nvComp

Training LLMs requires periodic checkpoints. These full snapshots of model weights, optimizer states, and gradients are saved to storage so training can resume...

Compression is a common technique to reduce storage costs and accelerate input/output transfer times across databases, data-center communications,...

The CUDA library, nvCOMP, now offers support for zStandard and Deflate standards, as well as modified-CRC32 checksum support and improved ANS performance.

Compression can improve performance in a variety of use cases such as DL workloads, databases, and general HPC. On the GPU, compression can accelerate inter-GPU...