Y Hacker News new | ask | show | jobs

user: skidrow created: 2024-07-02 karma: 364 submissions: Creating custom kernels for the AMD MI300 2 points | 0 comments Implementing a Fast Tensor Core Matmul on the Ada Architecture 4 points | 0 comments Matrix Core Programming on AMD GPUs 116 points | 5 comments Implementing a Fast Tensor Core Matmul on the Ada Architecture 3 points | 0 comments Matrix Core Programming on AMD GPUs 2 points | 0 comments Creating custom kernels for the AMD MI300 1 points | 0 comments Implementing a Fast Tensor Core Matmul on the Ada Architecture 2 points | 0 comments Matrix Core Programming on AMD CDNA3 and CDNA4 Architecture 24 points | 3 comments Creating custom kernels for the AMD MI300 2 points | 0 comments Implementing a Fast Tensor Core Matmul on the Ada Architecture 2 points | 0 comments Advanced Matrix Multiplication Optimization on Multi-Core Processors (2024) 85 points | 3 comments Creating custom kernels for the AMD MI300 2 points | 0 comments Introduction to Matrix Core Programming on AMD CDNA3 and CDNA4 Architecture 2 points | 0 comments Creating custom kernels for the AMD MI300 2 points | 0 comments Implementing a Fast Tensor Core Matmul on the Ada Architecture 2 points | 0 comments Creating custom kernels for the AMD MI300 1 points | 0 comments Implementing a Fast Tensor Core Matmul on the Ada Architecture 4 points | 0 comments Implementing a Fast Tensor Core Matmul on the Ada Architecture 2 points | 1 comments Compiler Explorer: An Essential Kernel Playground for CUDA Developers 2 points | 0 comments Creating custom kernels for the AMD MI300 1 points | 0 comments DeepSeek-R1 and FP8 Mixed-Precision Training 2 points | 0 comments How to Write a Fast Matrix Multiplication from Scratch with Tensor Cores (2024) 147 points | 17 comments DeepSeek-R1 and FP8 Mixed-Precision Training 2 points | 0 comments Implementing a Fast Tensor Core Matmul on the Ada Architecture 1 points | 0 comments How to Write a Fast Matrix Multiplication from Scratch with Tensor Cores 2 points | 0 comments Understanding Peak, Max-Achievable and Delivered FLOPs 1 points | 0 comments DeepSeek-R1 and FP8 Mixed-Precision Training 1 points | 0 comments Outperforming cuBLAS on H100: A Worklog 3 points | 0 comments