mediaic / UMI

This project implements the paper "Unrolled Memory Inner-Products: An Abstract GPU Operator for Efficient Vision-Related Computations" (ICCV 2017).

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

English

Brief

This project implements the UMI Operator of the paper "Unrolled Memory Inner-Products: An Abstract GPU Operator for Efficient Vision-Related Computations" (ICCV 2017).

Compile Test Files

CMake and Glog library is required for compiling.

mkdir build
cmake ..
make

The directory main/ show some examples you can program with UMI operator. I will prepare more document in the future.

FAQ

If you are using new Linux such as Arch, you have to install gcc/g++ version 5. You also have to set CMAKE_CXX_COMPILER, CMAKE_C_COMPILER and CUDA_HOST_COMPILER accordingly.

中文版 README

I made this part since I am a native Mandarin Chinese speaker (zh_TW).

概述

本專案實作了 "Unrolled Memory Inner-Products: An Abstract GPU Operator for Efficient Vision-Related Computations" (ICCV 2017) 的 UMI Operator

如何編譯

你需要 CMake 以及 Glog。

mkdir build
cmake ..
make

main/ 下面的檔案提供了一些使用 UMI operator 的例子,我未來會提供更完整的文件。

常見問題

如果你是用新一點的 Linux 像是 Arch,你得想辦法安裝 g++ version 5。 你也要設定一下 CMake 的 CMAKE_CXX_COMPILER、CMAKE_C_COMPILER、CUDA_HOST_COMPILER. 我還沒測過其他平台能不能使用。

About

This project implements the paper "Unrolled Memory Inner-Products: An Abstract GPU Operator for Efficient Vision-Related Computations" (ICCV 2017).

License:GNU General Public License v3.0


Languages

Language:C++ 63.7%Language:Cuda 28.1%Language:C 4.6%Language:CMake 3.6%