Back to results
US 2012/0072701 A1Published application
Method Macro Expander
AssigneeNVIDIA Corporation
InventorsJerome Francis Duluk, JR., Jesse David Hall, Patrick R. Brown, Gregory Scott Palmer, Eric S. Werness
Life of this patent
Estimating the term…
Abstract
One embodiment of the present invention sets forth a [TODO once claims are reviewed]
Claims22
A computer-implemented method for transmitting packets of work to a processing pipeline, the method comprising:
- receiving from a driver a first packet of work that includes a first header and a first input parameter, wherein the first header indicates a macro stored in an instruction memory that is to be loaded, and the first input parameter specifies an address associated with a first instruction of the macro;
- identify one or more additional input parameters that are to be processed via the macro;
- processing a first of the one or more additional input parameters via the first instruction to generate a second packet of work;
- and
- transmitting the second packet of work to the processing pipeline for execution.
2 more elements
A method macro expander, configured to:
- receive from a driver a first packet of work that includes a first header and a first input parameter, wherein the first header indicates a macro stored in an instruction memory that is to be loaded, and the first input parameter specifies an address associated with a first instruction of the macro;
- identify one or more additional input parameters that are to be processed via the macro;
- process a first of the one or more additional input parameters via the first instruction to generate a second packet of work;
- and
- transmit the second packet of work to the processing pipeline for execution.
2 more elements
A computer system, comprising:
- a driver;
- and
- a method macro expander coupled to the driver and including an instruction memory for storing one or more macros received from the driver, configured to:
- receive from the driver a first packet of work that includes a first header and a first input parameter, wherein the first header indicates a macro of the one or more macros stored in the instruction memory that is to be loaded, and the first input parameter specifies an address associated with a first instruction of the macro,
- receive a second packet of work that includes a second header and one or more additional input parameters, wherein the second header indicates that the one or more additional input parameters are to be processed via the macro,
- process a first of the one or more additional input parameters via the first instruction to generate a second packet of work, and
- transmit the second packet of work to the processing pipeline for execution.
4 more elements
Dependencies are inferred from the claim text.