Back to results
US 2026/0300182 A1Published application
Data transfer technique
AssigneeNVIDIA Corporation
InventorsRachit Garg, Sukru Burc Eryilmaz, Michael Andersch, Apoorv Parle, Herbert Hum, Kefeng Duan, Jiang Jiang
Life of this patent
Estimating the term…
Abstract
Apparatuses, systems, and techniques are to transfer data based, at least in part, on a computational graph. In at least one embodiment, a processor causes a compiler to generate instructions to prefetch one or more data values from dynamic random access memory (DRAM) into an in-processor cache based, at least in part, on a computational graph.
Claims20
One or more processors, comprising:
- circuitry to, in response to an application programming interface (API) call:
- receive a computational graph representing one or more data dependencies, the one or more data dependencies indicating that one or more data values used by a first software instruction are to be used by a second software instruction;
- generate, based on the computational graph, one or more prefetch instructions;
- and
- cause, based on the one or more prefetch instructions, the one or more data values to be copied from a first storage location to a second storage location prior to execution of the second software instruction.
2 more elements
A system, comprising:
- one or more processors;
- and
- memory storing instructions that, when executed by the one or more processors, cause the one or more processors, in response to an application programming interface (API) call, to:
- receive a computational graph representing one or more data dependencies, the one or more data dependencies indicating that one or more data values used by a first software instruction are to be used by a second software instruction;
- generate, based on the computational graph, one or more prefetch instructions;
- and
- cause, based on the one or more prefetch instructions, the one or more data values to be copied from a first storage location to a second storage location prior to execution of the second software instruction.
4 more elements
A method, comprising:
- in response to an application programming interface (API) call:
- receiving a computational graph representing one or more data dependencies, the one or more data dependencies indicating that one or more data values used by a first software instruction are to be used by a second software instruction;
- generating, based on the computational graph, one or more prefetch instructions;
- and
- causing, based on the one or more prefetch instructions, the one or more data values to be copied from a first storage location to a second storage location prior to execution of the second software instruction.
2 more elements
Dependencies are inferred from the claim text.