公开(公告)号：US12153539B1

公开(公告)日：2024-11-26

申请号：US18201073

申请日：2023-05-23

Applicant: Oracle International Corporation

Inventor： Kangnyeon Kim ,  Weiwei Gong ,  James Kearney ,  Harshada Chavan

IPC: G06F15/16 ,  G06F15/167

Abstract: An append operation is provided for using a plurality of threads on a plurality of streaming multiprocessors of a graphical processing unit. The append operation writes results into a result buffer. Executing the append operation comprises claiming, by each given thread within the plurality of threads having a result to write, a portion of a selected WCB, writing, by the given thread, the result to the portion of the selected WCB, and in response to a flush condition being met for the selected WCB, copying contents of the selected WCB to a result buffer.

公开(公告)号：US20250138830A1

公开(公告)日：2025-05-01

申请号：US18384734

申请日：2023-10-27

Applicant: Oracle International Corporation

Inventor： Weiwei Gong ,  Harshada Chavan ,  James Kearney ,  Kangnyeon Kim

IPC: G06F9/38

Abstract: An approach is provided for a thorough and clean way of handling graph overflows in graph execution on single instruction, multiple threads (SIMT) hardware with resumable graph support. The solution does not assume that the input and output fit in the buffers allocated in the SIMT hardware. The approach maintains state of the execution for each kernel and uses multiple iterations of graph execution, making progress in each iteration until all data items are processed through the graph on SIMT hardware. This iterative processing of the graph is transparent to the end user. For resumability, the approach treats buffers as circular buffers instead of serial buffers. With the help of counters, the approach keeps track of the start and end indexes of input and output buffers, thus achieving seamless graph resumability when re-execution is required for only a subset of kernels.