# US 2021/0263853 A1 · Multi-tile memory management mechanism

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

- Number: US 2021/0263853 A1
- Document ID: US20210263853A1
- Status: Published application
- Kind: A1
- Application number: 16/802,427
- Filed: 2020-02-26
- Published: 2021-08-26
- Earliest claimed: 2020-02-26
- Assignees: Intel Corporation
- Applicants: Intel Corporation
- Inventors: ZACK S. WATERS; TRAVIS SCHLUESSLER; MICHAEL APODACA; ANKUR SHAH
- Web page: https://patentgrep.com/patent/US20210263853A1

## CPC

- G06F 12/0882: Page mode
- G06F 9/4411: {Configuring for operating with peripheral devices; Loading of device drivers}
- G06F 9/5016: {the resource being the memory}
- G06F 9/544: {Buffers; Shared memory; Pipes}
- G06F 11/3006: {where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems (multiprogramming arrangements G06F9/46; allocation of resources G06F9/50)}
- G06F 11/3037: {where the computing system component is a memory, e.g. virtual memory, cache (accessing, addressing or allocating within memory systems or architectures G06F12/00; checking stores for correct operation G11C29/00)}
- G06F 12/0607: {Interleaved addressing}
- G06F 12/0837: with software control, e.g. non-cacheable data
- G06F 12/1054: {the data cache being concurrently physically addressed}
- G06F 12/1063: {the data cache being concurrently virtually addressed}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Graphics processors for implementing multi-tile memory management are disclosed. In one embodiment, a graphics processor includes a first graphics device having a local memory, a second graphics device having a local memory, and a graphics driver to provide a single virtual allocation with a common virtual address range to mirror a resource to each local memory of the first and second graphics devices.

## Claims

### Claim 1

A graphics processor for a multi-tile architecture, comprising:
a first graphics device having a local memory;
a second graphics device having a local memory; and
a graphics driver to provide a single virtual allocation with a common virtual address range to mirror a resource to each local memory of the first and second graphics devices.

### Claim 2 (depends on 1)

The graphics processor of claim 1, wherein the single virtual allocation includes a first page table for the first graphics device and a second page table for the second graphics device with the first and second page tables providing a unified physical address space.

### Claim 3 (depends on 2)

The graphics processor of claim 2, wherein the first graphics device is communicatively coupled to the second graphics device and each graphics device comprises a graphics tile of the multi-tile architecture for a process.

### Claim 4 (depends on 2)

The graphics processor of claim 2, wherein the single virtual allocation to mirror the resource including physical pages for each local memory of the first and second graphics device.

### Claim 5 (depends on 1)

The graphics processor of claim 1, wherein the mirrored resource comprises a read only resource.

### Claim 6 (depends on 1)

The graphics processor of claim 1, wherein the first graphics device comprises a graphics processing unit.

### Claim 7 (depends on 1)

The graphics processor claim 1, wherein the graphics driver comprises a kernel mode graphics driver.

### Claim 8 (depends on 7)

The graphics processor claim 7, further comprising:
a user mode graphics driver to provide private data with an allocation creation request to the kernel mode graphics driver to indicate whether the resource needs to be mirrored or not along with a node mask to indicate which graphics devices will be accessing the resource.

### Claim 9

A graphics processor for a multi-tile architecture, comprising:
a first graphics device having a local memory;
a second graphics device having a local memory; and
a graphics driver to provide a single virtual allocation with a common virtual address range to interleave physical pages of a shared resource to local memory of the first and second graphics devices.

### Claim 10 (depends on 9)

The graphics processor of claim 9, wherein the single virtual allocation interleaves a first physical page to the local memory of the first graphics device and a second physical page to the local memory of the second graphics device.

### Claim 11 (depends on 9)

The graphics processor of claim 9, wherein the single virtual allocation interleaves a first subset of a physical page to the local memory of the first graphics device and a second subset of the physical page to the local memory of the second graphics device.

### Claim 12 (depends on 9)

The graphics processor of claim 9, wherein the first graphics device is communicatively coupled to the second graphics device.

### Claim 13 (depends on 9)

The graphics processor of claim 9, wherein the shared resource comprises a shared read/write buffer.

### Claim 14 (depends on 9)

The graphics processor of claim 9, further comprising:
a third graphics device having a local memory; and
a fourth graphics device having a local memory, wherein the graphics driver comprises a kernel mode graphics driver to form a first group including the first and third graphics devices for rendering frame N of a display device based on a node mask.

### Claim 15 (depends on 14)

The graphics processor of claim 14, wherein the kernel mode graphics driver to form a second group including the second and the fourth graphics devices for rendering frame N+1 of the display device based on the node mask.

### Claim 16 (depends on 9)

The graphics processor of claim 9, wherein the first graphics device comprises a graphics tile of the multi-tile hierarchy.

### Claim 17 (depends on 9)

The graphics processor of claim 9, wherein the single virtual allocation with a common virtual address range to partition a render target on a per graphics device basis.

### Claim 18 (depends on 17)

The graphics processor of claim 17, wherein the single virtual allocation to allocate a first page of the render target to the local memory of the first graphics device and to allocate a second page of the render target to the local memory of the second graphics device.

### Claim 19

A graphics processor for a multi-tile architecture, comprising:
a first graphics device having a local memory;
a second graphics device having a local memory; and
a graphics driver to provide a heap with a common virtual address range for mapping virtual addresses of at least one resource into the heap and to interleave the at least one resource into local memory of the first and second graphics devices.

### Claim 20 (depends on 19)

The graphics processor of claim 19, wherein the heap interleaves a first physical page to the local memory of the first graphics device and a second physical page to the local memory of the second graphics device.

### Claim 21 (depends on 19)

The graphics processor of claim 19, wherein the heap interleaves a first subset of a physical page to the local memory of the first graphics device and a second subset of the physical page to the local memory of the second graphics device.

### Claim 22 (depends on 19)

The graphics processor of claim 19, wherein the at least one resource comprises a first resource including a texture and a second resource including a buffer.

### Claim 23

A graphics processor, comprising:
a first graphics device having a local memory;
a second graphics device having a local memory;
a communication link to couple the first and second graphics devices; and
a memory management unit is configured to utilize a single virtual allocation with a common virtual address range to mirror a resource or to interleave physical pages of the resource to the local memory of the first and second graphics devices.

### Claim 24 (depends on 23)

The graphics processor of claim 23, wherein the memory management unit is further configured with a monitoring feature to monitor accesses to the communication link, the local memory of the first graphics device, the local memory of the second graphics device, to allocate data from resources to the first and second graphics devices, and to change allocation of the data to the first and second graphics devices based on the monitoring feature.

### Claim 25 (depends on 23)

The graphics processor of claim 23, wherein the single virtual allocation interleaves a first physical page to the local memory of the first graphics device and a second physical page to the local memory of the second graphics device.
