# US 2025/0094349 A1 · Processor

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0094349 A1
- Document ID: US20250094349A1
- Status: Published application
- Kind: A1
- Application number: 18/818,781
- Filed: 2024-08-29
- Published: 2025-03-20
- Earliest claimed: 2023-09-14
- Assignees: Fujitsu Limited
- Applicants: Fujitsu Limited
- Inventors: Takahiro SHIKIBU
- Web page: https://patentgrep.com/patent/US20250094349A1

## CPC

- G06F 12/0815: Cache consistency protocols
- G06F 9/3834: {Maintaining memory consistency}
- G06F 9/3842: {Speculative instruction execution}
- G06F 9/3861: {Recovery, e.g. branch miss-prediction, exception handling (error detection or correction G06F11/00)}
- G06F 12/0811: with multilevel cache hierarchies
- G06F 12/0855: Overlapped cache accessing, e.g. pipeline (G06F12/0846 takes precedence)
- G06F 12/0862: with prefetch
- G06F 12/0895: of parts of caches, e.g. directory or tag array
- G06F 12/0897: with two or more cache hierarchy levels (with multilevel cache hierarchies G06F12/0811)
- G06F 2212/1024: Latency reduction
- G06F 2212/507: using speculative control
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

The processor includes core circuits and a cache unit having L2 to LN caches (N is 3 or more). L1 cache has a move-in buffer including entries in which memory access instruction resulted in cache miss in L1 cache is stored. The move-in buffer, when issuing a normal memory request to L2 cache, issues a pseudo memory request to L3 to LN caches in parallel and receives a pseudo data response that has coherency-unsecured data from any one cache. The re-order buffer executes a normal instruction execution completion process in response to the normal data response, executes a pseudo instruction execution completion process in response to the pseudo data response, and, when the pseudo data response is a failure, rewinds an arithmetic operation circuit that speculatively executed instructions after the memory access instruction in response to the pseudo instruction execution completion process, back to a state before the speculative execution.

## Claims

### Claim 1

A processor comprising:
a plurality of core circuits each having an arithmetic operation circuit executing an issued instruction, a load store unit that executes an issued memory access instruction, a level 1 cache, and a re-order buffer that controls an instruction execution completion process of the arithmetic operation circuit and the load store unit; and
a cache unit having a plurality of caches from a level 2 cache to a level N cache, N being an integer equal to or greater than 3,
wherein the level 1 cache has a move-in buffer that has a plurality of entries in which a memory access instruction that results in a cache miss in the level 1 cache is stored,
wherein the move-in buffer
issues a normal memory request, which performs memory access of the memory access instruction that results in the cache miss, to the level 2 cache, the normal memory request being issued from the level 2 cache to the level N cache serially, and
receives a normal data response that has coherency-secured data from any one of the level 2 cache to the level N cache in which the normal memory request result in a cache hit,
wherein, when the normal memory request is issued to the level 2 cache,
a pseudo memory request that performs the memory access is issued to a level 3 cache in a case where N is 3, or
the pseudo memory request is issued to the level 3 cache to the level N cache in parallel in a case where N is 4 or more,
wherein the move-in buffer receives a pseudo data response that has coherency-unsecured data from any one cache in which the pseudo memory request result in a cache hit, and
wherein the re-order buffer
executes a normal instruction execution completion process in response to the normal data response,
executes a pseudo instruction execution completion process in response to the pseudo data response, and,
when the pseudo data response is a failure, rewinds the arithmetic operation circuit, which speculatively executed an instruction after the memory access instruction that result in the cache miss in response to the pseudo instruction execution completion process, back to a state before the speculative execution.

### Claim 2 (depends on 1)

The processor according to claim 1, wherein
the level 1 cache
stores data of the normal data response in a cache memory in the level 1 cache in response to the normal data response and
does not store data of the pseudo data response in the cache memory in the level 1 cache in response to the pseudo data response.

### Claim 3 (depends on 2)

The processor according to claim 2, wherein
the data of the normal data response is stored in a register in the core circuit in response to the normal data response,
the data of the pseudo data response is stored in the register in the core circuit in response to the pseudo data response, and
when the pseudo data response is a failure, the data of the normal data response is overwritten into the register in the core circuit.

### Claim 4 (depends on 1)

The processor according to claim 1, wherein,
when the normal data response is received after reception of the pseudo data response, the move-in buffer determines whether or not data of the pseudo data response coincides with data of the normal data response, determines that the pseudo data response is a success in case of coincidence, and determines that the pseudo data response is a failure in case of no coincidence, and
the success or the failure of the pseudo data response is notified to the re-order buffer.

### Claim 5 (depends on 1)

The processor according to claim 1, wherein
the arithmetic operation circuit executes the issued instructions in an out-of-order different from a program order,
the re-order buffer executes the normal instruction execution completion process of instructions executed by the arithmetic operation circuit in an in-order that is the program order, and
the arithmetic operation circuit executes a first instruction and a second instruction after the first instruction, the second instruction having a dependent relation on the first instruction, executes the second instruction after the normal instruction execution completion process of the first instruction, and further speculatively executes the second instruction after the pseudo instruction execution completion process of the first instruction.

### Claim 6 (depends on 1)

The processor according to claim 1, wherein when the normal memory request is issued by the move-in buffer, the pseudo memory request is issued by the move-in buffer.

### Claim 7 (depends on 1)

The processor according to claim 1, wherein, when the normal memory request is issued by the move-in buffer, the pseudo memory request is issued by the level 2 cache.
