NVIDIA's new Blackwell architecture has set new records for large language model (LLM) inference, delivering up to 4x more performance compared to its predecessor, the H100 GPU, in the MLPerf Inference v4.1 benchmarks. The H200 GPU also showcased remarkable performance improvements across various benchmarks. Key advancements in
Table of contents
NVIDIA Blackwell shines in MLPerf Inference debutNVIDIA H200 Tensor Core GPU delivers outstanding performance on every benchmarkA giant generative AI leap on Jetson AGX OrinConclusionSort: