Tags: LLM training

A side‑by‑side look at three leading AI accelerators—Huawei's Ascend 950 series, Nvidia's H200 (GH100 Hopper), and AMD's Radeon Instinct MI300 (Aqua Vanjaram). The comparison covers architecture, process technology, transistor counts, die size, memory type and capacity, bandwidth, compute performance across FP8, FP16, FP32 and FP64, and target scenarios such as large‑scale LLM training, inference, and high‑performance computing. Availability timelines differ, with each vendor positioning its chip for data‑center and HPC workloads. Ler mais

Scale AI has filed a lawsuit against former sales employee Eugene Ling and competitor Mercor, alleging that Ling took more than 100 confidential documents and tried to use them to win a major Scale customer for Mercor. Mercor co‑founder Surya Midha says the firm did not use any of Scale’s trade secrets and that Ling merely stored old files on a personal drive. Ling has responded on X, expressing regret and denying any malicious intent. The case highlights growing tensions in the AI data‑training market, especially after Meta’s multibillion‑dollar investment in Scale AI‑training services. Ler mais