# US 2026/0228156 A1 · Scale-Up Memory Access over IEEE 802.3 PMA in Processors and Accelerators including ARM-Based SoCs

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0228156 A1
- Document ID: US20260228156A1
- Status: Published application
- Kind: A1
- Application number: 19/576,413
- Filed: 2026-03-24
- Published: 2026-08-06
- Earliest claimed: 2023-12-13
- Assignees: UnifabriX Ltd.
- Applicants: UnifabriX Ltd.
- Inventors: Ronen Aharon Hyatt; Gaya Opal Hyatt; Ethan Sharon Hyatt; Gil Thieberger
- Web page: https://patentgrep.com/patent/US20260228156A1

## CPC

- G06F 13/4027: {using bus bridges (G06F13/4022 takes precedence)}
- G06F 13/4221: {being an input/output bus, e.g. ISA bus, EISA bus, PCI bus, SCSI bus}
- G06F 13/1668: {Details of memory controller}
- G06F 13/28: using burst mode transfer, e.g. direct memory access {DMA}, cycle steal (G06F13/32 takes precedence)
- G06F 13/4022: {using switching circuits, e.g. switching matrix, connection or expansion network (G06F13/4009 takes precedence)}
- G06F 2213/0026: PCI express
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Large-scale AI and HPC systems are expected to increasingly deploy heterogeneous accelerators interconnected via high-bandwidth fabrics built on IEEE 802.3 PMA. Implementations disclose a processor comprising processing cores coupled via an interconnect, an interconnect component coupled to the interconnect, an RPU coupled to the interconnect component, a physical layer based on IEEE 802.3 PMA coupled to the RPU for communicating with an external entity, and a CXL device implemented in at least one of the RPU or the interconnect component. At least one of the RPU or the interconnect component translates between CXL transactions associated with data communicated with the external entity via the physical layer and interconnect transactions communicated via the interconnect. In other implementations, a CXL root port is coupled to the interconnect, and at least one of the RPU or the CXL root port translates between the CXL transactions and interconnect transactions communicated via the interconnect.

## Claims

### Claim 1

An apparatus comprising:
a processor comprising processing cores coupled via an interconnect;
an interconnect component coupled to the interconnect;
a resource provisioning unit (RPU) coupled to the interconnect component;
a physical layer based on IEEE 802.3 physical medium attachment (PMA), coupled to the RPU, configured to communicate with an external entity;
a Compute Express Link (CXL) device implemented in at least one of the RPU or the interconnect component; and
wherein at least one of the RPU or the interconnect component is configured to translate between CXL transactions and interconnect transactions; wherein the CXL transactions are associated with data communicated with the external entity via the physical layer based on IEEE 802.3 PMA, and the interconnect transactions are communicated via the interconnect.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the interconnect component comprises an ARM CXL/CCIX Gateway (CCG).

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein the interconnect comprises a mesh interconnect, the mesh interconnect comprising crosspoints (XPs) configured to route interconnect transactions between the processing cores, the interconnect component, and memory controllers based on packet identifiers.

### Claim 4 (depends on 1)

The apparatus of claim 1, further comprising an I/O-coherent Request Node with Distributed Virtual Memory support (RN-D) coupled to the interconnect, the RN-D configured to handle CXL.io or non-coherent traffic between the CXL device and the processing cores.

### Claim 5 (depends on 1)

The apparatus of claim 1, further comprising Subordinate Node (SN-F) nodes coupled to memory controllers, the memory controllers coupled to DRAM via DDR PHY and memory channels.

### Claim 6 (depends on 1)

The apparatus of claim 1, wherein the interconnect component comprises a CXL Streaming (CXS) interface and a Coherent Multichip Link (CML) gateway or a Cache Coherent Interconnect for Accelerators (CCIX) Gateway (CXG) that utilizes the CXS interface.

### Claim 7 (depends on 1)

The apparatus of claim 1, wherein the interconnect component is configured to initiate a CHI allocating ReadShared request to a Home Node in response to a CXL.mem request, wherein the Home Node is coupled to a memory controller.

### Claim 8 (depends on 7)

The apparatus of claim 7, wherein the Home Node is configured to send a ReadNoSnp request to the memory controller, and the memory controller is configured to return data to the interconnect component using a CompData response.

### Claim 9 (depends on 1)

The apparatus of claim 1, wherein the CXL device comprises a Global Fabric-Attached Memory Device (GFD) supporting CXL.mem transactions, and the interconnect component is optimized for handling CXL.mem traffic.

### Claim 10 (depends on 1)

The apparatus of claim 1, wherein the data communicated with the external entity via the physical layer based on IEEE 802.3 PMA comprises protocol data units (PDUs) of a carrier protocol encapsulating CXL PDUs, and the RPU is configured to extract CXL PDUs from the carrier protocol PDUs.

### Claim 11 (depends on 1)

The apparatus of claim 1, wherein the external entity comprises at least one of a GPU, an accelerator, or a switch, and the apparatus is configured to enable the external entity to access memory coupled to memory controllers of the processor via the RPU and the interconnect component.

### Claim 12

A method comprising:
receiving, via a physical layer based on IEEE 802.3 physical medium attachment (PMA), data from an external entity;
translating, utilizing a Compute Express Link (CXL) device implemented in at least one of a resource provisioning unit (RPU) or an interconnect component coupled to an interconnect of a processor, between CXL transactions and interconnect transactions communicated via the interconnect; and
wherein the processor comprises processing cores coupled via the interconnect, and the CXL transactions are associated with the data received from the external entity via the physical layer based on IEEE 802.3 PMA.

### Claim 13 (depends on 12)

The method of claim 12, wherein the data comprises protocol data units (PDUs) of a carrier protocol encapsulating CXL PDUs, and the translating comprises extracting CXL PDUs from the carrier protocol PDUs and reconstructing CXL transactions from the extracted CXL PDUs.

### Claim 14 (depends on 12)

An active cable comprising first and second pluggable modules coupled by a physical medium, wherein the active cable further comprises hardware circuitry, integrated into the active cable, configured to perform the method of claim 12.

### Claim 15 (depends on 12)

An apparatus configured to operate as a switch, wherein the apparatus comprises switching circuitry and is configured to perform the method of claim 12.

### Claim 16

An apparatus comprising:
a processor comprising processing cores coupled via an interconnect;
a Compute Express Link (CXL) root port coupled to the interconnect;
a resource provisioning unit (RPU) coupled to the CXL root port;
a physical layer based on IEEE 802.3 physical medium attachment (PMA), coupled to the RPU, configured to communicate with an external entity; and
wherein at least one of the RPU or the CXL root port is configured to translate between CXL transactions and interconnect transactions; wherein the CXL transactions are associated with data communicated with the external entity via the physical layer based on IEEE 802.3 PMA, and the interconnect transactions are communicated via the interconnect.

### Claim 17 (depends on 16)

The apparatus of claim 16, wherein the CXL root port is coupled to a fully coherent request node (RN-F) and a fully coherent home node (HN-F) on the interconnect.

### Claim 18 (depends on 17)

The apparatus of claim 17, wherein the RN-F enables the external entity to access memory coupled to memory controllers of the processor via the physical layer based on IEEE 802.3 PMA, the RPU, the CXL root port, and the RN-F.

### Claim 19 (depends on 17)

The apparatus of claim 17, wherein the HN-F enables the processing cores to access a resource coupled to the external entity via the interconnect, the CXL root port, the RPU, and the physical layer based on IEEE 802.3 PMA.

### Claim 20 (depends on 16)

The apparatus of claim 16, wherein the CXL root port is coupled to the interconnect via a CXL/CCIX Gateway (CCG) and an I/O-coherent Request Node with Distributed Virtual Memory support (RN-D).

### Claim 21 (depends on 20)

The apparatus of claim 20, wherein the CCG is optimized for handling CXL.mem traffic.

### Claim 22 (depends on 20)

The apparatus of claim 20, wherein the RN-D is configured to handle CXL.io or PCIe traffic communicated via the CXL root port.

### Claim 23 (depends on 16)

The apparatus of claim 16, further comprising Subordinate Nodes (SN-F) coupled to memory controllers, the memory controllers coupled to DRAM via DDR PHY and memory channels.

### Claim 24 (depends on 16)

The apparatus of claim 16, wherein the data communicated with the external entity via the physical layer based on IEEE 802.3 PMA comprises protocol data units (PDUs) of a carrier protocol encapsulating CXL PDUs, and the RPU is configured to extract CXL PDUs from the carrier protocol PDUs.

### Claim 25 (depends on 16)

The apparatus of claim 16, wherein the external entity comprises at least one of a GPU, an accelerator, or a switch, and the apparatus is configured to enable the external entity to access memory coupled to memory controllers of the processor via the RPU, the CXL root port, and the interconnect.

### Claim 26 (depends on 16)

The apparatus of claim 16, wherein the interconnect comprises a mesh interconnect, the mesh interconnect comprising crosspoints (XPs) configured to route interconnect transactions between the processing cores, the CXL root port, and memory controllers based on packet identifiers.

### Claim 27

A method comprising:
receiving, via a physical layer based on IEEE 802.3 physical medium attachment (PMA), data from an external entity;
translating, by at least one of a resource provisioning unit (RPU) or a Compute Express Link (CXL) root port coupled to an interconnect of a processor, between CXL transactions and interconnect transactions;
wherein the processor comprises processing cores coupled via the interconnect, the CXL transactions are associated with the data received from the external entity via the physical layer based on IEEE 802.3 PMA, and the interconnect transactions are communicated via the interconnect.

### Claim 28 (depends on 27)

The method of claim 27, wherein the CXL root port is coupled to a fully coherent request node (RN-F) and a fully coherent home node (HN-F) on the interconnect, the RN-F enabling the external entity to access memory coupled to memory controllers of the processor, and the HN-F enabling the processing cores to access a resource coupled to the external entity.

### Claim 29 (depends on 27)

An active cable comprising first and second pluggable modules coupled by a physical medium, wherein the active cable further comprises hardware circuitry, integrated into the active cable, configured to perform the method of claim 27.

### Claim 30 (depends on 27)

An apparatus configured to operate as a switch, wherein the apparatus comprises switching circuitry and is configured to perform the method of claim 27.
