Re:MegaTrain: Full Precision Training of 100B+ Parameter LLM...
by rldjbpin on Thursday April 09, 2026 @04:15AM
(#47700675)
even if it scales down, the host-to-gpu memory ratio is around 10 to 1 (1.5 TB vs 141 GB) [1].
might need some odd system builds, improbable with current pricing, to replicate or better the ratio. such as a ~256 GB system for a 4090 (with 24 GB VRAM).
might need some odd system builds, improbable with current pricing, to replicate or better the ratio. such as a ~256 GB system for a 4090 (with 24 GB VRAM).
[1] https://www.nvidia.com/en-us/data-center/h200/