3-D stacked multiprocessor structures and methods for multimodal operation of same
원문보기
IPC분류정보
국가/구분
United States(US) Patent
등록
국제특허분류(IPC7판)
G06F-015/00
G06F-015/76
G06F-015/173
G06F-009/38
출원번호
US-0452113
(2012-04-20)
등록번호
US-9471535
(2016-10-18)
발명자
/ 주소
Buyuktosunoglu, Alper
Emma, Philip G.
Hartstein, Allan M.
Healy, Michael B.
Kailas, Krishnan Kunjunny
출원인 / 주소
International Business Machines Corporation
대리인 / 주소
Davis, Jennifer R.
인용정보
피인용 횟수 :
0인용 특허 :
15
초록▼
Three-dimensional (3-D) processor devices are provided, which are constructed by connecting processors in a stacked configuration. For instance, a processor system includes a first processor chip comprising a first processor and a second processor chip comprising a second processor. The first and se
Three-dimensional (3-D) processor devices are provided, which are constructed by connecting processors in a stacked configuration. For instance, a processor system includes a first processor chip comprising a first processor and a second processor chip comprising a second processor. The first and second processor chips are connected in a stacked configuration with the first and second processors connected through vertical connections between the first and second processor chips. The processor system further includes a mode control circuit to selectively operate the processor system in one of a plurality of operating modes. For example, in a one mode of operation, the first and second processors are configured to implement a run-ahead function, wherein the first processor operates a primary thread of execution and the second processor operates a run-ahead thread of execution.
대표청구항▼
1. A processor system, comprising: a first processor chip comprising a first processor;a second processor chip comprising a second processor,wherein the first and second processor chips are connected in a stacked configuration with the first and second processors connected through vertical connectio
1. A processor system, comprising: a first processor chip comprising a first processor;a second processor chip comprising a second processor,wherein the first and second processor chips are connected in a stacked configuration with the first and second processors connected through vertical connections between the first and second processor chips; anda mode control circuit to selectively operate the processor system in one of a plurality of operating modes, wherein in a first mode of operation, the first and second processors are logically configured as a single processor as seen by other entities, which is configured to implement a run-ahead function, wherein the first processor operates a primary thread of execution and the second processor operates a run-ahead thread of execution, wherein the primary thread and the run-ahead thread execute concurrently; andwherein in the first mode of operation, program instructions of an executing program are fetched and stored in a shared memory, wherein the first processor executes the program instructions in the shared memory in program sequence, wherein the second processor speculatively runs ahead of the first processor by identifying and executing only data fetch instructions, data store instructions, and branch instructions of the program instructions, while disregarding all other program instructions, and wherein in the first mode of operation, the first processor maintains and modifies an architected state of the executing program and the second processor does not maintain and modify the architected state of the executing program but rather maintains a speculative state;wherein in the first mode of operation, the second processor is configured to identify an upcoming data store instruction stored in the shared memory, and to determine if a cache line exists in a cache storage for the upcoming date store instruction, wherein if it is determined that the cache line does not exist in the cache storage, the second processor is configured to (i) have a cache line allocated for the upcoming data store instruction and (ii) ensure that the allocated cache line is in a data store ready state which enables data to be stored in the allocated cache line, andwherein if it is determined that the cache line does exist in the cache storage, the second processor is configured to (i) determine if the existing cache line is in a data store ready state and (ii) proceed to place the existing cache line in a data store ready state, if the existing cache line is determined by the second processor to not be in a data store ready state. 2. The processor system of claim 1, wherein the first and second processor chips are substantially the same. 3. The processor system of claim 1, wherein in the first mode of operation, the first and second processors communicate with each other using the vertical connections between the first and second processor chips. 4. The processor system of claim 1, wherein in the first mode of operation where the first and second processors are logically configured as a single processor, one or more portions of the cache storage used by at least the first processor or the second processor is not used as cache storage, but used instead as a shared private storage with an application specific structure, which is not visible to other entities in the processor system except the first and second processors, wherein the cache storage comprises one or more L1 caches of the first processor, the second processor or both. 5. The processor system of claim 1, wherein a portion of the shared memory is reconfigured to be a private storage area that is only accessible by the first and second processors to hold state that is not visible to any entity other than the first and second processors. 6. The processor system of claim 1, wherein the shared memory is a shared cache that is configured by aggregating two vertically aligned caches associated with the first and second processors.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (15)
Karlsson, Martin, Allocating processor resources during speculative execution using a temporal ordering policy.
Zhou, Qing A; Lu, Daoqiang; Shi, Wei; He, Jiangqi, Electronic assembly with stacked IC's using two or more different connection technologies and methods of manufacture.
Tanguay ; Jr. Armand R. (Fullerton CA) Jenkins B. Keith (Long Beach CA), Modulator-based photonic chip-to-chip interconnections for dense three-dimensional multichip module integration.
Carson John C. (Corona del Mar CA) DeCaro Robert E. (San Juan Capistrano CA) Hsu Ying (Huntington Beach CA) Miyake Michael K. (Westminster CA), Stackable modules and multimodular assemblies.
Segelken John M. (Morristown NJ) Shively Richard R. (Convent Station NJ) Stanziola Christopher A. (Hyde Park NY) Wu Lesley J. (Denville NJ), Stacked board assembly for computing machines, including routing boards.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.