社区帖子称消费级RTX存在FP32累加性能限制
一名Reddit作者称,NVIDIA为区分消费级与专业版、数据中心产品,故意将RTX 20、30、40和50系列部分使用FP32累加的张量运算性能减半。作者表示,各代显卡受影响的精度有所不同,RTX 50系列的FP4据称不受影响。作者列举了受影响的场景,包括LoRA训练、微调、扩散任务和LLM预填充,以及大批量文本生成。帖子还介绍了尝试解除限制的方法,并认为RTX 20和30系列的成功可能性较高,RTX 40系列相对较高,RTX 50系列目前较低。
如果相关限制确实存在,消费级RTX GPU在训练、扩散和LLM工作负载中的表现可能明显慢于专业级替代产品,因此这些限制值得购买AI硬件的用户关注。
Hello guys, hoping you're doing fine!
Small post I want to make since seeing a lot of posts about the unlocks and such.
What is the unlock in simple terms?
NVIDIA on purpose, neutered the performance of operations that use FP32 accumulation with GEMMs, with the objective of market segmentation and making consumer GPUs slower to train, or do some inference tasks slowly as well when compute is involved (like diffusion, prompt generation).
Those are "unlocked" on the prosumer and datacenters gpus.
Cards with "unlocked" FP32 accumulation performance:
#NVIDIA#GeForce RTX 20 Series#GeForce RTX 30 Series#GeForce RTX 40 Series#GeForce RTX 50 Series#NVIDIA RTX A6000#NVIDIA RTX 6000 Ada#NVIDIA RTX 6000 PRO