--- title: KOURAN Morphable Compute Lattice Architecture | Garmo Labs url: https://garmolabs.com/ip/kouran-morphable-compute-lattice-architecture-ip-3645 description: One morphable PE lattice is intended to map dense matrix, sparse tensor, vector, reduction, graph and solver/search primitives rather than separate fixed-function dense/sparse islands. --- [Ip](https://garmolabs.com/marketplace) › IP-3645 AI & tensor compute # KOURAN Morphable Compute Lattice Architecture One morphable PE lattice is intended to map dense matrix, sparse tensor, vector, reduction, graph and solver/search primitives rather than separate fixed-function dense/sparse islands. Source / reference evidence [Request evaluation](https://garmolabs.com/contact?products=ip-ip-3645) KOU AI & tensor compute THE TECHNICAL ROLE ## A clear integration boundary. KOURAN Morphable Compute Lattice Architecture provides the documented ai / hpc / general compute architecture responsibility: One morphable PE lattice is intended to map dense matrix, sparse tensor, vector, reduction, graph and solver/search primitives rather than separate fixed-function dense/sparse islands. Evaluate the exact KOURAN-IP-015 implementation and confirm its integration contract before using it inside the target design. 01 One morphable PE lattice is intended to map dense matrix, sparse tensor, vector, reduction, graph and solver/search primitives rather than separate fixed-function dense/sparse islands. TECHNICAL OVERVIEW ## The details matter. [Download the product brief](https://garmolabs.com/static/briefs/ip-ip-3645.pdf) Canonical identity IP-3645 Native implementation identity KOURAN-IP-015 Documented responsibility One morphable PE lattice is intended to map dense matrix, sparse tensor, vector, reduction, graph and solver/search primitives rather than separate fixed-function dense/sparse islands. Recorded evidence class Source / reference evidence Interface and physical parameters Bound to the selected source/configuration; not inferred from name or family. ### Before integration. KOURAN Morphable Compute Lattice Architecture provides the documented ai / hpc / general compute architecture responsibility: One morphable PE lattice is intended to map dense matrix, sparse tensor, vector, reduction, graph and solver/search primitives rather than separate fixed-function dense/sparse islands. Evaluate the exact KOURAN-IP-015 implementation and confirm its integration contract before using it inside the target design. - Exact implementation and configuration manifest - Documented interfaces and clock/reset contract - Available reference or verification collateral - Recorded physical views where authorized and applicable - Integration acceptance criteria and unresolved gate list The selected release must confirm: - Source/configuration identity - Interface and dependencies - Verification evidence - Process/library where applicable - Authorized package and acceptance criteria Unpublished values are not inferred from family names or architecture targets. [Discuss the integration](https://garmolabs.com/contact?products=ip-ip-3645) Source / reference evidence ## Know the release. Know what it establishes. Architecture docs + morphable PE + 4x4 mesh RTL. This is a controlled-portfolio summary, not a fresh execution of the chip qualification flow. [Explore the evidence center](https://garmolabs.com/evidence) [Qualification record](https://garmolabs.com/qualification/ip-ip-3645) ### Release-specific qualification Evidence shown is our record for this canonical implementation, not a fresh tool rerun. External foundry acceptance, silicon measurements, standards certification and unlisted interface/physical parameters are not inferred. Exact delivery and rights are agreed before licensing. Canonical record: P1 / IMPLEMENTATION-BACKED ARCHITECTURE; FULL 64-PE TILE NOT PRESENT A PATH THAT FITS THE PROJECT ## Build with KOURAN Morphable Compute Lattice Architecture. Technical scope, rights, support and release configuration are agreed before delivery. evaluation ### Review the exact recorded implementation and agreed test scope Inspect the named release and establish technical fit. [Discuss this scope](https://garmolabs.com/contact?products=ip-ip-3645&scope=evaluation) production ### Named-design rights for the agreed qualified release Agree deployment or design rights, deliverables and support. [Discuss this scope](https://garmolabs.com/contact?products=ip-ip-3645&scope=production) custom ### Target integration, verification or physical qualification Define the customer-specific work and acceptance criteria. [Discuss this scope](https://garmolabs.com/contact?products=ip-ip-3645&scope=custom) ## Keep exploring. AI & tensor compute [## KX Activation](https://garmolabs.com/ip/kx-activation-ip-2837) Digital RTL implementation indexed as kx_activation. The recorded role is activation; exact interfaces, parameters and functional coverage are release-bound. Internal physical evidence IP-2837 · Internal physical evidence AI & tensor compute [## KX Ai Avgpool 2X2](https://garmolabs.com/ip/kx-ai-avgpool-2x2-ip-2845) Digital RTL implementation indexed as kx_ai_avgpool_2x2. The recorded role is ai avgpool 2x2; exact interfaces, parameters and functional coverage are release-bound. Internal physical evidence IP-2845 · Internal physical evidence AI & tensor compute [## KX Ai Bias Adder](https://garmolabs.com/ip/kx-ai-bias-adder-ip-2846) Digital RTL implementation indexed as kx_ai_bias_adder. The recorded role is ai bias adder; exact interfaces, parameters and functional coverage are release-bound. Internal physical evidence IP-2846 · Internal physical evidence ## Your next big idea. Let’s build it together. Start with a chip, a software release or a single IP block. [Talk to Garmo](https://garmolabs.com/contact) [Start exploring](https://garmolabs.com/developers) CANONICAL EVIDENCE RECORD · IP-3645 ### Source / reference evidence Source and/or executable reference evidence is recorded. This does not establish HDL or physical qualification unless separately stated. **Record integrity and qualification boundary**4e753d22c51d9abd5a4212e33f3ba7752b196396a3397a8bd616170e3226450c Evidence ranking for evaluation prioritization; not a probability or certification. [How evidence priority is determined ](https://garmolabs.com/marketplace/evidence)