1.. _libc_gpu_motivation: 2 3========================== 4Motivation and Limitations 5========================== 6 7.. contents:: Table of Contents 8 :depth: 4 9 :local: 10 11Motivation 12========== 13 14This project aims to provide a large subset of the C standard library to users 15of GPU accelerators. We deliberately choose to only implement a subset of the C 16library as some features are not expressly useful or easily implemented on the 17GPU. This will be discussed further in `Limitations <libc_gpu_limitations>`_. 18The main motivation behind this project is to provide the well understood C 19library as a firm base for GPU development. 20 21The main idea behind this project is that programming GPUs can be as 22straightforward as programming on CPUs. This project aims to validate the GPU as 23a more general-purpose target. The implementations here will also enable more 24complex implementations of other libraries on the GPU, such as ``libc++``. 25 26Host services and C library features are currently provided sparsely by the 27different GPU vendors. We wish to provide these functions more completely and 28make their implementations available in a common format. This is useful for 29targets like OpenMP offloading or SYCL which wish to unify the offloading 30toolchain. We also aim to provide these functions in a format compatible with 31offloading in ``Clang`` so that we can treat the C library for the GPU as a 32standard static library. 33 34A valuable use for providing C library features on the GPU is for testing. For 35this reason we build `tests on the GPU <libc_gpu_testing>`_ that can run a unit 36test as if it were being run on the CPU. This also helps users port applications 37that traditionally were run on the CPU. With this support, we can expand test 38coverage for the GPU backend to the existing LLVM C library tests. 39 40.. _libc_gpu_limitations: 41 42Limitations 43=========== 44 45We only implement a subset of the standard C library. The GPU does not 46currently support thread local variables in all cases, so variables like 47``errno`` are atomic and global. Furthermore, the GPU under the OpenCL execution 48model cannot safely provide a mutex interface. This means that features like 49file buffering are not implemented on the GPU. We can also not easily provide 50threading features on the GPU due to the execution model so these will be 51ignored, as will features like ``locale`` or ``time``. 52