News & releases

What shipped, and what changed.

Announcements from the lab and a versioned changelog. We publish numbers only once they're measured on production silicon — if a post says "up to," that's a ceiling, not a promise.

From the lab

Notes on what we're shipping.

2026-08-12
Hardware

NP-2 first silicon is back from the fab

Our next-generation die is powered up and passing the first functional vectors. Bring-up is early, but the tiled compute array is already running transformer workloads at a rate that changes what we can promise at the same power envelope. We'll publish measured numbers once they're on production silicon — not from simulation, and not before.

2026-06-24
Toolchain

PeakStack 2.1: transformers and small language models

On-device LLM inference joins the toolchain. INT4 weight quantization, activation sparsity, and a KV-cache path tuned for the on-die SRAM. A 1.5B-parameter model now decodes at 40 tokens/s on a single NP-1 inside a single-digit watt envelope.

2026-04-15
Reference design

The NP-1 reference board, now in a 4 mm envelope

We rebuilt the reference design around thermal headroom. The new board fits a sealed industrial housing with no fan and no vent, holding the full NP-1 performance envelope across -40 °C to 105 °C junction. What that unlocks for field-hardened equipment is the point of the post.

2026-02-02
Documentation

NP-1 datasheet rev 2.4

Expanded thermal characterization, BF16 support documented, and a full electrical spec for the LPDDR5 PHY. Nothing changed on the die — this is documentation catching up with what the silicon already does.

2025-11-19
Product

NP-C1 sampling to select customers

The compact vision processor moves from engineering samples to first design-ins with camera and sensor partners. Sub-4 mm package, 1.5 W typical, and the same toolchain as NP-1 — a model compiled for one runs on the other.

2025-09-03
Toolchain

PeakStack 2.0 goes GA

Automatic weight and activation sparsity, and a mixed-precision policy engine, ship to all partners. One calibration pass, a per-layer precision policy, flash — no hand-written kernels.

Changelog

Releases, by version.

Software and silicon, in the order they shipped. Datasheet revisions are tracked separately from silicon revisions — a datasheet change doesn't always mean a new die.

Software — PeakStack

PeakStack 2.12026-06

Transformer and small-LLM inference

INT4 weight quantization, activation sparsity, and a KV-cache scheduler tuned for on-die SRAM. Adds a decode path for on-device language models.

PeakStack 2.02025-09

Automatic sparsity & mixed precision

A sparsity pass applied without touching model code, and a per-layer precision policy engine. The first release with hardware-in-the-loop profiling as default.

PeakStack 1.02023

First stable release

PyTorch and ONNX import, INT8 and FP16, calibration-aware quantization, and the one-pass-to-flash flow that's still the backbone of the toolchain.

Hardware — silicon & datasheets

NP-2 · first silicon2026-08

Next-generation die, bring-up underway

First silicon returned from the fab and passing initial functional vectors. Numbers to follow on production silicon.

NP-1 · datasheet rev 2.42026-02

BF16, LPDDR5 PHY, expanded thermal

Documentation release. BF16 support, LPDDR5 electrical specification, and expanded junction-temperature characterization. No die change.

NP-C1 · datasheet rev 1.82025-11

Sampling revision

Engineering samples to first design-ins. Shares NP-ISA v2 and the PeakStack toolchain with NP-1.

NP-1 · production2023

First production silicon

First design wins ship in robotics and smart-camera systems. PeakStack reaches a stable 1.0 alongside it.

Revisions are additive and backward-compatible within a silicon generation. A model compiled for an earlier PeakStack release runs on a later one without recompiling.