Skip to knowledge
VLSI Hub / Knowledge

Auditable claim record

NVIDIA A100 Tensor Core GPU · uses memory

HBM2 · SUPPORTED

urn:vlsi:claim:60cd0a44a83d9943e183d4010aa1b2d8f3ae5b30a61df13f4d550f2118aa02cb

This is the cited 40 GB configuration; later A100 variants use HBM2e.

Back to NVIDIA A100 Tensor Core GPU · Complete history as JSON · JSON-LD

Evidence

NVIDIA Ampere Architecture In-Depth

NVIDIA specifies 40 GB of HBM2 for the A100 discussed in the architecture article.

NVIDIA Technical Blog · lines 82-84 / 40 GB HBM2 and 40 MB L2 cache · supports

Published 2020-05-14 · Retrieved 2026-09-08T15:56:30.763Z

urn:vlsi:evidence:67f3357f3f7b99795077d217c0ebf2b799d42747b0bbb46db5b3ed53f4cbaa37

Assessment history

New assessments preserve earlier records. A correction changes the current interpretation without rewriting what VLSI knew before it.

Recorded assessments, oldest first
RevisionRecorded at (UTC)StatusValidityReviewer & reason
12026-09-08T15:57:21.457ZSUPPORTEDpoint
2020-05-14
curated-source-agent (agent)

Evidence ingested by curated-primary-sources/1.0.0. This is the cited 40 GB configuration; later A100 variants use HBM2e.