Systems and methods for task parallelization

• US 9,836,305 B1
• Filed: 11/10/2015
• Issued: 12/05/2017
• Est. Priority Date: 03/18/2015
• Status: Active Grant

• Current Assignee: Finastra Global Ltd.
• Original Assignee: Misys Global Limited
• Inventors: Leenhouwers, Bram, Gabreau, Edouard, Grimaud, Meggie, Antonov, Alexis
• Primary Examiner(s): Sun, Charlie
• Application Number:
  US14/936,925
• Time in Patent Office: 756 Days
• CPC Class Codes: G06F 8/427 Parsing
• CPC Class Codes: G06F 8/443 Optimisation
• CPC Class Codes: G06F 8/456 Parallelism detection
• CPC Class Codes: G06F 9/5016 the resource being the memory

Abstract

Various embodiments of the present disclosure can include systems, methods, and non-transitory computer readable media configured to obtain at least one script and at least one document, wherein the script includes one or more instructions to be translated for execution in a parallelized computing environment, and wherein the document includes data that is referenced by the script. A syntax tree for the script can be determined. At least one approach for optimizing the syntax tree can be applied. Parallelized code for execution in the parallelized computing environment can be generated. A binary representation of the document can be determined. The parallelized code can be processed based at least in part on the binary representation of the document.

Claims

1. A computer-implemented method comprising:

obtaining, by a computing system, at least one script and at least one document, wherein the script includes one or more instructions to be translated for execution in a parallelized computing environment, and wherein the document includes data that is referenced by the script;

determining, by the computing system, a syntax tree for the script;

applying, by the computing system, at least one approach for optimizing the syntax tree;

generating, by the computing system, parallelized code for execution in the parallelized computing environment;

determining, by the computing system, a binary representation of the document;

pre-computing, by the computing system, a result of at least one instruction included in the script;

in response to the pre-computing, storing, by the computing system, the result in the binary representation of the document;

in response to the pre-computing, updating, by the computing system, the syntax tree to replace a reference to the instruction with a reference to the pre-computed result; and

processing, by the computing system, the parallelized code based at least in part on the binary representation of the document.

2. The computer-implemented method of claim 1, wherein the script is written in Groovy programming language, and wherein the determining, by the computing system, the syntax tree for the script further comprises:

obtaining, by the computing system, the syntax tree using an application programming interface (API) of the Groovy programming language.

3. The computer-implemented method of claim 1, wherein the applying, by the computing system, the at least one approach for optimizing the syntax tree further comprises:

determining, by the computing system, at least one inferred type for at least one variable referenced in the syntax tree.

4. The computer-implemented method of claim 3, wherein the determining, by the computing system, the inferred type for the at least one variable further comprises:

determining, by the computing system, that the variable corresponds to a binary statement that includes a minus operator; and

determining, by the computing system, that the variable is one of a “

double”

type, a “

date”

type;

a string type, a Boolean type, custom structured data, an array of some type, a matrix of some type, or a cube of some type.

5. The computer-implemented method of claim 3, wherein the determining, by the computing system, the inferred type for the at least one variable further comprises:

determining, by the computing system, that the variable corresponds to a binary statement that includes a multiplication operator; and

determining, by the computing system, that the variable is one of a “

double”

type, a “

date”

type, a string type, a Boolean type, custom structured data, an array of some type, a matrix of some type, or a cube of some type.

6. The computer-implemented method of claim 3, wherein the determining, by the computing system, the inferred type for the at least one variable further comprises:

determining, by the computing system, that the at least one variable corresponds to an assignment statement that references a second variable; and

determining, by the computing system, that the at least one variable corresponds to a same type as the second variable.

7. The computer-implemented method of claim 1, wherein the applying, by the computing system, the at least one approach for optimizing the syntax tree further comprises:

replacing, by the computing system, at least one call in the script with a call to a raw function corresponding to the parallelized code.

8. The computer-implemented method of claim 1, wherein the parallelized code corresponds to a OpenCL parallelization framework, a Compute Unified Device Architecture (CUDA) parallelization framework, or Java bytecode.

9. The computer-implemented method of claim 1, wherein the generating, by the computing system, the parallelized code further comprises:

determining, by the computing system, the parallelized code by at least transforming the optimized syntax tree to a parallelization framework code.

10. The computer-implemented method of claim 1, wherein the determining, by the computing system, the binary representation of the document further comprises:

determining, by the computing system, a dictionary corresponding to the document, wherein the dictionary stores the data included in the document, and wherein data stored in the dictionary is ordered to optimize memory usage.

11. The computer-implemented method of claim 10, wherein data corresponding to a string value type is translated to a first number of bits, wherein data corresponding to a numerical value type is translated to a second number of bits, and wherein data corresponding to a date value type is translated to a third number of bits.

12. The computer-implemented method of claim 1, wherein the determining, by the computing system, the binary representation of the document further comprises:

pre-computing, by the computing system, an amount of memory needed for the document, wherein information indicating the amount of memory is stored in the binary representation of the document.

13. The computer-implemented method of claim 12, the method further comprising:

allocating, by the computing system, memory prior to executing the parallelized code based at least in part on the information indicating the amount of memory needed for the document.

14. The computer-implemented method of claim 1, the method further comprising:

obtaining, by the computing system, at least one second document, wherein the second document includes the data that is referenced by the script;

determining, by the computing system, a binary representation of the second document; and

processing, by the computing system, the parallelized code based at least in part on the binary representation of the second document.

15. The computer-implemented method of claim 1, the method further comprising:

obtaining, by the computing system, at least one second document, wherein the second document includes the data that is referenced by the script;

determining, by the computing system, a binary representation of the second document;

determining;

by the computing system, an interlaced binary structure based at least in part on the binary representation of the document and the binary representation of the second document; and

processing, by the computing system, the parallelized code with respect to the document and the second document based at least in part on the interlaced binary representation.

16. The computer-implemented method of claim 15, the method further comprising:

allocating, by the computing system, memory for the document prior to executing the parallelized code;

allocating, by the computing system, memory for the second document prior to executing the parallelized code, wherein the memory allocated for the document and the memory allocated for the second document is coalesced.

17. A system comprising:

at least one processor; and

a memory storing instructions that, when executed by the at least one processor, cause the system to perform;

obtaining at least one script and at least one document, wherein the script includes one or more instructions to be translated for execution in a parallelized computing environment, and wherein the document includes data that is referenced by the script;

determining a syntax tree for the script;

applying at least one approach for optimizing the syntax tree;

generating parallelized code for execution in the parallelized computing environment;

determining a binary representation of the document;

pre-computing a result of at least one instruction included in the script;

in response to the pre-computing, storing the result in the binary representation of the document;

in response to the pre-computing, updating the syntax tree to replace a reference to the instruction with a reference to the pre-computed result; and

processing the parallelized code based at least in part on the binary representation of the document.

18. A non-transitory computer-readable storage medium including instructions that, when executed by at least one processor of a computing system, cause the computing system to perform:

obtaining at least one script and at least one document, wherein the script includes one or more instructions to be translated for execution in a parallelized computing environment, and wherein the document includes data that is referenced by the script;

determining a syntax tree for the script;

applying at least one approach for optimizing the syntax tree;

generating parallelized code for execution in the parallelized computing environment;

determining a binary representation of the document;

pre-computing a result of at least one instruction included in the script;

in response to the pre-computing, storing the result in the binary representation of the document;

in response to the pre-computing, updating the syntax tree to replace a reference to the instruction with a reference to the pre-computed result; and

processing the parallelized code based at least in part on the binary representation of the document.