Reliable GPU compute for large training & inference jobs

T1Covered by 1 source · first reported by Stripe / Shopify Eng / Netflix TechBlog / Dropbox Tech / Slack Eng / Spotify Eng at 14 Oct, 19:47 UTC

Authors: Nick Rutigliano, Argha C, Samartha Chandrashekar How we detect failing GPUs and network fabric, remediate machines automatically, and let long-running jobs survive hardware loss. Open Netflix, and the home screen is assembled for you: which rows appear, the order they appear in, the titles…