The HPE ProLiant Compute DL384 Gen12 server with NVIDIA GH200 Grace Hopper Superchip 144 GB achieved unprecedented MLPerf Inference Datacenter v4.1 results, ranking first in 7 of 16 models, including DLRM and Mixtral-8x7B. This marks the first submission with the 144 GB configuration, outperforming systems with 96 GB and 141 GB GPUs in generative AI inference. Ideal for memory-intensive workloads, the server offers up to 10X higher performance for large-scale AI and HPC tasks. HPE and NVIDIA also co-engineered HPE Private Cloud AI to help enterprises securely deploy and scale AI solutions.

Join for free to read