Data processing systems

Patent metadata

Field	Value
Publication number	US-10089709-B2
Application number	US-201615208459-A
Country	US
Kind code	B2
Filing date	Jul 12, 2016
Priority date	Jul 15, 2015
Publication date	Oct 2, 2018
Grant date	Oct 2, 2018

Abstract

A graphics processing unit 3 includes a rasterizer 25 , a thread spawner 40 , a programmable execution unit 41 , a varying interpolator 42 , a texture mapper 43 , and a blender 29 . The programmable execution unit 41 is able to communicate with the varying interpolator 42 , the texture mapper 43 and the blender 29 to request processing operations by those graphic specific accelerators. In addition to this, these graphics-specific accelerators are also able to communicate directly with each other and with the thread spawner 40 , independently of the programmable execution unit 41 . This allows for certain graphics processing operations to be performed using direct communication between the graphics-specific accelerators of the graphics processing unit, instead of executing instructions in the programmable execution unit to trigger the performance of those operations by the graphics-specific accelerators.

First claim

What is claimed is:

1. A processing unit comprising:
a programmable execution unit operable to execute programs to perform processing operations; and
a plurality of accelerators each operable to perform respective processing operations; wherein:
the programmable execution unit is operable to communicate with the accelerators to cause the accelerators to perform processing operations, and to receive responses from the accelerators;
two or more of the accelerators are operable to communicate directly with each other, in order to trigger processing operations by at least one of the accelerators; and further wherein:
the processing unit is configured to determine for a processing entity or group of processing entities to be processed by the processing unit, whether that entity or group of entities should be processed solely by executing a program using the programmable execution unit for that entity or group of entities, or whether that entity or group of entities should be processed solely by means of direct communication between the accelerators, or whether that entity or group of entities should be processed both by executing a program using the programmable execution unit for that entity or group of entities and by means of direct communication between the accelerators; and

the processing unit comprises processing circuitry configured to:
when it is determined that a processing entity or entities is to be processed by executing a program using the programmable execution unit, issue an execution thread or threads for execution by the programmable execution unit in order to trigger processing operations for the entity or entities;
when it is determined that a processing entity or entities is to be processed by means of direct communication between the accelerators, issue a message directly to one or more of the accelerators, in order to trigger processing operations for the entity or entities; and
when it is determined that a processing entity or entities is to be processed both by executing a program using the programmable execution unit for the entity or group of entities and by means of direct communication between the accelerators, issue an execution thread or threads for execution by the programmable execution unit in order to trigger processing operations for the entity or entities, and issue a message directly to one or more of the accelerators, in order to trigger processing operations for the entity or entities.

2. The processing unit of claim 1 , wherein:
the processing unit is a graphics processing unit;
the accelerators include a varying interpolator, a texture mapper and a blender; and
the varying interpolator can communicate directly with the texture mapper to request texturing operations, and/or the texture mapper can communicate directly with the blender to request blending operations.

3. The processing unit of claim 1 , wherein there is a set of programs that are predefined as being suitable to be performed using direct communication between the accelerators.

4. The processing unit of claim 1 , wherein:
the processing entity or entities comprise a graphics fragment or fragments; and
the determination is based on one or more of:
whether the fragment to be processed has full coverage;
whether the fragment to be processed has an allocated tile buffer; and
whether the fragment to be processed is the oldest at its position in the render output.

5. The processing unit of claim 1 , wherein the programmable execution unit is operable to be able to issue a message or messages to one or more of the accelerators that will then cause the accelerators to communicate directly with each other to perform processing operations.

6. The processing unit of claim 5 , wherein the programmable execution unit is operable to issue a message or messages to one or more of the accelerators that will then cause the accelerators to communicate directly with each other to perform processing operations in response to an instruction in a program that it is executing.

7. A method of performing processing operations in a processing unit that comprises:
a programmable execution unit operable to execute programs to perform processing operations; and
a plurality of accelerators each operable to perform respective processing operations; and in which:
the programmable execution unit is operable to communicate with the accelerators to cause the accelerators to perform processing operations, and to receive responses from the accelerators; and
two or more of the accelerators are operable to communicate directly with each other, in order to trigger processing operations by at least one of the accelerators;
the method comprising:
the processing unit:
determining for a processing entity or group of processing entities to be processed by the processing unit, whether that entity or group of entities should be processed solely by executing a program using the programmable execution unit for that entity or group of entities, or whether that entity or group of entities should be processed solely by means of direct communication between the accelerators, or whether that entity or group of entities should be processed both by executing a program using the programmable execution unit for that entity or group of entities and by means of direct communication between the accelerators;
when it is determined that a processing entity or entities is to be processed by executing a program using the programmable execution unit, issuing an execution thread or threads for execution to the programmable execution unit in order to trigger processing operations for the entity or entities;
when it is determined that a processing entity or entities is to be processed by means of direct communication between the accelerators, issuing a message directly to one or more of the accelerators, in order to trigger processing operations for the entity or entities; and
when it is determined that a processing entity or entities is to be processed, both by executing a program using the programmable execution unit for that entity or group of entities and by means of direct communication between the accelerators, issuing an execution thread or threads for execution to the programmable execution unit in order to trigger processing operations for the entity or entities, and issuing a message directly to one or more of the accelerators in order to trigger processing operations for the entity or entities.

8. The method of claim 7 , wherein:
the processing unit is a graphics processing unit;
the accelerators include a varying interpolator, and a texture mapper; and
the method comprises:
the varying interpolator communicating directly with the texture mapper to request texturing operations.

9. The method of claim 7 , wherein:
the processing unit is a graphics processing unit;
the accelerators include a texture mapper and a blender; and
the method comprises:
the texture mapper communicating directly with the blender to request blending operations.

10. The method of claim 7 , wherein there is a set of programs that are predefined as being suitable to be performed using direct communication between the accelerators.

11. The method of claim 7 , wherein:
the processing entity or entities comprise a graphics fragment or fragments; and
the determination is based on one or more of:
whether the fragment to be processed has full coverage;
whether the fragment to be processed has an allocated tile buffer; and
whether the fragment to be processed is the oldest at its position in the render output.

12. The method of claim 7 , comprising the programmable execution unit issuing a message to one of the accelera

Assignees

• Advanced Risc Mach Ltd

Inventors

• Engh-Halstvedt Andreas Due
• Bermingham David James
• Kleen Amir
• Nystad Jørn
• Østby Kenneth Edvard

Classifications

• G06T15/005

  General purpose rendering architectures · CPC title

• G06T1/20 Primary

  Processor architectures; Processor configuration, e.g. pipelining · CPC title

• G06F9/3877

  using a secondary processor, e.g. coprocessor (peripheral processor G06F13/12) · CPC title

• G06T17/10

  Constructive solid geometry [CSG] using solid primitives, e.g. cylinders, cubes · CPC title

• G06T15/04

  Texture mapping · CPC title