inference/research
verifiedhigh · 0.85updated 2026-10-11

SambaNova SN40L: Scaling the AI Memory Wall with Dataflow and Composition of Experts (arXiv:2405.07518)

whitepaper

Fields

id
sources/ainv-sambanova-sn40l-paper
Publisher
SambaNova Systems
Kind
whitepaper
Published
2024-05-11
Accessed
2026-10-11
notes
First-party technical paper by SambaNova Systems architecture team (Prabhakar et al.) detailing the SN40L Reconfigurable Dataflow Unit (RDU), TSMC 5nm process, 2.5D CoWoS package, dual Reconfigurable Dataflow Dies (RDD), 638 BF16 TFLOPS, 520 MiB on-chip PMU SRAM, 64 GiB co-packaged HBM, and up to 1.5 TiB DDR DRAM.

Notes

First-party technical paper by SambaNova Systems architecture team (Prabhakar et al.) detailing the SN40L Reconfigurable Dataflow Unit (RDU), TSMC 5nm process, 2.5D CoWoS package, dual Reconfigurable Dataflow Dies (RDD), 638 BF16 TFLOPS, 520 MiB on-chip PMU SRAM, 64 GiB co-packaged HBM, and up to 1.5 TiB DDR DRAM.

Cross-references

This record references no other records.