From ea2ce177dd444b5670e12d5b427359ce77b8cc89 Mon Sep 17 00:00:00 2001 From: Inki Dae Date: Thu, 9 Apr 2020 10:01:44 +0900 Subject: Introduce CL kernel cache feature This patch introduces CL kernel cache feature for NN runtimes such as ARMNN or other in-house versions. Originally, this feature came from utils/Utils.cpp and utils/Utils.h files but was used just for testing. Caching CL kernel binaries is mandatory and required for real product because inference performance at first run is one of very important factors. This patch exposes two API - save_program_cache_to_file and restore_program_cache_from_file so that NN runtimes can use these API to cache the compiled CL kernel binaries. Changelog v1. - Reordered including headers suggested by Manuel Bottini. - Updated description. Changelog v2. - Drop function descriptions. Changelog v3. - Fix build error with opencl = 0. As for this, it moves two functions, save_program_cache_to_file and restore_program_cache_from_file, to src/runtime/CL/Utils.cpp and arm_compute/runtime/CL/Utils.h, and set ARM_COMPUTE_CL flag only when opencl = 1. This change makes CL kernel cache feature to depend on OpenCL support. Changelog v4. - Fix coding_standard_checks error. Changelog v5. - Add a new cpp file to Android.bp to fix coding_standard_checks error. Changelog v6. - rebased on top of master, and dropped _new_built_programs_map object so it adds compiled CL kerne binaries per model to a given file. Changelog v7. - Correct descriptions. - Correct arm_compute/runtime/CL/Utils.h file description. - Correct src/runtime/CL/Utils.cpp file description. - Drop ARM_COMPUTE_CL ifdef ~ endif from examples/cl_cache.cpp - Do not add 'ARM_COMPUTE_CL' flag at SConscript. Change-Id: I8c640fb1809ca79191daf5b1ff72a50dd492c36d Signed-off-by: Inki Dae Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2914 Tested-by: Arm Jenkins Reviewed-by: Georgios Pinitas Comments-Addressed: Arm Jenkins --- utils/Utils.cpp | 82 --------------------------------------------------------- utils/Utils.h | 12 --------- 2 files changed, 94 deletions(-) (limited to 'utils') diff --git a/utils/Utils.cpp b/utils/Utils.cpp index 80b47d7675..70c2004d7a 100644 --- a/utils/Utils.cpp +++ b/utils/Utils.cpp @@ -274,87 +274,5 @@ uint64_t get_mem_free_from_meminfo() // Nothing found or an error during opening the file return 0; } - -/** This function loads prebuilt opencl kernels from a file - * - * @param[in] filename Name of the file to be used to load the kernels - */ -void restore_program_cache_from_file(const std::string &filename) -{ -#ifdef ARM_COMPUTE_CL - std::ifstream cache_file(filename, std::ios::binary); - if(cache_file.is_open()) - { - if(!CLScheduler::get().is_initialised()) - { - arm_compute::CLScheduler::get().default_init(); - } - - while(!cache_file.eof()) - { - size_t name_len = 0; - size_t binary_len = 0; - cache_file.read(reinterpret_cast(&name_len), sizeof(size_t)); - cache_file.read(reinterpret_cast(&binary_len), sizeof(size_t)); - if(name_len == 0 || binary_len == 0) - { - break; - } - std::vector tmp(name_len); - std::vector binary(binary_len); - std::string name; - cache_file.read(tmp.data(), name_len); - name.assign(tmp.data(), name_len); - tmp.resize(binary_len); - cache_file.read(reinterpret_cast(binary.data()), binary_len); - cl::Context context = arm_compute::CLScheduler::get().context(); - cl::Program::Binaries binaries{ binary }; - std::vector devices = context.getInfo(); - cl::Program program(context, devices, binaries); - program.build(); - CLKernelLibrary::get().add_built_program(name, program); - } - cache_file.close(); - } -#else /* ARM_COMPUTE_CL */ - ARM_COMPUTE_UNUSED(filename); -#endif /* ARM_COMPUTE_CL */ -} - -/** This function saves opencl kernels library to a file - * - * @param[in] filename Name of the file to be used to save the library - */ -void save_program_cache_to_file(const std::string &filename) -{ -#ifdef ARM_COMPUTE_CL - if(CLScheduler::get().is_initialised()) - { - std::ofstream cache_file(filename, std::ios::binary); - if(cache_file.is_open()) - { - for(const auto &it : CLKernelLibrary::get().get_built_programs()) - { - std::vector> binaries = it.second.getInfo(); - ARM_COMPUTE_ERROR_ON(binaries.size() != 1); - const std::string kernel_name = it.first; - size_t kernel_name_size = kernel_name.length(); - size_t binary_size = binaries[0].size(); - cache_file.write(reinterpret_cast(&kernel_name_size), sizeof(size_t)); - cache_file.write(reinterpret_cast(&binary_size), sizeof(size_t)); - cache_file.write(kernel_name.c_str(), kernel_name_size); - cache_file.write(reinterpret_cast(binaries[0].data()), binaries[0].size()); - } - cache_file.close(); - } - else - { - ARM_COMPUTE_ERROR("Cannot open cache file"); - } - } -#else /* ARM_COMPUTE_CL */ - ARM_COMPUTE_UNUSED(filename); -#endif /* ARM_COMPUTE_CL */ -} } // namespace utils } // namespace arm_compute diff --git a/utils/Utils.h b/utils/Utils.h index 1c577d9eea..23846cfc53 100644 --- a/utils/Utils.h +++ b/utils/Utils.h @@ -822,18 +822,6 @@ int compare_tensor(ITensor &tensor1, ITensor &tensor2, T tolerance) return num_mismatches; } - -/** This function saves opencl kernels library to a file - * - * @param[in] filename Name of the file to be used to save the library - */ -void save_program_cache_to_file(const std::string &filename = "cache.bin"); - -/** This function loads prebuilt opencl kernels from a file - * - * @param[in] filename Name of the file to be used to load the kernels - */ -void restore_program_cache_from_file(const std::string &filename = "cache.bin"); } // namespace utils } // namespace arm_compute #endif /* __UTILS_UTILS_H__*/ -- cgit v1.2.1