Event processing by hardware accelerator

Patent metadata

Field	Value
Publication number	US-12554540-B2
Application number	US-202318230744-A
Country	US
Kind code	B2
Filing date	Aug 7, 2023
Priority date	Aug 7, 2023
Publication date	Feb 17, 2026
Grant date	Feb 17, 2026

Abstract

A hardware acceleration circuit, comprising a communication interface for connecting to one or more event-driven circuits, a memory, an event handling circuit, and a hardware acceleration engine. The event handling circuit is adapted to detect one or more events triggered by one or more of the event-driven circuits, update one or more pointers pointing to one or more event handling routines stored in the memory and to a context memory segment in the memory storing a plurality of context parameters relating to the one or more events, and transmit the pointer(s) to the hardware acceleration engine. The hardware acceleration engine is adapted to receive the pointer(s) from the event handling circuit, and execute the event handling routine(s) pointed by the pointer(s) to process data relating to the event(s) according to at least some of the context parameters retrieved from the context memory segment using the pointer(s).

First claim

What is claimed is:

1 . A hardware acceleration circuit coupled to a host processor, comprising:
a communication interface for connecting to at least one event-driven circuit; a memory; a hardware acceleration engine; and an event handling circuit configured to:
autonomously detect at least one event triggered by the at least one event-driven circuit and independent of an execution of a currently executed routine by the host processor,
in response to the detected at least one event, update at least one pointer pointing to at least one event handling routine stored in the memory and to a context memory segment in the memory storing a plurality of context parameters relating to the at least one event, and
directly transmit the at least one pointer to the hardware acceleration engine without intervention of the host processor in the pointer transmission;
wherein the hardware acceleration engine is configured to:
receive the at least one pointer from the event handling circuit, and
execute the at least one event handling routine pointed by the at least one pointer to process data relating to the at least one event according to at least some of the plurality of context parameters retrieved from the context memory segment using the at least one pointer;
wherein the hardware acceleration circuit processes the at least one event without requiring the host processor to perform a context switch from the currently executed routine to the at least one event handling routine.

2 . The hardware acceleration circuit of claim 1 , wherein the hardware acceleration engine is adapted to process the at least one event with no intervention of a host processor.

3 . The hardware acceleration circuit of claim 1 , wherein the event handling circuit is further adapted to allocate at least one hardware resource to support processing of the data relating to the at least one event.

4 . The hardware acceleration circuit of claim 1 , wherein the at least one event handling routine is preloaded into the memory by at least one host processor.

5 . The hardware acceleration circuit of claim 1 , wherein the at least one event handling routine is hard coded in the memory.

6 . The hardware acceleration circuit of claim 1 , wherein at least one of the plurality of context parameters is updated in the context memory segment by the at least one event-driven circuit.

7 . The hardware acceleration circuit of claim 1 , wherein at least one of the plurality of context parameters is updated in the context memory segment by at least one host processor.

8 . The hardware acceleration circuit of claim 1 , wherein the at least one event handling routine and the plurality of context parameters are pre-loaded into the memory before the at least one event is detected.

9 . The hardware acceleration circuit of claim 1 , wherein the event handling circuit is further configured to allocate hardware resources of the hardware acceleration engine based on a type of the at least one event.

10 . The hardware acceleration circuit of claim 1 , wherein the memory comprises a local memory that is separate from a system memory accessible to the host processor, and wherein the event handling routines and the plurality of context parameters are stored in the local memory.

11 . The hardware acceleration circuit of claim 1 , wherein the event handling circuit is configured to detect the at least one event through hardware trigger mechanisms including at least one of: interrupt line assertion by the at least one event-driven circuit, or monitored memory location access by the at least one event-driven circuit.

12 . The hardware acceleration circuit of claim 1 , wherein the event handling circuit is further configured to allocate hardware resources of the hardware acceleration engine based on a type of the detected at least one event, wherein different types of events receive different amounts of allocated hardware resources.

13 . The hardware acceleration circuit of claim 1 , wherein the at least one event handling routine is pre-loaded into the memory prior to detection of the at least one event by the host processor during at least one of: a boot sequence or an initialization process.

14 . The hardware acceleration circuit of claim 1 , wherein the at least one event-driven circuit is configured to write the plurality of context parameters directly to the context memory segment without host processor intervention.

15 . A method of accelerating event processing coupled to a host processor, comprising:
using an event handling circuit for:
autonomously detecting at least one event independent of an execution of a currently executed routine by the host processor and triggered by at least one event-driven circuit,
in response to the detected at least one event, updating at least one pointer pointing to at least one event handling routine stored in a memory and to a context memory segment in the memory storing a plurality of context parameters relating to the at least one event; and
directly transmit the at least one pointer to a hardware acceleration engine without intervention of the host processor in the pointer transmission;
using a hardware acceleration engine for:
receive the at least one pointer from the event handling circuit, and

executing the at least one event handling routine pointed by the at least one pointer to process data relating to the at least one event according to at least some of the plurality of context parameters retrieved from the context memory segment using the at least one pointer;
wherein the hardware acceleration circuit processes the at least one event without requiring the host processor to perform a context switch from the currently executed routine to the at least one event handling routine.

16 . The method of claim 15 , wherein the hardware acceleration engine is adapted to process the at least one event with no intervention of a host processor.

17 . The method of claim 15 , wherein the event handling circuit is further adapted to allocate at least one hardware resource to support processing of the data relating to the at least one event.

18 . The method of claim 15 , wherein the at least one event handling routine is preloaded into the memory by at least one host processor.

19 . The method of claim 15 , wherein the at least one event handling routine is hard coded in the memory.

20 . The method of claim 15 , wherein at least one of the plurality of context parameters is updated in the context memory segment by the at least one event-driven circuit.

21 . The method of claim 15 , wherein at least one of the plurality of context parameters is updated in the context memory segment by at least one host processor.

22 . The hardware acceleration circuit of claim 10 , wherein data relating to the at least one event is transferred only once over a system bus connecting the hardware acceleration circuit to the system memory, eliminating a need for double transfer of the data.

Assignees

• Next Silicon Ltd

Inventors

• Margolin Alexander
• Daskal Menashe
• Nishry Oren

Classifications

• G06F9/542 Primary

  Event management; Broadcasting; Multicasting; Notifications · CPC title

• G06F9/4812

  by interrupt, e.g. masked · CPC title

• G06F9/5027 Primary

  the resource being a machine, e.g. CPUs, Servers, Terminals · CPC title

• G06F2209/509

  Offload · CPC title