Skip to content

Instantly share code, notes, and snippets.

💫
in a crazy daze

Yuan Shuai ysh329

💫
in a crazy daze
Block or report user

Report or block ysh329

Hide content and notifications from this user.

Learn more about blocking users

Contact Support about this user’s behavior.

Learn more about reporting abuse

Report abuse
View GitHub Profile
@ysh329
ysh329 / Matrix.md
Created Sep 21, 2018 — forked from nadavrot/Matrix.md
Efficient matrix multiplication
View Matrix.md

High-Performance Matrix Multiplication

This is a short post that explains how to write a high-performance matrix multiplication program on modern processors. In this tutorial I will use a single core of the Skylake-client CPU with AVX2, but the principles in this post also apply to other processors with different instruction sets (such as AVX512).

Intro

Matrix multiplication is a mathematical operation that defines the product of

@ysh329
ysh329 / JetsonTK1_nvcc_fatal_computer_60.md
Created Jun 23, 2018 — forked from MarsMSJ/JetsonTK1_nvcc_fatal_computer_60.md
Jetson TK1 - caffe compilation issue: nvcc fatal : Unsupported gpu architecture 'compute_60'
View JetsonTK1_nvcc_fatal_computer_60.md

Jetson TK1 - caffe compilation issue: nvcc fatal : Unsupported gpu architecture 'compute_60'

Summary

I'm posting here a detailed summary of the problem and the situation with the Jetson TK1 board and compiling the final caffe release. The final caffe release requires cudnn v.5. If you attempt to compile on the Jetson TK1 board you will give an error from nvcc saying "Unsupported gpu architecture". I describe the

Problem

On the Jetson TK1 if you clone the final caffe release and compile:
make all -j 4

The nvcc compiler will show the following error:

You can’t perform that action at this time.