公开(公告)号：US20240143330A1

公开(公告)日：2024-05-02

申请号：US17974314

申请日：2022-10-26

Applicant: Nvidia Corporation

Inventor： Michael Houston ,  Ryan Kelsey Albright ,  Benjamin Goska ,  Siddha Ganju ,  Elad Mentovich

IPC: G06F9/38 ,  G06F1/26

CPC classification number: G06F9/3802 ,  G06F1/26

Abstract: Systems and methods herein address power for one or more processing units, using one of a plurality of power profiles during execution of a group of real-time instructions, the one of the plurality of power profiles determined based in part on a relationship determined between the one of the plurality of power profiles and a power profile of the group of real-time instructions, the relationship limited by a threshold, and the plurality of power profiles are associated with a plurality of groups of reference instructions.

公开(公告)号：US20150081866A1

公开(公告)日：2015-03-19

申请号：US14149327

申请日：2014-01-07

Applicant: Nvidia Corporation

Inventor： Jonathan Cohen ,  Michael Houston ,  Frank Jargstorff ,  Eric Young ,  Roy Kim

IPC: H04L29/06

CPC classification number: H04L69/03 ,  H04L67/02 ,  H04L67/1097 ,  H04L67/2828 ,  H04L67/40

Abstract: A special-purpose processing system, a method of carrying out sharing special-purpose processing resources and a graphics processing system. In one embodiment, the special-purpose processing system includes: (1) a special-purpose processing resource and (2) a Representational State Transfer (ReST) application programming interface operable to process data using the special-purpose processing resource in response to stateless commands based on a standard protocol selected from the group consisting of: (2a) a standard network protocol and (2b) a standard database query protocol.

Abstract translation: 专用处理系统，实现共享专用处理资源的方法和图形处理系统。 在一个实施例中，专用处理系统包括：（1）专用处理资源和（2）表示状态转移（ReST）应用程序编程接口，可操作以响应于无状态而使用专用处理资源处理数据 基于从由以下组成的组中选择的标准协议的命令：（2a）标准网络协议和（2b）标准数据库查询协议。

公开(公告)号：US11983536B1

公开(公告)日：2024-05-14

申请号：US17974314

申请日：2022-10-26

Applicant: Nvidia Corporation

Inventor： Michael Houston ,  Ryan Kelsey Albright ,  Benjamin Goska ,  Siddha Ganju ,  Elad Mentovich

IPC: G06F9/38 ,  G06F1/26

CPC classification number: G06F9/3802 ,  G06F1/26

Abstract: Systems and methods herein address power for one or more processing units, using one of a plurality of power profiles during execution of a group of real-time instructions, the one of the plurality of power profiles determined based in part on a relationship determined between the one of the plurality of power profiles and a power profile of the group of real-time instructions, the relationship limited by a threshold, and the plurality of power profiles are associated with a plurality of groups of reference instructions.

公开(公告)号：US12299577B2

公开(公告)日：2025-05-13

申请号：US15624577

申请日：2017-06-15

Applicant: NVIDIA CORPORATION

Inventor： Boris Ginsburg ,  Sergei Nikolaev ,  Ahmad Kiswani ,  Hao Wu ,  Amir Gholaminejad ,  Slawomir Kierat ,  Michael Houston ,  Alex Fit-Florea

IPC: G06N3/084 ,  G06F17/16 ,  G06N3/04 ,  G06N3/045 ,  G06N3/088

Abstract: Aspects of the present invention are directed to computer-implemented techniques for improving the training of artificial neural networks using a reduced precision (e.g., float16) data format. Embodiments of the present invention rescale tensor values prior to performing matrix operations (such as matrix multiplication or matrix addition) to prevent overflow and underflow. To preserve accuracy throughout the performance of the matrix operations, the scale factors are defined using a novel data format to represent tensors, wherein a matrix is represented by the tuple X, where X=(a, v[.]), wherein a is a float scale factor and v[.] are scaled values stored in the float16 format. The value of any element X[i] according to this data format would be equal to a*v[i].

公开(公告)号：US20240004705A1

公开(公告)日：2024-01-04

申请号：US17856368

申请日：2022-07-01

Applicant: NVIDIA Corporation

Inventor： Chad Robert Plummer ,  Pratikkumar Dilipkumar Patel ,  Jun Gu ,  Tao Li ,  Divya Ramakrishnan ,  Michael Houston

IPC: G06F9/48

CPC classification number: G06F9/4893

Abstract: A device comprises one or more circuits that dynamically adjust a load profile of one or more processing devices processing a workload in a bulk-synchronous mode.

公开(公告)号：US10200508B2

公开(公告)日：2019-02-05

申请号：US14149327

申请日：2014-01-07

Applicant: Nvidia Corporation

Inventor： Jonathan Cohen ,  Michael Houston ,  Frank Jargstorff ,  Eric Young ,  Roy Kim

IPC: H04L29/06 ,  H04L29/08

Abstract: A special-purpose processing system, a method of carrying out sharing special-purpose processing resources and a graphics processing system. In one embodiment, the special-purpose processing system includes: (1) a special-purpose processing resource and (2) a Representational State Transfer (ReST) application programming interface operable to process data using the special-purpose processing resource in response to stateless commands based on a standard protocol selected from the group consisting of: (2a) a standard network protocol and (2b) a standard database query protocol.