US2022027432A1PendingUtilityA1

System, method and computer program product for dense/sparse linear system solver accelerator

Assignee: IBMPriority: Jan 24, 2017Filed: Sep 23, 2021Published: Jan 27, 2022
Est. expiryJan 24, 2037(~10.5 yrs left)· nominal 20-yr term from priority
G06F 7/5443G06F 17/16G06F 17/11
54
PatentIndex Score
0
Cited by
0
References
0
Claims

Abstract

An addressable high throughput multiply accumulator (AHT-MAC), including processing of data in row-major format where column-major is required or vice versa by multiplying a first argument by a second argument corresponding to an address in a memory to obtain a first result and adding the first result to a prior result having a prior address that matches the address corresponding to the first argument and the second argument while concurrently starting a multiplication operation in a second address.

Claims

exact text as granted — not AI-modified
What is claimed is: 
     
         1 . An addressable high throughput multiply accumulator (AHT-MAC), comprising:
 processing of data in row-major format where column-major is required or vice versa by:
 multiplying a first argument by a second argument corresponding to an address in a memory to obtain a first result; and 
 adding the first result to a prior result having a prior address that matches the address corresponding to the first argument and the second argument while concurrently starting a multiplication operation in a second address.

Join the waitlist — get patent alerts

Track US2022027432A1 — get alerts on status changes and closely related new filings.

We store only your email — no account needed. See our privacy policy.