US 8,769,510 B2Granted

Identification and translation of program code executable by a graphical processing unit (GPU)

AssigneeThe MathWorks, Inc.

InventorsJocelyn Luke Martin, Joseph F. Hicklin

Life of this patent

Estimating the term…

Loading patent…

Abstract

A device receives program code, and receives size/type information associated with inputs to the program code. The device determines, prior to execution of the program code and based on the input size/type information, a portion of the program code that is executable by a graphical processing unit (GPU), and determines, prior to execution of the program code and based on the input size/type information, a portion of the program code that is executable by a central processing unit (CPU). The device compiles the GPU-executable portion of the program code to create a compiled GPU-executable portion of the program code, and compiles the CPU-executable portion of the program code to create a compiled CPU-executable portion of the program code. The device provides, to the GPU for execution, the compiled GPU-executable portion of the program code, and provides, to the CPU for execution, the compiled CPU-executable portion of the program code.

Claims27

Dependent claims

A method comprising:

receiving, by a computing device, program code;
receiving, by the computing device, an input size and type information associated with inputs to the program code;
performing, by the computing device, a static analysis of the program code to identify portions of the program code;
determining, by the computing device prior to execution of the program code and based on the input size and the type information, a portion of the program code, of the portions of the program code, that is executable by a graphical processing unit (GPU)
determining the portion of the program code that is executable by the GPU comprising:
comparing the portions of the program code to a size threshold;
comparing the portions of the program code to a degree of parallelism threshold;
and
determining the portion of the program code that is executable by the GPU based on comparing the portions of the program code to the size threshold and comparing the portions of the program code to the degree of parallelism threshold;
determining, prior to execution of the program code and based on the input size and the type information, a portion of the program code, of the portions of the program code, that is executable by a central processing unit (CPU)
the portion of the program code that is executable by the CPU being different than the portion of the program code that is executable by the GPU;
compiling, by the computing device, the portion of the program code that is executable by the GPU to create a compiled GPU-executable portion of the program code;
compiling, by the computing device, the portion of the program code that is executable by the CPU to create a compiled CPU-executable portion of the program code;
providing, to the GPU for execution and by the computing device, the compiled GPU-executable portion of the program code;
and
providing, to the CPU for execution and by the computing device, the compiled CPU-executable portion of the program code.

13 more elements

One or more non-transitory computer-readable media storing instructions, the instructions comprising:

one or more instructions which, when executed by a computing device, cause the computing device to receive a program;
one or more instructions which, when executed by the computing device, cause the computing device to receive an input size and type information associated with inputs to the program;
one or more instructions which, when executed by the computing device, cause the computing device to perform a static analysis of the program to identify portions of the program;
one or more instructions which, when executed by the computing device, cause the computing device to determine, based on the input size and the type information, a portion of the program, that is executable by a graphical processing unit (GPU), as a GPU-executable portion of the program and a portion of the program, that is executable by a central processing unit (CPU), as a CPU-executable portion of the program,
the GPU-executable portion of the program being different than the CPU-executable portion of the program,
the one or more instructions to determine the GPU-executable portion and the CPU-executable portion of the program including:
one or more instructions to compare the portions of the program to a size threshold,
one or more instructions to compare the portions of the program to a degree of parallelism threshold, and
one or more instructions to determine the GPU-executable portion and the CPU-executable portion of the program based on comparing the portions of the program to the size threshold and comparing the portions of the program to the degree of parallelism threshold;
one or more instructions which, when executed by the computing device, cause the computing device to compile the GPU-executable portion of the program to create a compiled GPU-executable portion of the program;
one or more instructions which, when executed by the computing device, cause the computing device to compile the CPU-executable portion of the program to create a compiled CPU-executable portion of the program;
one or more instructions which, when executed by the computing device, cause the computing device to provide, to the GPU, the compiled GPU-executable portion of the program;
and
one or more instructions which, when executed by the computing device, cause the computing device to provide, to the CPU, the compiled CPU-executable portion of the program.

11 more elements

A device, comprising:

a memory to store a plurality of instructions;
and
a processor to execute one or more instructions of the plurality of instructions in the memory to:
receive program code created using a technical computing environment,
receive an input size and type information associated with inputs to the program code,
perform a static analysis of the program code to identify portions of the program code,
determine, prior to execution of the program code and based on the input size and the type information, a first portion of the program code to be executed by a graphical processing unit (GPU) and a second portion of the program code to be executed by a central processing unit (CPU)
when determining the first portion of the program code and the second portion of the program code, the processor is to execute instructions in the memory to:
compare the portions of the program code to a size threshold,
compare the portions of the program code to a degree of parallelism threshold, and
determine the first portion of the program code and the second portion of the program code based on comparing the portions of the program code to the size threshold and comparing the portions of the program code to the degree of parallelism threshold,
compile the first portion of the program code, to create a compiled first portion of the program code, and the second portion of the program code to create a compiled second portion of the program code,
provide, to the GPU for execution, the compiled first portion of the program code, and
provide, to the CPU for execution, the compiled second portion of the program code.

11 more elements

Dependencies are inferred from the claim text.

Facts

Application: 12/894,479Filed 30 Sep 2010 · us provisional application
Priority: 8 Apr 2010
Published: US 2011/0252411 A113 Oct 2011
Granted: 1 Jul 2014
Term ends: Not estimatedThe source records don’t support an estimate.

Classification3 codes

G06F 8/41
Transformation of program code: compilation
G06F 8/456
Exploiting coarse grain parallelism in compilation, i.e. parallelism between groups of instructions: parallelism detection
G06F
Computing and data processing