I know this is a hardware related question but i did not find any article regarding their speed improvements in data science field, only gaming or os benchmark. I want to upgrade my storage system but can not decide between nvme or sata ssd.
My gpu is gtx 1070, cpu i7 7700k.might upgrade later. I am using cuda 10 version. I want to know your experiences about training on gpu with data on nvme vs sata. My datasets often includes huge number of small size individual files, often i convert them into numpy files which takes total size of more than 100gb(combined).
So guys, should i go with nvme or sata ssd? In the same price range i can get more storage with sata ssds.
Have a look at this great post by @rwightman.
Based on his experiments he recommends buying NVME drives:
- Quick fix: buy an NVME SSD drive, or two.
- SATA SSD is not necessarily going to cut it, you can saturate them with small to medium image files + default loader/dataset setups feeding multiple GPUs.