US 11,853,193 B2
Inverse performance driven program analysis
Budirijanto Purnomo, Santa Clara, CA (US); and Chen Shen, Santa Clara, CA (US)
Assigned to Advanced Micro Devices, Inc., Santa Clara, CA (US)
Filed by Advanced Micro Devices, Inc., Santa Clara, CA (US)
Filed on Oct. 29, 2021, as Appl. No. 17/514,398.
Prior Publication US 2023/0140822 A1, May 4, 2023
Int. Cl. G06F 11/36 (2006.01); G06F 9/50 (2006.01); G06F 9/54 (2006.01); G06F 8/41 (2018.01)
CPC G06F 11/3612 (2013.01) [G06F 8/433 (2013.01); G06F 9/5038 (2013.01); G06F 9/54 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A method comprising:
accessing profile data from an execution of a plurality of tasks executed on a plurality of computing resources;
constructing a dependency graph for the plurality of tasks executed on the plurality of computing resources based on the profile data;
causing a user interface to be presented, wherein the user interface includes a representation of the plurality of tasks executed on the plurality of computing resources;
receiving, via the user interface, an input for a modification of one or more execution attributes of one or more target tasks of the plurality of tasks executed on the plurality of computing resources;
determining that the modification of the one or more execution attributes of the one or more target tasks of the plurality of tasks executed on the plurality of computing resources is projected to improve a performance metric of the plurality of computing resources while maintaining a validity of the dependency graph;
presenting, via the user interface, one or more steps to implement the modification of the one or more execution attributes of the one or more target tasks of the plurality of tasks executed on the plurality of computing resources; and
updating the user interface to reflect the modification of the one or more execution attributes of the one or more target tasks of the plurality of tasks executed on the plurality of computing resources, wherein the updating of the user interface includes the projected improvement to the performance metric of the plurality of computing resources.