The H200 NVL AI accelerator, set to be generally available in December 2024, boasts a 1.5x memory increase and is ideal for efficient data center operations.
Companies can fine-tune large language models within a few hours with the upcoming H200 NVL GPU, showcasing significant advancements in GPU technology.
NVIDIA's director, Dion Harris, highlighted that the upcoming H200 NVL is designed for lower power requirements, making it suitable for air-cooled data centers.
Partnering vendors like Dell, Lenovo, and Supermicro will support the H200 NVL, ensuring a wide availability of this advanced GPU technology.
Collection
[
|
...
]