Hardware
Run High-Performance Core Math at Scale with NVIDIA nvmath-python
NVIDIA **nvmath‑python v1.0** is a Python library that wraps CUDA‑X and NVPL math libraries (cuFFT, cuBLASLt, cuDSS, cuSPARSE, cuTENSOR, cuBLASMp) to give NumPy, CuPy, and PyTorch users GPU‑accelerate
NVIDIA nvmath‑python v1.0 is a Python library that wraps CUDA‑X and NVPL math libraries (cuFFT, cuBLASLt, cuDSS, cuSPARSE, cuTENSOR, cuBLASMp) to give NumPy, CuPy, and PyTorch users GPU‑accelerated performance for common math operations on CPUs, single GPUs, or distributed multi‑GPU/multi‑node systems.
It adds Universal Sparse Tensor (UST) support via a domain‑specific language, offering custom sparse formats, plus both generic and hardware‑optimized APIs—including fused composite operations and just‑in‑time kernel fusion with cuBLASLt—while providing stateful autotuning, fast installation, flexible backend selection, and detailed computational tracing through Python’s logging module.
Related
- Simplify Sparse Deep Learning with Universal Sparse Tensor in nvmath-python
- NVIDIA CUDA 13.3 Enhances GPU Development with Tile Programming in C++, Compiler Autotuning, and Python Updates
- Develop High-Performance GPU Kernels in C++ with NVIDIA CUDA Tile
Source: NVIDIA Developer | 2026-07-30