公开(公告)号：US20230185570A1

公开(公告)日：2023-06-15

申请号：US18107374

申请日：2023-02-08

Applicant: NVIDIA Corporation

Inventor： Andrew KERR ,  Jack Choquette ,  Xiaogang Qiu ,  Omkar Paranjape ,  Poornachandra Rao ,  Shirish Gadre ,  Steven J. Heinrich ,  Manan Patel ,  Olivier Giroux ,  Alan Kaatz

IPC: G06F9/30 ,  G06F12/0808 ,  G06F12/0888 ,  G06F9/32 ,  G06F9/38 ,  G06F9/52 ,  G06F9/54

CPC classification number: G06F9/30043 ,  G06F12/0808 ,  G06F12/0888 ,  G06F9/3009 ,  G06F9/321 ,  G06F9/3871 ,  G06F9/522 ,  G06F9/542 ,  G06F9/544 ,  G06F9/546 ,  G06F9/3838 ,  G06F2212/621 ,  G06F9/3004

Abstract: A technique for block data transfer is disclosed that reduces data transfer and memory access overheads and significantly reduces multiprocessor activity and energy consumption. Threads executing on a multiprocessor needing data stored in global memory can request and store the needed data in on-chip shared memory, which can be accessed by the threads multiple times. The data can be loaded from global memory and stored in shared memory using an instruction which directs the data into the shared memory without storing the data in registers and/or cache memory of the multiprocessor during the data transfer.

公开(公告)号：US20210326137A1

公开(公告)日：2021-10-21

申请号：US17363561

申请日：2021-06-30

Applicant: NVIDIA Corporation

Inventor： Andrew KERR ,  Jack CHOQUETTE ,  Xiaogang QIU ,  Omkar PARANJAPE ,  Poornachandra RAO ,  Shirish GADRE ,  Steven J. HEINRICH ,  Manan PATEL ,  Olivier GIROUX ,  Alan KAATZ

IPC: G06F9/30 ,  G06F12/0808 ,  G06F12/0888 ,  G06F9/32 ,  G06F9/38 ,  G06F9/52 ,  G06F9/54

Abstract: A technique for block data transfer is disclosed that reduces data transfer and memory access overheads and significantly reduces multiprocessor activity and energy consumption. Threads executing on a multiprocessor needing data stored in global memory can request and store the needed data in on-chip shared memory, which can be accessed by the threads multiple times. The data can be loaded from global memory and stored in shared memory using an instruction which directs the data into the shared memory without storing the data in registers and/or cache memory of the multiprocessor during the data transfer.