Do EnCharge EN100's 200 TOPS / 8.25 W and 'up to 20x performance per watt' figures hold against a primary or independent benchmark?
claim-encharge-en100-switched-capacitor-in-memory-compute records that the EN100 delivers 200 TOPS at 8.25 W and "performance per watt up to 20 times better than competing chips." These are company-sourced figures reported through IEEE Spectrum (Tier 2). Quantitative claims require Tier 1–2 with a primary home; a vendor number relayed by journalism does not clear the floor, hence the [unverified-quant] flag.
Why it matters
The "20x better perf/watt" multiplier is exactly the kind of figure that garbles in translation and that a vendor has every incentive to state under favorable conditions (which precision, which sparsity, which competing chip, measured or peak). The capture's own note says analog AI is a field where "efficiency claims have not yet beaten digital's commercial gravity" — so the efficiency number is the load-bearing part of the optimistic case and deserves a primary before the vault leans on it.
What would answer it
- An EnCharge AI primary source: the EN100 datasheet / spec sheet, or a technical white paper stating the benchmark conditions (numeric precision, workload, comparison baseline).
- A peer-reviewed / conference paper from Naveen Verma's group (e.g. ISSCC, VLSI, or a journal) reporting the switched-capacitor in-memory-compute macro's measured TOPS/W — the academic lineage behind the product.
- Any independent third-party measurement of EN100 perf-per-watt on a named model.
Note specifically: what is the 20x measured against, and at what precision? Resolve to updating the claim-note's flag once a primary confirms or corrects the figures.
Progress log
- 2026-07-31: Fresh search (product page, launch press release, ~12 trade-press pieces, EN100 EAP-status pages) found no independent or primary benchmark and no named comparison baseline for "20x" — see claim-no-independent-benchmark-of-encharge-en100-efficiency-figures and claim-encharge-en100-20x-per-watt-claim-has-no-stated-baseline. The closest primary found is a 2021 ISSCC paper from Verma's lab on the predecessor academic chip (claim-2021-isscc-verma-lab-chip-is-primary-precedent-not-en100-evidence) — real Tier-1 evidence for the switched-capacitor mechanism, but not for the EN100 product's own 200 TOPS/8.25 W or 20x figures, which remain unconfirmed. Status stays
open: still needs a datasheet/white paper stating benchmark conditions, or an independent third-party measurement, neither of which surfaced this round. EN100 remains in Early Access as of this search, which is itself part of why no outside test exists yet.