Read-only communication operator

Patent metadata

Field	Value
Publication number	US-9430204-B2
Application number	US-94990810-A
Country	US
Kind code	B2
Filing date	Nov 19, 2010
Priority date	Nov 19, 2010
Publication date	Aug 30, 2016
Grant date	Aug 30, 2016

Abstract

A high level programming language provides a read-only communication operator that prevents a computational space from being written. An indexable type with a rank and element type defines the computational space. For an input indexable type, the read-only communication operator produces an output indexable type with the same rank and element type as the input indexable type but ensures that the output indexable type may not be written. The read-only communication operator ensures that any attempt to write to the output indexable type will be detected as an error at compile time.

First claim

What is claimed is:

1. A computer readable storage device storing computer-executable instructions that, when executed by a computer system, cause the computer system to perform a method comprising:
applying a read-only communication operator to an input indexable type to form an immutable computational space, the read-only communication operator applied in a data parallel source code configured for execution on one or more data parallel compute nodes;
generating an output indexable type from said application, the output indexable type is a data type forming an immutable computational space; and
performing a data parallel algorithm using the output indexable type,
wherein the read-only communication operator is based on a template operator that defines one or more of constant accessors, index operators, or subscript operators for the output indexable type,
wherein the read-only communication operator includes a template-enclosed code portion that checks a Parent type and range of the input indexable type using a conditional operator that returns a read-only Parent range, wherein the template-enclosed code portion optionally includes an internal casting of the Parent type to prevent a cast inadvertently made to a lvalue-reference of the input indexable type that might occur outside of the template-enclosed code portion, and
in case the input indexable type returns its index operators and subscript operators as by-lvalue-references, a template-based conditional base class is implemented to enforce the corresponding return type as a rvalue-reference or a by-value depending on the Parent type.

2. The computer readable storage device of claim 1 , wherein the input indexable type has a rank and an element type, and wherein the output indexable type has the rank and the element type.

3. The computer readable storage device of claim 1 , wherein the data parallel algorithm is a visual shader.

4. The computer readable storage device of claim 1 , wherein the read-only communication operator is applied in an invocation of a vector function.

5. The computer readable storage device of claim 4 , wherein the read-only communication operator is applied to a parameter in an invocation of a kernel function.

6. The computer readable storage device of claim 1 , wherein the data parallel source code is written in a programming language with data parallel extensions.

7. The computer readable storage device of claim 1 , wherein the data parallel source code is written in a high level data parallel programming language.

8. The computer readable storage device of claim 1 , wherein the one or more data parallel compute nodes include at least one graphics processing unit.

9. The computer readable storage device of claim 1 , wherein the one or more data parallel compute nodes include at least one general purpose processor.

10. A method implemented on a computing device having at least one processor, the method comprising:
identifying a read-only communication operator in data parallel source code configured for execution on one or more data parallel compute nodes, the read-only communication operator used to form an immutable computational space; and
generating data parallel executable code from the data parallel source code such that the data parallel executable code implements read-only communication operator by generating an output indexable type from an input indexable type, the output indexable type forming the immutable computational space in memory,
wherein the read-only communication operator is based on a template operator that defines one or more of constant accessors, index operators, and subscript operators for the output indexable type,
wherein the read-only communication operator includes a template-enclosed code portion that checks a Parent type and range of the input indexable type using a conditional operator that returns a read-only Parent range, wherein the template-enclosed code portion optionally includes an internal casting of the Parent type to prevent a cast inadvertently made to a lvalue-reference of the input indexable type that might occur outside of the template-enclosed code portion, and
in case the input indexable type returns its index operators and subscript operators as by-lvalue-references, a template-based conditional base class is implemented to enforce the corresponding return type as a rvalue-reference or a by-value depending on the Parent type.

11. The method of claim 10 wherein the input indexable type has a rank and an element type, and wherein the output indexable type has the rank and the element type.

12. The method of claim 10 further comprising:
generating an error in response to detecting a write to the output indexable type in the data parallel source.

13. The method of claim 10 , wherein the read-only communication operator is applied to a parameter in an invocation of a kernel function.

14. The method of claim 10 , wherein the read-only communication operator is applied in an invocation of a vector function.

15. The method of claim 10 wherein the data parallel source code is written in a high level general purpose programming language with data parallel extensions.

16. The method of claim 10 wherein the data parallel source code is written in a high level data parallel programming language.

17. The method of claim 10 wherein the one or more data parallel compute nodes include at least one graphics processing unit.

18. The method of claim 10 wherein the one or more data parallel compute nodes include at least one general purpose processor.

19. A computer readable storage device storing computer-executable instructions that, when executed by at least one processor, perform a method comprising:
applying a read-only communication operator to form an immutable computational space by generating an output indexable type with a rank and an element type from an input indexable type with the rank and the element type; and
performing a data parallel algorithm using the output indexable type;
wherein the read-only communications operator is included in data parallel source code written in a high level general purpose programming language with data parallel extensions,
wherein the read-only communication operator includes a template-enclosed code portion that checks a Parent type and range of the input indexable type using a conditional operator that returns a read-only Parent range, wherein the template-enclosed code portion optionally includes an internal casting of the Parent type to prevent a cast inadvertently made to a lvalue-reference of the input indexable type that might occur outside of the template-enclosed code portion,
in case the input indexable type returns its index operators and subscript operators as by-lvalue-references, a template-based conditional base class is implemented to enforce the corresponding return type as a rvalue-reference or a by-value depending on the Parent type,
wherein the output indexable type is a data type that forms an immutable computational space in memory, and
wherein the read-only communication operator defines one or more of constant accessors, index operators, or subscript operators for the output indexable type.

20. The computer readable storage device of claim 19 , wherein the read-only communication operator is applied in an invocation of a vector function.

Assignees

• Levanoni Yosseff
• Ringseth Paul F
• Zhu Weirong
• Zhang Lingli
• Microsoft Technology Licensing Llc

Inventors

• Levanoni Yosseff
• Ringseth Paul F
• Zhu Weirong
• Zhang Lingli

Classifications

• G06F8/458 Primary

  Synchronisation, e.g. post-wait, barriers, locks (synchronisation among tasks G06F9/52) · CPC title

• G06F8/314 Primary

  Parallel programming languages (G06F8/313 takes precedence) · CPC title