NVIDIA H100
CoreWeave's CAIOS hits 7+ GB/s per GPU on NVIDIA Blackwell Ultra
CoreWeave says its AI Object Storage service (CAIOS) averaged 7+ GB/s per GPU in tests on 16 NVIDIA Blackwell Ultra nodes. According to the company, that is a more than…
CoreWeave Reports 6.5x DeepSeek R1 Inference Gain on NVIDIA GB300 NVL72
CoreWeave says its new NVIDIA GB300 NVL72 instances, built on Blackwell Ultra GPUs, delivered a 6.5x performance improvement on inference for the DeepSeek R1 reasoning…
Red Hat AI 3.5 lets chat and batch share one GPU without latency hits
Red Hat AI 3.5 makes priority-aware serving generally available, letting chat, batch and agentic traffic share one GPU pool while interactive requests go first.…
Databricks AI Research Team Reports 8x More Training Jobs with AI Runtime
Databricks says its own AI Research team has grown the number of training jobs 8x across its AI Functions workloads since it began using agents to run experiments on AI…
Flo Health fine-tunes open model with Databricks, lifting medical safety to about 61%
Flo Health, the women's health app with 81 million monthly active users, has fine-tuned an open model with Databricks AI Runtime and raised its combined medical safety…
CoreWeave reports 51-52% MFU and 3.66-day MTTF in 1,024-GPU H100 test
CoreWeave says a six-week benchmark of up to 1,024 NVIDIA H100 GPUs reached 51-52% Model FLOPs Utilization (MFU) and a mean time to failure (MTTF) of 3.66 days. The…
CoreWeave Gets $100M From Magnetar and Plans NVIDIA H100 Access in Q1 2023
CoreWeave says Magnetar has backed the company with a $100 million investment, which will go into scaling its NVIDIA GPU-accelerated cloud infrastructure. In a year-end…
CoreWeave and NVIDIA Train MLPerf GPT-3 Benchmark in Under 11 Minutes
CoreWeave and NVIDIA say they trained the new MLPerf GPT-3 175B benchmark in under 11 minutes using more than 3,500 NVIDIA H100 Tensor Core GPUs. According to CoreWeave…
CoreWeave adds torchforge support to scale reinforcement learning on its cloud
CoreWeave has added support for torchforge, a PyTorch-scalable reinforcement learning framework, on its cloud. Researchers can run it through CoreWeave's…