Narteau (discuter | contributions)
(Page créée avec « *Day 1 (April 22nd) **GPU vs CPU architecture. Memory hierarchy (briefly). **Approaches to distribute data among threads and thread blocks. **Shared memory. **Asynchronous op… »)
(Page créée avec « *Day 1 (April 22nd) **GPU vs CPU architecture. Memory hierarchy (briefly). **Approaches to distribute data among threads and thread blocks. **Shared memory. **Asynchronous op… »)
Version actuelle en date du 15 avril 2015 à 12:33
- Day 1 (April 22nd)
- GPU vs CPU architecture. Memory hierarchy (briefly).
- Approaches to distribute data among threads and thread blocks.
- Shared memory.
- Asynchronous operations. CUDA streams.
- Visualization of time lines.
- OpenACC. Compiler's directives to program GPU.
- Practical part.
- Day 2 (April 24th)
- Application of GPU to solving challenging tasks. Different approaches and application are discussed.
- Multiple GPU.
- GPU optimized libraries.
- Practical part.
- Day 3 (May 26th)
- GPU program optimization
- CUDA + FORTRAN/Java/Python/C
- Answering question and discussions. Here I plan to help researchers to find the way to adapt their code to GPU and make it effectively use compute resources.
