Name: Kun Wu
Type: User
Company: Univ. of Illinois at Urbana-Champaign
Bio: Making Composable & Scalable Stack for Data-Intensive Applications!⚓@illinois-impact PhD student on GPU compilers & libraries⚓BEng (Tsinghua)
Location: IMPACT Group, Urbana, IL
Blog: kunwu.me
Kun Wu's Projects
mips assembly implementation of counting 8queen solution
aufs, a simple linux kernel FS, with write enabled
Implementation of BEAR and SlashBurn.
Efficient GPU kernels for block-sparse matrix multiplication and convolution
Automatically exported from code.google.com/p/blueprint-ywjtffdq
Stanford NLP Coursera
C++ Primer 5ed answers
CSAPP网站上的9个labs
CSE 351: The Hardware/Software Interface (taught by Luis Ceze)
code for benchmarking GPU performance based on cublasSgemm and cublasHgemm
Samples for CUDA Developers which demonstrates features in CUDA Toolkit
Hooked CUDA-related dynamic libraries by using automated code generation tools.
CUDA Library Samples
[In progress] Performant memcpy within cuda kernel
Private clone of https://github.com/cupy/cupy
CUDA Templates for Linear Algebra Subroutines
All hail, Thy Highest University (THU)
Timing code backed up at the 1.1.2.timing branch
private clone of https://github.com/ydwu4/dgl-hack