일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | ||
6 | 7 | 8 | 9 | 10 | 11 | 12 |
13 | 14 | 15 | 16 | 17 | 18 | 19 |
20 | 21 | 22 | 23 | 24 | 25 | 26 |
27 | 28 | 29 | 30 |
- sequential consistency
- Subquery
- dynamic scheduling
- sql
- cache coherence miss
- Cache
- atomic exchange
- theta join
- cache coherence
- load linked
- multibanekd cache
- 관계형 모델
- pipelined
- way prediction
- pipelined cache
- cache optimization
- pipline hazards
- branch prediction
- dependence
- transactional memory
- directory based coherence protocol
- store conditional
- structural hazard
- speculative execution
- register renaming
- moesi
- relational model
- mesi
- ISA
- nonblocking cache
- Today
- Total
목록분류 전체보기 (32)
공대생의 공부흔적

참고: Computer Architecture: A Quantitative Approach (5th edition) - 3. Instruction-Level Parallelism and its Exploitation 지난 글에 이어, 이번 글에서는 Tomasulo 알고리즘에 대해 살펴볼 것이다. 2024.04.14 - [Computer Architecture] - [컴퓨터구조#7-1] ILP (Instruction-Level Parallelism) - dependence, Out-of-Order execution, 동적 스케줄링 Tomasulo Alrgorithm out-of-order execution을 위한 알고리즘으로, 여러 functional unit 간 컨트롤과 로직을 분산시켜 동적 스케줄링을 ..
참고: Computer Architecture: A Quantitative Approach (5th edition) - 3. Instruction-Level Parallelism and its Exploitation 이번 글과 다음 글에서는 ILP에 대해 다룰 것이다. 이번 글에서는 여러 가지 의존성(data/name/control/memory dependence)와 out-of-order execution, 그리고 동적 스케줄링에 대해 소개하고 다음 글에서는 Tomasulo 알고리즘에 대해 논의할 것이다. 목차 Dependence Out-of-Order Execution Dynamic Scheduling Tomasulo Algorithm 0. ILP 본격적인 내용에 들어가기 앞서, ILP 란, 여러 명..

이번 글에서는 캐시와 관련된 고급 주제들에 대해 다룰 것이다. 목차 1. 대체 정책: LRU, PLRU, NRU, Adaptive policy 2. 삽입 정책: LRU, OPT, LIP, BIP 3. 멀티코어에서의 캐시 위계 1. 캐시 대체 정책 Cache Replacement Policy 캐시 내 데이터가 다 찬 경우, 새로운 데이터를 넣기 위해서는 이미 있던 데이터 중 하나를 빼야 한다. 어떤 데이터를 뺄지 고르는 데 여러 방법이 존재할 수 있다. 이때 사용하는 정책을 캐시 대체 정책이라고 한다. 간단히 생각해보면, 가장 긴 재사용 distance를 가지고 있는 블록, 즉 해당 블록에 대한 다음 접근이 가장 먼 미래에 있는 블록을 evict하면 된다. 하지만 이는 미래에 대한 지식을 필요로 한다. 따..

이번 글에서는 프리페칭에 대해 알아볼 것이다. 목차 1. 프리페칭 2. What, When, Where, How 3. 하드웨어 프리페처: Next-line, Stride, Locality-based 4. 프리페처 성능 5. Markov Prefetching 1. 프리페칭 (Prefetching) 프로그램이 데이터를 필요로 하기 전에 미리 fetch해오는 것. → miss rate과 miss latency를 모두 줄일 수 있다. 왜 필요한가? 메모리 지연 시간은 매우 길기 때문에, 정확하고 빠르게 프리페치할 수 있다면 이러한 긴 지연 시간을 줄이거나 없앨 수 있다. Compulsory cache miss를 없앨 수 있다. capacity, conflict, coherence로 인한 캐시 미스는 여전히 발생..

참고: Computer Architecture: A Quantitative Approach (5th edition) - 2.1~2.3. 2024.04.12 - [Computer Architecture] - [컴퓨터구조#4-1] 캐시 최적화 (Cache Optimization) - hit time 줄이기, 캐시 대역폭 늘리기 지난 글에 이어, 이번 글에서는 miss penalty와 miss rate을 줄이는 캐시 최적화 기술에 대해서 알아볼 것이다. 목차 Small and simple first level caches Way prediction Pipelining cache Nonblock cache Multibanked cache Critial word first, Early restart Merging..

참고: Computer Architecture: A Quantitative Approach (5th edition) - 2.1~2.3. 메모리 위계에 대해 살펴본 지난 글에 이어, 이번 글과 다음 글에서는 10가지 고급 캐시 최적화 기술에 대해서 살펴볼 것이다. 이번 글에서는 hit time을 줄이는 small and simple first level caches, way prediction, 그리고 캐시 대역폭을 늘리는 pipelined cache access, nonblocking, multibanked cache에 대해 알아볼 것이다. 2024.04.12 - [Computer Architecture] - [컴퓨터구조#3] 메모리 위계 (Memory Hierarchy) [컴퓨터구조#3] 메모리 위계 ..