New cuBLAS 12.0 Features and Matrix Multiplication Performance on NVIDIA Hopper GPUs

New cuBLAS 12.0 Features and Matrix Multiplication Performance on NVIDIA Hopper GPUs

New cuBLAS 12.0 Features and Matrix Multiplication Performance on NVIDIA Hopper GPUs | NVIDIA Technical Blog

“The cuBLAS library is an implementation of Basic Linear Algebra Subprograms (BLAS) on top of the NVIDIA CUDA runtime, and is designed to leverage NVIDIA GPUs for various matrix multiplication operations. This post mainly discusses the new capabilities of the cuBLAS and cuBLASLt APIs. However, the cuBLAS library also offers cuBLASXt API targeting multiGPU distributed GEMMs. The cuBLASXt API is set to be available in Early Access in 2023 and targets GEMMs and their fusion inside device functions…”

Source: developer.nvidia.com/blog/new-cublas-12-0-features-and-matrix-multiplication-performance-on-nvidia-hopper-gpus/