IP Library › Granted Patent US 9,983,655
Granted Patent B2
US 9,983,655 · App. 14/963,352 · Granted May 29, 2018

Method and apparatus for performing inter-lane power management

Inventors: Mitesh R. Meswani (Austin, TX); David A. Roberts (Sunnyvale, CA); Dmitri Yudanov (Austin, TX); Arkaprava Basu (Austin, TX); Sergey Blagodurov (Bellevue, WA)
Assignee: Advanced Micro Devices, Inc.
G06F1/3243G06F9/3885
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 9,983,655
App. No.
14/963,352
Granted
May 29, 2018
Kind
B2
Abstract

A method and apparatus for performing inter-lane power management includes de-energizing one or more execution lanes upon a determination that the one or more execution lanes are to be predicated. Energy from the predicated execution lanes is redistributed to one or more active execution lanes.

Claims (20)

1. A method comprising:

de-energizing one or more of a first plurality of execution lanes upon a determination that the one or more of the first plurality of execution lanes are to be predicated, wherein the de-energizing is performed when the one or more of the lanes of the first plurality of execution lanes are predicted to be active an amount that exceeds a pre-defined threshold; and

redistributing energy from the predicated execution lanes to one or more of a second plurality of active execution lanes, wherein one of more lanes of the second plurality of active execution lanes performing latency critical computations are redistributed with more energy than one or more lanes of the second plurality of active execution lanes that are not performing latency critical computations.

2. The method of claim 1 wherein redistributing the energy includes redistributing power.

3. The method of claim 1 wherein redistributing the energy includes increasing clock frequency to the one or more active execution lanes.

4. The method of claim 1 wherein the energy is redistributed evenly among all active execution lanes when all lanes of the second plurality of active execution lanes are performing latency critical computations.

5. An apparatus comprising:

a plurality of execution lane circuits;

an encoder; and

one or more power steering supplies;

wherein the encoder is configured to control the one or more power steering supplies to de-energize one or more of a first plurality of the execution lane circuits upon a determination that the one or more execution lane circuits are to be predicated, wherein the de-energizing is performed when the one or more of the execution lane circuits of the first plurality of execution lanes are predicted to be active an amount that exceeds a pre-defined threshold, and to redistribute energy from the predicated execution lane circuits to one or more of a second plurality of active execution lane circuits, wherein one of more execution lane circuits of the second plurality of active execution lane circuits performing latency critical computations are redistributed with more energy than one or more execution lane circuits of the second plurality of active execution lane circuits that are not performing latency critical computations.

6. The apparatus of claim 5 wherein the encoder is configured to redistribute power.

7. The apparatus of claim 5 wherein the encoder is configured to increase clock frequency to the one or more active execution lane circuits.

8. The apparatus of claim 5 wherein the energy is redistributed evenly among all active execution lane circuits when all lanes of the second plurality of active execution lane circuits are performing latency critical computations.

9. A non-transitory computer-readable medium, having instructions recorded thereon that, when executed by a computing device, cause the computing device to perform operations comprising:

de-energizing one or more of a first plurality of execution lanes upon a determination that the one or more of the first plurality of execution lanes are to be predicated, wherein the de-energizing is performed when the one or more of the lanes of the first plurality of execution lanes are predicted to be active an amount that exceeds a pre-defined threshold; and

redistributing energy from the predicated execution lanes to one or more of a second plurality of active execution lanes, wherein one of more lanes of the second plurality of active execution lanes performing latency critical computations are redistributed with more energy than one or more lanes of the second plurality of active execution lanes that are not performing latency critical computations.

10. The non-transitory computer-readable medium of claim 9 wherein redistributing the energy includes redistributing power.

11. The non-transitory computer-readable medium of claim 9 wherein redistributing the energy includes increasing clock frequency to the one or more active execution lanes.

12. The non-transitory computer-readable medium of claim 9 wherein the energy is redistributed evenly among all active execution lanes when all lanes of the second plurality of active execution lanes are performing latency critical computations.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 9, 2015
From: MESWANI, MITESH R.; ROBERTS, DAVID A.; YUDANOV, DMITRI; BASU, ARKAPRAVA; BLAGODUROV, SERGEY
To: ADVANCED MICRO DEVICES, INC.
Reel/Frame 037252/0092 →
Continuity (1)
Related Publication 20170168546A1 · Jun 15, 2017