Implement GPU parallel programming with CUDA C and OpenACC
When: December 24th
Venue: A125, Zhugongshan building
Morning: CUDA C/C++ Basics and hands on (9:00--11:30)
The main content:1. Basic idea of parallel computing and GPU heterogeneous parallel computing
2. CUDA C programming model and basic optimization
3.Unified Memory use
4. Streams and Concurrency
5. Code profile
6. Error handling
7. hands on
Afternoon:Introduction to OpenACC parallel programming and Handson(14:00--17:00)
The main content:1. Introduction to OpenAcc basic programming model
2. parallel/kernel directives
3. data directives
4. optimizations
5. Basic optimization method
6. hands on
Alternative: Cuda10 latest programming features
Teaching expert: Fu Qingping NVIDIA senior solution architect
Since he started his work, professor fu has been engaged in HPC system operation and platform optimization, HPC system architecture and gpu-based deep learning algorithm platform development, testing and tuning. He is currently responsible for HPC/DL GPU supercomputing architecture at NVIDIA.
Please download cuda, pgi compiler in advance!
pgicompiler:https://www.pgroup.com/products/community.htm