Parallel program generation method
AssigneeAtostek Oy
InventorJuhana Helovuo
Life of this patent
Estimating the term…
Abstract
A method for compiling a source code into a parallel executable form, in which the execution order of the executable is partially undefined. During the compilation process a partial execution order is first defined for instructions having ordering constraints related to the source code level. The partial execution order is then completed with architecture related ordering constraints in order to produce an executable code.
Claims11
A method for compiling a computer program from a source code into an executable code, the method comprising the steps of:
- parsing the source code for defining an abstract parse tree;
- synthesizing said abstract parse tree for defining data flow and data dependency graphs;
- refining said data flow and data dependency graphs into a machine data flow graph in order to form ordering constraints;
- defining a partial execution order with a first portion of the ordering constraints prescribed in the source code, wherein the first portion of the ordering constraints consists of the ordering constraints necessary to satisfy the dependencies in said data flow and data dependency graphs;
- grouping instructions in said machine data flow graph into parallel executable code by adding a second portion of ordering constraints, wherein the second portion of the ordering constraints consists of ordering constraints necessary to map said machine data flow graph to a computation device;
- and
- producing an executable form from said parallel executable code;
- wherein the method further comprises a step of defining a final execution order for said parallel executable code when said parallel executable code is executed in a computing device or in said step of grouping instructions.
A computer program product stored in a memory for compiling a computer program from a source code into an executable code, which computer program product is arranged to execute the following steps when executed in a computing device:
- parsing the source code for defining an abstract parse tree;
- synthesizing said abstract parse tree for defining data flow and data dependency graphs;
- refining said data flow and data dependency graphs into a machine data flow graph in order to form ordering constraints;
- defining a partial execution order with a first portion of the ordering constraints prescribed in the source code, wherein the first portion of the ordering constraints consists of the ordering constraints necessary to satisfy the dependencies in said data flow and data dependency graphs;
- and
- grouping instructions in said machine data flow graph into parallel executable code by adding a second portion of ordering constraints, wherein the second portion of the ordering constraints consists of ordering constraints necessary to map said machine data flow graph to a computation device;
- and
- producing an executable form from said machine data flow graph;
- wherein the steps further comprise a step of defining a final execution order for said parallel executable code when said parallel executable code is executed in a computing device or in said step of grouping instructions.
An apparatus comprising a computer program product stored in a memory for compiling a computer program from a source code into an executable code, which apparatus is arranged to perform the following steps when executing said computer program product:
- parsing the source code for defining an abstract parse tree;
- synthesizing said abstract parse tree for defining data flow and data dependency graphs;
- refining said data flow and data dependency graphs into a machine data flow graph in order to form ordering constraints;
- defining a partial execution order with a first portion of the ordering constraints prescribed in the source code, wherein the first portion of the ordering constraints consists of the ordering constraints necessary to satisfy the dependencies in said data flow and data dependency graphs;
- grouping instructions in said machine data flow graph into parallel executable code by adding a second portion of ordering constraints, wherein the second portion of the ordering constraints consists of ordering constraints necessary to map said machine data flow graph to a computation device;
- and
- producing an executable form from said machine data flow graph;
- wherein the steps further comprise a step of defining a final execution order for said parallel executable code when said parallel executable code is executed in a computing device or in said step of grouping instructions.
Dependencies are inferred from the claim text.