US 12,242,861 B2
Methods and apparatus to load data within a machine learning accelerator
Arnab Raha, San Jose, CA (US); Deepak Mathaikutty, Chandler, AZ (US); Debabrata Mohapatra, San Jose, CA (US); Sang Kyun Kim, San Jose, CA (US); Gautham Chinya, Sunnyvale, CA (US); and Cormac Brick, San Francisco, CA (US)
Assigned to Intel Corporation, Santa Clara, CA (US)
Filed by Intel Corporation, Santa Clara, CA (US)
Filed on Jan. 18, 2024, as Appl. No. 18/416,303.
Application 18/416,303 is a continuation of application No. 17/359,392, filed on Jun. 25, 2021, granted, now 11,922,178.
Prior Publication US 2024/0231839 A1, Jul. 11, 2024
Int. Cl. G06F 9/445 (2018.01); G06F 9/30 (2018.01); G06F 9/50 (2006.01); G06N 20/00 (2019.01); H03K 19/177 (2020.01); H03K 19/20 (2006.01)
CPC G06F 9/445 (2013.01) [G06F 9/3001 (2013.01); G06F 9/5027 (2013.01); G06N 20/00 (2019.01); H03K 19/177 (2013.01); H03K 19/20 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A method, comprising:
writing compressed data into a memory of a processing element, the compressed data generated by compressing data based on sparsity in the data, the data comprising first operable data;
performing, by the processing element, a first machine learning operation using at least part of the compressed data and a first sparsity bitmap, the at least part of the compressed data generated by compressing the first operable data, the first sparsity bitmap indicating sparsity in the first operable data;
determining whether the data comprises second operable data based on a second sparsity bitmap indicating sparsity in the second operable data;
in response to determining that the data comprises the second operable data, performing, by the processing element, a second machine learning operation using another part of the compressed data that is generated by compressing the second operable data; and
in response to determining that the data does not comprise the second operable data:
writing new data into the memory of the processing element, and
performing, by the processing element, the second machine learning operation using the new data.