Granted Patent B2

US 12,608,201 · App. 18/886,639 · Granted Apr 21, 2026

Inventors: Robert Valentine (Kiryat Tivon, IL); Galina Ryvchin (Haifa, IL); Piotr Majcher (Straszyn, PL); Mark J. Charney (Lexington, MA); Elmoustapha Ould-Ahmed-Vall (Chandler, AZ); Jesus Corbal (King City, OR); Milind B. Girkar (Sunnyvale, CA); Zeev Sperber (Zichron Yackov, IL); Simon Rubanovich (Haifa, IL); Amit Gradstein (Binyamina, IL)

Assignee: Intel Corporation

G06F9/30014G06F7/5443G06F9/30018G06F9/30036G06F9/30038G06F9/30105G06F9/3818

Abstract

Claims (24)

1 . An apparatus, comprising:

circuitry to fetch one or more instructions, the one or more instructions to indicate a first source vector comprising a first plurality of integer data elements, a second source vector comprising a second plurality of integer data elements, and one or more accumulation integer data elements, wherein each of the one or more accumulation integer data elements is four times larger than each data element of the first plurality of integer data elements and the second plurality of integer data elements, and wherein the first plurality of integer data elements and the one or more accumulation integer data elements are signed integer data elements and the second plurality of integer data elements are unsigned integer data elements;

on-chip storage to store the first plurality of integer data elements, the second plurality of integer data elements, and the one or more accumulation integer data elements; and

execution circuitry to execute the one or more instructions to generate one or more result integer data elements, wherein to generate the one or more result integer data elements, the execution circuitry is to:

multiply each data element of the first plurality of integer data elements with a corresponding data element of the second plurality of integer data elements to generate a plurality of products, and

accumulate the plurality of products in groups of four, each group of four products to be accumulated with a corresponding accumulation integer data element of the one or more accumulation integer data elements with saturation to generate a corresponding one or more result integer data elements.

2 . The apparatus of claim 1 , wherein to accumulate with saturation, a maximum value is to be set for a corresponding data element of the one or more result integer data elements if a width of an accumulated value of a corresponding group of four products and the corresponding accumulation integer data element is larger than a threshold.

3 . The apparatus of claim 1 , further comprising: at least a first register to store the first plurality of integer data elements, at least a second register to store the second plurality of integer data elements, and at least a third register to store both the one or more accumulation integer data elements and the one or more result integer data elements.

4 . The apparatus of claim 1 , wherein the on-chip storage comprises at least one of a cache memory and a plurality of registers.

5 . The apparatus of claim 1 , wherein the first plurality of integer data elements and the second plurality of integer data elements comprise 8-bit integer data elements and the one or more accumulation data elements comprise 32-bit integer data elements.

6 . The apparatus of claim 1 , wherein the one or more accumulation integer data elements comprise a plurality of accumulation integer data elements and the one or more result data elements comprise a corresponding plurality of result data elements.

7 . The apparatus of claim 6 wherein each group of four products is to be accumulated with a corresponding accumulation integer data element within a different processing lane of a plurality of processing lanes, wherein each lane of the plurality of processing lanes has a size equal to a size of each of the accumulation integer data elements.

8 . A machine-readable storage medium having program code stored thereon which, when compiled by a machine, causes the machine to generate one or more instructions, the machine to:

fetch the one or more instructions, the one or more instructions to indicate a first source vector comprising a first plurality of integer data elements, a second source vector comprising a second plurality of integer data elements, and one or more accumulation integer data elements, wherein each of the one or more accumulation integer data elements is four times larger than each data element of the first plurality of integer data elements and the second plurality of integer data elements, and wherein the first plurality of integer data elements and the one or more accumulation integer data elements are signed integer data elements and the second plurality of integer data elements are unsigned integer data elements;

execute the one or more instructions to perform operations, comprising:

reading the first plurality of integer data elements, the second plurality of integer data elements, and the one or more accumulation integer data elements from an on-chip storage; and

multiplying each data element of the first plurality of integer data elements with a corresponding data element of the second plurality of integer data elements to generate a plurality of products, and

accumulating the plurality of products in groups of four, each group of four products to be accumulated with a corresponding accumulation integer data element of the one or more accumulation integer data elements with saturation to generate a corresponding one or more result integer data elements.

9 . The machine-readable storage medium of claim 8 , wherein to accumulate with saturation, a maximum value is to be set for a corresponding data element of the one or more result integer data elements if a width of an accumulated value of a corresponding group of four products and the corresponding accumulation integer data element is larger than a threshold.

10 . The machine-readable storage medium of claim 8 , wherein the first plurality of integer data elements are to be stored in at least a first register, the second plurality of integer data elements are to be stored in at least a second register, and the one or more accumulation integer data elements and the one or more result integer data elements are to be stored in at least a third register.

11 . The machine-readable storage medium of claim 8 , wherein the on-chip storage comprises at least one of a cache memory and a plurality of registers.

12 . The machine-readable storage medium of claim 8 , wherein the first plurality of integer data elements and the second plurality of integer data elements comprise 8 -bit integer data elements and the one or more accumulation data elements comprise 32-bit integer data elements.

13 . The machine-readable storage medium of claim 8 , wherein the one or more accumulation integer data elements comprise a plurality of accumulation integer data elements and the one or more result data elements comprise a corresponding plurality of result data elements.

14 . The machine-readable storage medium of claim 13 , wherein each group of four products is to be accumulated with a corresponding accumulation integer data element within a different processing lane of a plurality of processing lanes, wherein each lane of the plurality of processing lanes has a size equal to a size of each of the accumulation integer data elements.

Continuity (5)

References Cited (197)