COHERENT COMPUTING
KMJ-X130

Less disconnected. More capable.

CPU, accelerator and memory execution designed to share a computational fabric.

KMJ-X130 architectural concept visualization; not fabricated silicon

Architectural concept. Not a die photograph, floorplan or measured-silicon demonstration.

CPUScalar control
AcceleratorSpecialized work
CoherenceShared architectural fabric

One coherent view. More ways to compute.

KMJ-X130 brings scalar control, specialized execution and memory together. Coherence and the delivery boundary are specified for the selected implementation, not assumed across arbitrary devices.

KMJ-X130 / conceptual operation01Control02Share03Accelerate04RecoverArchitectural explanation. Not a floorplan, timing diagram or hardware simulation.Conceptual operation · choose a stage, or play the explanation
Explanatory sequence, not hardware simulation.
CPU execution manages the application and its work.
SEE THE REASON BEHIND THE ARCHITECTURE

A CPU-led accelerator application

Trace the relationship between host control, shared state and accelerator work. Package, memory and advanced-node implementation options require their own qualification.

Try a related software workflow
Continue to GSP ↗
01

Coherent CPU + accelerator architecture

02

AI, HPC, graph and network composition

03

Memory/fabric and RAS integration

04

R1.0 package-ready implementation

TECHNICAL OVERVIEW

The details matter.

Download the product brief
Canonical identity
IP-062
Responsibility
Extreme-scale superchip
Documented building blocks
Full KMJ CPU/AI/HPC/memory/graph/network stack; chiplet fabric; HBM; RAS/security.
Delivery boundary
Release-specific physical views
Evidence scope
Qualified release
Qualified release

Know the release.
Know what it establishes.

The later R1.0 record identifies KMJ-X130 as signoff-complete and ChipIgnite package-ready. Its controlled release is distinct from broader family roadmap configurations.

Source basis: Bible v7.42 §§126–127, R1.0 controlled release ledger. This is a controlled-portfolio summary, not a fresh execution of the chip qualification flow.

Release-specific qualification

Package-ready is pre-silicon evidence, not foundry acceptance, returned devices or measured silicon characterization.

Canonical record: Engineering build-spec family

A PATH THAT FITS THE PROJECT

Build with KMJ-X130.

Technical scope, rights, support and release configuration are agreed before delivery.

evaluation

Architecture and release evaluation

Inspect the named release and establish technical fit.

Discuss this scope
production

Chip / SoC licensing program

Agree deployment or design rights, deliverables and support.

Discuss this scope
custom

Custom silicon integration

Define the customer-specific work and acceptance criteria.

Discuss this scope

Your next big idea.
Let’s build it together.

Start with a chip, a software release or a single IP block.