US 12,277,499 B2
Vector computation unit in a neural network processor
Gregory Michael Thorson, Waunakee, WI (US); Christopher Aaron Clark, Madison, WI (US); and Dan Luu, Madison, WI (US)
Assigned to Google LLC, Mountain View, CA (US)
Filed by Google LLC, Mountain View, CA (US)
Filed on Apr. 16, 2024, as Appl. No. 18/636,640.
Application 18/636,640 is a continuation of application No. 18/176,640, filed on Mar. 1, 2023, granted, now 12,014,272.
Application 18/176,640 is a continuation of application No. 16/245,406, filed on Jan. 11, 2019, granted, now 11,620,508, issued on Apr. 4, 2023.
Application 16/245,406 is a continuation of application No. 14/845,117, filed on Sep. 3, 2015, granted, now 10,192,162, issued on Jan. 29, 2019.
Claims priority of provisional application 62/165,022, filed on May 21, 2015.
Prior Publication US 2024/0273368 A1, Aug. 15, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 5/08 (2006.01); G06F 7/544 (2006.01); G06N 3/063 (2023.01); G06N 3/08 (2023.01); G06N 5/04 (2023.01)
CPC G06N 3/08 (2013.01) [G06F 5/08 (2013.01); G06F 7/544 (2013.01); G06N 3/063 (2013.01); G06N 5/04 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A vector computation unit for performing neural network computations comprising:
normalization circuitry configured to:
receive a vector of activation values;
receive one or more control signals specifying a normalization function; and
apply the normalization function to the activation values to generate respective normalized values for each activation value.