US 10,372,431 B2Granted

Unified intermediate representation

AssigneeApple Inc.

InventorsAaftab Munshi, Rahul U. Joshi, Mon P. Wang, Kelvin C. Chiu

Life of this patent

Estimating the term…

Loading patent…

Abstract

A system decouples the source code language from the eventual execution environment by compiling the source code language into a unified intermediate representation that conforms to a language model allowing both parallel graphical operations and parallel general-purpose computational operations. The intermediate representation may then be distributed to end-user computers, where an embedded compiler can compile the intermediate representation into an executable binary targeted for the CPUs and GPUs available in that end-user device. The intermediate representation is sufficient to define both graphics and non-graphics compute kernels and shaders. At install-time or later, the intermediate representation file may be compiled for the specific target hardware of the given end-user computing system. The CPU or other host device in the given computing system may compile the intermediate representation file to generate an instruction set architecture binary for the hardware target, such as a GPU, within the system.

Claims20

Dependent claims

A non-transitory machine-readable medium, on which are stored instructions, comprising instructions that when executed cause a machine to:

receive a pre-compiled library, wherein the pre-compiled library is a post-compilation intermediate representation of code, the intermediate representation of code being suitable for a subsequent compilation into machine language for a plurality of different processing devices, and wherein the pre-compiled library comprises both graphics operations and non-graphics operations;
compile at least a portion of the pre-compiled library from the intermediate representation into a first binary file comprising both one or more kernels and one or more shaders that are directly executable by a first target processor;
compile at least a portion of the pre-compiled library from the intermediate representation into a second binary file comprising both one or more kernels and one or more shaders that are directly executable by a second target processor, which is different from the first target processor,
wherein at least a first kernel from the first binary file and at least a second kernel from the second binary file are each based upon the compilation of a same first portion of the pre-compiled library, and wherein at least a first shader from the first binary file and at least a second shader from the second binary file are each based upon the compilation of a same second portion of the pre-compiled library;
responsive to detecting a request for the first kernel, retrieve the first kernel from the first binary file for execution by the first target processor;
and
responsive to detecting a request for the second kernel, retrieve the second kernel from the second binary file for execution by the second target processor.

4 more elements

A method comprising:

receiving a pre-compiled library, wherein the pre-compiled library is a post-compilation intermediate representation of code, the intermediate representation of code being suitable for a subsequent compilation into machine language for a plurality of different processing devices, and wherein the pre-compiled library comprises both graphics operations and non-graphics operations;
compiling at least a portion of the pre-compiled library from the intermediate representation into a first binary file comprising both one or more kernels and one or more shaders that are directly executable by a first target processor;
compiling at least a portion of the pre-compiled library from the intermediate representation into a second binary file comprising both one or more kernels and one or more shaders that are directly executable by a second target processor, which is different from the first target processor,
wherein at least a first kernel from the first binary file and at least a second kernel from the second binary file are each based upon the compilation of a same first portion of the pre-compiled library, wherein at least a first shader from the first binary file and at least a second shader from the second binary file are each based upon the compilation of a same second portion of the pre-compiled library;
responsive to detecting a request for the first kernel, retrieving the first kernel from the first binary file for execution by the first target processor;
and
responsive to detecting a request for the second kernel, retrieving the second kernel from the second binary file for execution by the second target processor.

4 more elements

A system for processing graphics comprising:

a CPU;
a memory associated with the CPU for holding instructions for the CPU;
a first target GPU and a second target GPU for producing graphics, wherein the instructions for the CPU, when executed cause the production of graphics by:
receiving a pre-compiled library, wherein the pre-compiled library is a post-compilation intermediate representation of code, the intermediate representation of code being suitable for a subsequent compilation into machine language for a plurality of different processing devices, and wherein the pre-compiled library comprises both graphics operations and non-graphics operations;
compiling at least a portion of the pre-compiled library from the intermediate representation into a first binary file comprising both one or more kernels and one or more shaders that are directly executable by a first target processor;
compiling at least a portion of the pre-compiled library from the intermediate representation into a second binary file comprising both one or more kernels and one or more shaders that are directly executable by a second target processor, which is different from the first target processor,
wherein at least a first kernel from the first binary file and at least a second kernel from the second binary file are each based upon the compilation of a same first portion of the pre-compiled library, wherein at least a first shader from the first binary file and at least a second shader from the second binary file are each based upon the compilation of a same second portion of the pre-compiled library;
responsive to detecting a request for the first kernel, retrieving the first kernel from the first binary file for execution by the first target processor;
and
responsive to detecting a request for the second kernel, retrieving the second kernel from the second binary file for execution by the second target processor.

7 more elements

Dependencies are inferred from the claim text.

Facts

Application: 15/636,200Filed 28 Jun 2017 · continuation
Priority: 30 May 2014
Published: US 2017/0308364 A126 Oct 2017
Granted: 6 Aug 2019
Term ends: Not estimatedThe source records don’t support an estimate.

Classification5 codes

G06F 8/47
Compilation: retargetable compilers
G06F 8/41
Transformation of program code: compilation
G06F 8/44
Compilation: encoding
G06F 8/456
Exploiting coarse grain parallelism in compilation, i.e. parallelism between groups of instructions: parallelism detection

Family

Application62/005,632Us provisional application · filed 30 May 2014
Application14/501,873Parent of this continuation · filed 30 Sep 2014
US 2017/0308364 A1Pre-grant publication · published 26 Oct 2017
US 9,740,464 B2Parent of this continuation
US 2015/0347107 A1Published application
US 10,372,431 B2This patent

From the US filing history. Foreign family members aren’t listed.

Timeline

Citations

Open elsewhere

Building an agent or pulling many patents? Use the API or MCP server instead of scraping pages. Get an API key