US 12,271,339 B2
Instruction format and instruction set architecture for tensor streaming processor
Dennis Charles Abts, Eau Claire, WI (US); Jonathan Alexander Ross, Palo Alto, CA (US); John Thompson, Minneapolis, MN (US); and Gregory Michael Thorson, Palo Alto, CA (US)
Assigned to Groq, Inc., Mountain View, CA (US)
Filed by Groq, Inc., Mountain View, CA (US)
Filed on Oct. 9, 2023, as Appl. No. 18/483,026.
Application 18/483,026 is a continuation of application No. 17/684,337, filed on Mar. 1, 2022, granted, now 11,822,510.
Application 17/684,337 is a continuation of application No. 17/105,976, filed on Nov. 27, 2020, granted, now 11,360,934, issued on Jun. 14, 2022.
Application 17/105,976 is a continuation in part of application No. 17/103,910, filed on Nov. 24, 2020, abandoned.
Application 17/105,976 is a continuation in part of application No. 16/951,938, filed on Nov. 18, 2020, granted, now 11,868,804.
Application 17/105,976 is a continuation in part of application No. 16/526,966, filed on Jul. 30, 2019, granted, now 11,263,129, issued on Mar. 1, 2022.
Application 17/105,976 is a continuation in part of application No. 16/132,243, filed on Sep. 14, 2018, granted, now 11,243,880, issued on Feb. 8, 2022.
Claims priority of provisional application 63/114,500, filed on Nov. 16, 2020.
Claims priority of provisional application 62/940,815, filed on Nov. 26, 2019.
Claims priority of provisional application 62/937,123, filed on Nov. 18, 2019.
Claims priority of provisional application 62/559,333, filed on Sep. 15, 2017.
Prior Publication US 2024/0037064 A1, Feb. 1, 2024
Int. Cl. G06F 15/82 (2006.01); G06F 9/38 (2018.01); G06N 20/00 (2019.01)
CPC G06F 15/825 (2013.01) [G06F 9/3851 (2013.01); G06N 20/00 (2019.01)] 16 Claims
OG exemplary drawing
 
1. A compiler for operating a tensor streaming processor comprising: compiling a plurality of instructions for operation of the tensor streaming processor by the compiler operating on at least one computer processor and storing the plurality of instructions in a tangible storage medium, wherein the compiler:
allocates functional resources and data transport lanes of the tensor streaming processor for performing a selected operation as determined by each of the plurality of instructions,
schedules processing of one or more data flows through specific functional resources of the tensor streaming processor, and
schedules instruction flow for a plurality of instruction flows, each instruction flow causing a selected one or more functional resources to execute a specific instruction using specific data at a defined time instant.