DOE Patents title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Performance for GPU exceptions

Abstract

Techniques for improving performance of accelerated processing devices (“APDs”) when exceptions occur are provided. In APDs, the very large number of parallel processing execution units, and the complexity of the hardware used to execute a large number of work-items in parallel, means that APDs typically stall when an exception occurs (unlike in central processing units (“CPUs”), which are able to execute speculatively and out-of-order). However, the techniques provided herein allow at least some execution to occur past exceptions. Execution past an exception generating instruction occurs by executing instructions that would not lead to a corruption while skipping those that would lead to a corruption. After the exception has been satisfied, execution occurs in a replay mode in which the potentially exception-generating instruction is executed and in which instructions that did not execute in the exception-wait mode are executed. A mask and counter are used to control execution in replay mode.

Inventors:
Issue Date:
Research Org.:
Advanced Micro Devices, Inc., Santa Clara, CA (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1892615
Patent Number(s):
11249765
Application Number:
16/109,567
Assignee:
Advanced Micro Devices, Inc. (Santa Clara, CA)
Patent Classifications (CPCs):
G - PHYSICS G06 - COMPUTING G06F - ELECTRIC DIGITAL DATA PROCESSING
G - PHYSICS G06 - COMPUTING G06T - IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
DOE Contract Number:  
AC52-07NA27344; B620717
Resource Type:
Patent
Resource Relation:
Patent File Date: 08/22/2018
Country of Publication:
United States
Language:
English

Citation Formats

Gutierrez, Anthony T. Performance for GPU exceptions. United States: N. p., 2022. Web.
Gutierrez, Anthony T. Performance for GPU exceptions. United States.
Gutierrez, Anthony T. Tue . "Performance for GPU exceptions". United States. https://www.osti.gov/servlets/purl/1892615.
@article{osti_1892615,
title = {Performance for GPU exceptions},
author = {Gutierrez, Anthony T.},
abstractNote = {Techniques for improving performance of accelerated processing devices (“APDs”) when exceptions occur are provided. In APDs, the very large number of parallel processing execution units, and the complexity of the hardware used to execute a large number of work-items in parallel, means that APDs typically stall when an exception occurs (unlike in central processing units (“CPUs”), which are able to execute speculatively and out-of-order). However, the techniques provided herein allow at least some execution to occur past exceptions. Execution past an exception generating instruction occurs by executing instructions that would not lead to a corruption while skipping those that would lead to a corruption. After the exception has been satisfied, execution occurs in a replay mode in which the potentially exception-generating instruction is executed and in which instructions that did not execute in the exception-wait mode are executed. A mask and counter are used to control execution in replay mode.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {2022},
month = {2}
}

Works referenced in this record:

Method and Apparatus for Servicing Page Fault Exceptions
patent-application, June 2013


Instructions and Logic to Interrupt and Resume Paging in a Secure Enclave Page Cache
patent-application, December 2015


iGPU: Exception support and speculative execution on GPUs
conference, June 2012


Towards high performance paged memory for GPUs
conference, March 2016


vDNN: Virtualized deep neural networks for scalable, memory-efficient neural network design
conference, October 2016


Instruction and logic for interrupt and exception handling
patent, October 2019


Efficient exception handling support for GPUs
conference, October 2017