wowsuch ai
Live

GoogleDeveloper platform

Coral Compiler v1.0

Coral Compiler v1.0 is an open-source IREE/MLIR compiler and simulation stack for targeting Coral NPU hardware. It accepts models from JAX, PyTorch, and TensorFlow and lowers workloads to Coral NPU instructions.

What it does

Developers can compile and benchmark ML models for Coral NPU hardware and tune silicon configurations before fabrication.

Availability
v1.0 foundation released
Pricing model
Not established
Deployment
Open-source toolchain
Privacy
Not established
Licence
Open-source
Capability
traininganalysis
Modality
textimageaudiomultimodal
Access
CLISDKopen weights
Deployment
on-premises
Pricing
open-source
Target user
developerresearcherenterprise

Evidence

4 claims
  • Google is releasing the v1.0 foundation of its open-source MLIR/IREE compiler toolchain. [1]
    To make the hardware easy to target without writing custom assembly, we are also releasing the v1.0 foundation of our open-source MLIR/IREE compiler toolchain .
    Google. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The compiler accepts models from JAX, PyTorch, and TensorFlow. [1]
    The Coral Compiler v1.0, based on IREE/MLIR: An open-source, end-to-end compiler and simulation stack that takes JAX, PyTorch, and TensorFlow models straight to the Coral NPU hardware target.
    Google. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The compiler partitions workloads between the host CPU and Coral NPU core. [1]
    The compiler partitions workloads between the host CPU and the Coral NPU core, handles operator fusion, tiles matrices to fit local SRAM and matrix accumulators, and progressively lowers workloads directly to Coral NPU scalar, RVV v1.0 vector, and VME matrix instructions.
    Google. Quoted text is outside CC BY 4.0 and keeps its original rights.
  • The stack includes an open-source MPACT simulator and a reference FPGA design. [1]
    Alongside our open-source MPACT simulator and reference FPGA design, hardware and software teams can benchmark models and tune silicon configurations before a chip ever goes to fabrication.
    Google. Quoted text is outside CC BY 4.0 and keeps its original rights.

Sources

1

Versions

4
VersionEffectiveChangeChanged fieldsState
42026-10-28 00:15 UTCmaterial changecapability, access, deploymentCurrent
32026-10-23 01:22 UTCmaterial changecapability, modality, access, availability, deploymentCurrent
22026-10-22 00:40 UTCmaterial changecapability, access, availability, deploymentCurrent
12026-10-20 17:19 UTCpublicationNoneCurrent

In briefs

2

Open in the catalog

Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.