# US 2024/0202025 A1 · Hybrid virtual gpu co-scheduling

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

- Number: US 2024/0202025 A1
- Document ID: US20240202025A1
- Status: Published application
- Kind: A1
- Application number: 18/394,232
- Filed: 2023-12-22
- Published: 2024-06-20
- Earliest claimed: 2020-11-24
- Assignees: Intel Corporation
- Applicants: Intel Corporation
- Inventors: Yan Zhao; Zhi Wang; Weinan Li
- Web page: https://patentgrep.com/patent/US20240202025A1

## CPC

- G06F 9/4881: {Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues}
- G06F 9/5027: {the resource being a machine, e.g. CPUs, Servers, Terminals}
- G06F 9/3004: {to perform operations on memory}
- G06F 9/30079: {Pipeline control instructions, e.g. multicycle NOP}
- G06F 9/505: {considering the load}
- G06F 9/5077: {Logical partitioning of resources; Management or configuration of virtualized resources (specific details on emulation or internal functioning of virtual machines G06F9/455)}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

An embodiment of a semiconductor package apparatus may include technology to manage one or more virtual graphic processor units, and co-schedule the one or more virtual graphic processor units based on both general processor instructions and graphics processor instructions. Other embodiments are disclosed and claimed.

## Claims

### Claim 1

(canceled)

### Claim 2

A system, comprising:
a memory that stores a workload queue associated with a virtual graphics processor unit (vPGU); and
logic communicatively coupled to the memory to:
identify a semaphore associated with a graphics processing unit,
determine if the graphics processing unit is active, and
if the graphics processing unit is determined to be active, append a workload associated with the semaphore to the workload queue.

### Claim 3 (depends on 2)

The system of claim 2, wherein the logic is further to:
release the semaphore if the graphics processing unit is determined to be active.

### Claim 4 (depends on 2)

The system of claim 2, wherein the logic is further to:
if the graphics processing unit is idle, load the workload into a hardware execution queue associated with the graphics processing unit.

### Claim 5 (depends on 2)

The system of claim 2, wherein the workload queue is stored in a general memory space that is accessible by a central processing unit.

### Claim 6 (depends on 5)

The system of claim 5, wherein the logic is further to:
access, with the graphics processing unit and the central processing unit, the workload queue based on a general graphics translation table.

### Claim 7 (depends on 6)

The system of claim 6, wherein the logic is further to:
map the workload queue and scheduling accounting information associated with scheduling policies implemented by graphics processor commands into the general graphics translation table.

### Claim 8 (depends on 7)

The system of claim 7, wherein the logic is further to:
share the scheduling accounting information and the workload queue between the central processing unit and the graphics processing unit with the general graphics translation table.

### Claim 9

A semiconductor package apparatus, comprising:
one or more substrates; and
logic coupled to the one or more substrates, wherein the logic is at least partly implemented in one or more of configurable logic or fixed-functionality hardware logic, the logic coupled to the one or more substrates to:
identify a semaphore associated with a graphics processing unit,
determine if the graphics processing unit is active, and
if the graphics processing unit is determined to be active, append a workload associated with the semaphore to a workload queue associated with a virtual graphics processor unit (vPGU).

### Claim 10 (depends on 9)

The apparatus of claim 9, wherein the logic coupled to the one or more substrates is further to:
release the semaphore if the graphics processing unit is determined to be active.

### Claim 11 (depends on 9)

The apparatus of claim 9, wherein the logic coupled to the one or more substrates is further to:
if the graphics processing unit is idle, load the workload into a hardware execution queue associated with the graphics processing unit.

### Claim 12 (depends on 9)

The apparatus of claim 9, wherein the workload queue is stored in a general memory space that is accessible by a central processing unit.

### Claim 13 (depends on 12)

The apparatus of claim 12, wherein the logic coupled to the one or more substrates is further to:
access, with the graphics processing unit and the central processing unit, the workload queue based on a general graphics translation table.

### Claim 14 (depends on 13)

The apparatus of claim 13, wherein the logic coupled to the one or more substrates is further to:
map the workload queue and scheduling accounting information associated with scheduling policies implemented by graphics processor commands into the general graphics translation table.

### Claim 15 (depends on 14)

The apparatus of claim 14, wherein the logic coupled to the one or more substrates is further to:
share the scheduling accounting information and the workload queue between the central processing unit and the graphics processing unit with the general graphics translation table.

### Claim 16

A method comprising:
identifying a semaphore associated with a graphics processing unit;
determining if the graphics processing unit is active; and
if the graphics processing unit is determined to be active, appending a workload associated with the semaphore to a workload queue associated with a virtual graphics processor unit (vPGU).

### Claim 17 (depends on 16)

The method of claim 16, further comprising:
releasing the semaphore if the graphics processing unit is determined to be active.

### Claim 18 (depends on 16)

The method of claim 16, further comprising:
if the graphics processing unit is idle, loading the workload into a hardware execution queue associated with the graphics processing unit.

### Claim 19 (depends on 16)

The method of claim 16, wherein the workload queue is stored in a general memory space that is accessible by a central processing unit.

### Claim 20 (depends on 19)

The method of claim 19, further comprising:
accessing, with the graphics processing unit and the central processing unit, the workload queue based on a general graphics translation table.

### Claim 21 (depends on 20)

The method of claim 20, further comprising:
mapping the workload queue and scheduling accounting information associated with scheduling policies implemented by graphics processor commands into the general graphics translation table; and
sharing the scheduling accounting information and the workload queue between the central processing unit and the graphics processing unit with the general graphics translation table.
