精选理由
Together AI搞了个新基准ParallelKernelBench，专门测AI写多GPU CUDA代码的能力，比单核测试更贴近真实场景。
Together AI团队在@aiDotEngineer大会上发布了ParallelKernelBench，一个开源基准测试。该基准专门评估大语言模型能否为通信密集型多GPU工作负载编写高效的CUDA内核。与现有单GPU测试不同，ParallelKernelBench聚焦真实场景下的并行内核性能。开发者可使用该基准对比不同LLM的代码生成质量。
原文 · Together AI
Multi-GPU kernels are the real test for coding mod…
Multi-GPU kernels are the real test for coding models. Today at @aiDotEngineer, @simran_s_arora shared ParallelKernelBench, an open-source benchmark for evaluating whether LLMs can write fast CUDA kernels for real communication-heavy workloads.
Proud to see this work from the Together AI Frontier Performance team.