US 12,086,700 B2
Neural processor
Ilia Ovsiannikov, Porter Ranch, CA (US); Ali Shafiee Ardestani, San Jose, CA (US); Joseph H. Hassoun, Los Gatos, CA (US); Lei Wang, Burlingame, CA (US); Sehwan Lee, Hwaseong-si (KR); JoonHo Song, Hwaseong-si (KR); Jun-Woo Jang, Hwaseong-si (KR); Yibing Michelle Wang, Pasadena, CA (US); and Yuecheng Li, San Jose, CA (US)
Assigned to Samsung Electronics Co., Ltd., Yongin-si (KR)
Filed by Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed on Aug. 27, 2019, as Appl. No. 16/552,619.
Application 16/552,619 is a continuation of application No. 16/446,610, filed on Jun. 19, 2019.
Claims priority of provisional application 62/689,008, filed on Jun. 22, 2018.
Claims priority of provisional application 62/798,297, filed on Jan. 29, 2019.
Claims priority of provisional application 62/841,590, filed on May 1, 2019.
Claims priority of provisional application 62/841,606, filed on May 1, 2019.
Prior Publication US 2020/0026978 A1, Jan. 23, 2020
Int. Cl. G06N 3/04 (2006.01); G06F 9/30 (2018.01); G06F 17/15 (2006.01); G06F 17/16 (2006.01); G06N 3/08 (2023.01); G06T 9/00 (2006.01)
CPC G06N 3/04 (2013.01) [G06F 17/153 (2013.01); G06F 17/16 (2013.01); G06N 3/08 (2013.01); G06T 9/002 (2013.01); G06F 9/3001 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A processor, comprising:
a first tile,
a second tile,
a memory, and
a bus,
the bus being connected to:
the memory,
the first tile, and
the second tile,
the first tile comprising:
a first weight register,
a second weight register,
an activations buffer,
a first multiplier, and
a second multiplier,
the first tile being configured to perform a convolution of an array of activations with a kernel of weights, the performing of the convolution comprising, in order:
forming a tensor product of the kernel with a first subarray of the array of activations;
forming a tensor product of the kernel with a second subarray of the array of activations, the second subarray being offset from the first subarray by n array elements in a first direction, n being a positive integer; and
forming a tensor product of the kernel with a third subarray of the array of activations, the third subarray being offset from the second subarray by one array element in a second direction, perpendicular to the first direction,
wherein the second subarray and the third subarray are spaced apart from an end of a row of the array of activations.