News for hackers. Stuff that matters.

A comment on MegaTrain: Full Precision Training of 100B+ Parameter LLMs on a Single GPU. Parent comment

Re:MegaTrain: Full Precision Training of 100B+ Parameter LLM... by nekusar on Wednesday April 08, 2026 @11:35AM (#47691665)
When the comparison is again 128 H100's , yeah, this is a crazy good upgrade.

And you can rent H100's and H200s for not that much per hour.

[ Parent | on HN ]
Re:MegaTrain: Full Precision Training of 100B+ Parameter LLM... by p_stuart82 on Wednesday April 08, 2026 @01:47PM (#47693725)
$2-4/hr always sounds cheap until you multiply by wall clock and reruns
[ Parent | on HN ]
Re:MegaTrain: Full Precision Training of 100B+ Parameter LLM... by jeremyjh on Wednesday April 08, 2026 @02:10PM (#47694056)
Yes but they are getting only 341 tok/s. A 2.5 trillion run would take over 200 years.
[ Parent | on HN ]