", + device, platform, metadata); + +} + +// ================================================================================================= +} // namespace clblast + +// Shortcuts to the clblast namespace +using half = clblast::half; +using float2 = clblast::float2; +using double2 = clblast::double2; + +// Main function (not within the clblast namespace) +int main(int argc, char *argv[]) { + const auto command_line_args = clblast::RetrieveCommandLineArguments(argc, argv); + switch(clblast::GetPrecision(command_line_args)) { + case clblast::Precision::kHalf: clblast::TuneXgemm<half>(argc, argv); break; + case clblast::Precision::kSingle: clblast::TuneXgemm<float>(argc, argv); break; + case clblast::Precision::kDouble: clblast::TuneXgemm<double>(argc, argv); break; + case clblast::Precision::kComplexSingle: clblast::TuneXgemm<float2>(argc, argv); break; + case clblast::Precision::kComplexDouble: clblast::TuneXgemm<double2>(argc, argv); break; + } + return 0; +} + +// ================================================================================================= diff --git a/src/utilities/timing.hpp b/src/utilities/timing.hpp index 3d66de2a..4622aa99 100644 --- a/src/utilities/timing.hpp +++ b/src/utilities/timing.hpp @@ -14,14 +14,20 @@ #ifndef CLBLAST_TIMING_H_ #define CLBLAST_TIMING_H_ +#include <cstdio> +#include <utility> #include <vector> +#include <algorithm> #include <chrono> +#include "utilities/utilities.hpp" + namespace clblast { // ================================================================================================= template <typename F> double TimeFunction(const size_t num_runs, F const &function) { + function(); // warm-up auto timings = std::vector<double>(num_runs); for (auto &timing: timings) { const auto start_time = std::chrono::steady_clock::now(); @@ -33,6 +39,52 @@ double TimeFunction(const size_t num_runs, F const &function) { } // ================================================================================================= + +using Timing = std::pair<size_t, double>; + +template <typename T, typename F> +std::vector<Timing> TimeRoutine(const size_t from, const size_t to, const size_t step, + const size_t num_runs, const Queue& queue, + const std::vector<Buffer<T>>& buffers, F const &routine) { + auto timings = std::vector<Timing>(); + for (auto value = from; value < to; value += step) { + printf("[ RUN ] Running with value %zu\n", value); + try { + const auto FunctionToTune = [&]() { routine(value, queue, buffers); }; + const auto time_ms = TimeFunction(num_runs, FunctionToTune); + printf("[ OK ] Took %.2lf ms\n", time_ms); + timings.push_back({value, time_ms}); + } + catch (...) { + printf("[ ERROR ] Exception caught\n"); + timings.push_back({value, -1.0}); // invalid + } + } + return timings; +} + +// ================================================================================================= + +void PrintTimingsToFileAsJSON(const std::string &filename, + const Device& device, const Platform& platform, + const std::vector<std::pair<std::string,std::string>> &descriptions) { + auto file = fopen(filename.c_str(), "w"); + fprintf(file, "{\n"); + for (auto &description: descriptions) { + fprintf(file, " \"%s\": \"%s\",\n", description.first.c_str(), description.second.c_str()); + } + fprintf(file, " \"platform_version\": \"%s\",\n", platform.Version().c_str()); + fprintf(file, " \"device_name\": \"%s\",\n", GetDeviceName(device).c_str()); + fprintf(file, " \"device_vendor\": \"%s\",\n", platform.Vendor().c_str()); + fprintf(file, " \"device_type\": \"%s\",\n", device.Type().c_str()); + fprintf(file, " \"device_architecture\": \"%s\",\n", GetDeviceArchitecture(device).c_str()); + fprintf(file, " \"device_core_clock\": \"%zu\",\n", device.CoreClock()); + fprintf(file, " \"device_compute_units\": \"%zu\",\n", device.ComputeUnits()); + fprintf(file, "}\n"); + fclose(file); +} + +// ================================================================================================= } // namespace clblast // CLBLAST_TIMING_H_ |