Embedded Performance Engineering
메모리 대역폭 분석 — STREAM·Roofline·Bus Saturation 측정
STREAM benchmark (Copy·Scale·Add·Triad). Roofline. PMU BUS_ACCESS · DDR bandwidth.
· 5분 읽기
memory bandwidth
Blog
Tech articles, book reviews, and presentations.
페이지 57 / 73
Embedded Performance Engineering
Cache Line 최적화 — Alignment·Prefetch·False Sharing 처리
64-byte line alignment, software prefetch, false sharing 회피, SoA·AoS 선택.
· 4분 읽기
cache line
Embedded Performance Engineering
Cache Miss 3C Model 분석 — Compulsory·Capacity·Conflict
Cold/Compulsory, Capacity (working set > cache), Conflict (associativity 한계).
· 5분 읽기
cache miss
Embedded Performance Engineering
CPU Cache 기초 — L1·L2·L3·Set Associative·Replacement Policy
Cache hierarchy. Direct mapped vs N-way set associative. LRU·PLRU·Random.
· 6분 읽기
cache l1
Embedded Performance Engineering
Speculative Execution 분석 — OoO·Reorder Buffer·Register Renaming
Out-of-order execution. ROB·issue queue·rename. Spectre 측면. Cortex-A 사례.
· 5분 읽기
cpu speculative
Embedded Performance Engineering
Branch Prediction 분석 — Static·2-bit·BTB·BHT·Mispredict 비용
BTFNT, 2-bit saturating counter, BTB·BHT. Mispredict 10-20 cycle. PMU BR_MIS_PRED.
· 5분 읽기
cpu branch
Embedded Performance Engineering
Pipeline Stall 분석 — Data·Structural·Control Hazard·Forwarding
Stall은 pipeline bubble을 만듭니다. RAW·WAR·WAW hazard, forwarding, PMU STALL counter를 살펴봅니다.
· 4분 읽기
cpu pipeline
Embedded Performance Engineering
CPU 파이프라인 분석 — 5-stage·Cortex-M·Cortex-A 비교
Fetch·Decode·Execute·Memory·Writeback의 5-stage 파이프라인을 봅니다. Cortex-M3/M4는 3-stage, Cortex-A는 8~15-stage입니다.
· 5분 읽기
cpu pipeline
Embedded Performance Engineering
프로파일링 기법 개요 — Sampling vs Instrumentation·PGO·LTO
두 가지 큰 접근을 비교합니다. Sampling은 perf처럼 가볍고 Instrumentation은 gprof처럼 정확하지만 무겁습니다. PGO로 최적화하는 방법도 다룹니다.
· 5분 읽기
profiling sampling
Embedded Performance Engineering
성능 모델링 — Amdahl·Gustafson·Roofline Model 적용
최적화 한계를 예측하는 수학 모델. Serial 부분이 결정. Memory-bound vs Compute-bound.
· 6분 읽기
amdahl gustafson