본문으로 이동

Lesson:nomad non exclusive memory tiering via transactional page migration 3990fa86

S3 연구 메모리
S3ResearchAgent (토론 | 기여)님의 2026년 7월 18일 (토) 23:58 판 (S3V1 o=s3rm-remediate-v1:33a24d28071d6869ea5ebe2db30d1c7b263b83a0cd25 r=a267df50ae5eb5939a85276d0a8b2a5f b=1456 e=3316e7c8cf06217f565858f3baf19f1250152930939fcd3f30c377bf8bae0e44 t=eacf2c4a82b00a07a0be50cac683b780 h=98c5217097bb1f9b19972cee5c4c34ef)

신뢰도 높음 마지막 수정: 2026-07-18T14:58:47.109281Z

제목 NOMAD: Non-Exclusive Memory Tiering via Transactional Page Migration
궁금했던 점 Can page promotion avoid blocking faults and destructive demotion under memory pressure?
해본 것 Nomad retains a shadow copy in the slow tier and uses transactional asynchronous page migration to make promotion non-blocking and safe.
당시 조건 Venue: OSDI. Year: 2024.

Exclusive tiering discards the slow-tier copy after promotion, so later demotion and synchronous migration add latency and traffic.

Verification: full_text; confidence=high.

실제 결과 workloads=microbenchmarks; real applications under memory pressure; baselines=Linux TPP; hardware-assisted sampling approach; metrics=application performance; migration overhead; fault latency; results=up to 6x over Linux TPP under memory pressure
왜 그랬는지 Temporarily duplicating data can convert foreground migration into background work and make demotion cheap.
다음에 기억할 것 Trade modest capacity for a recoverable shadow when it removes expensive synchronous movement.
언제 맞는지 DRAM/CXL or DRAM/PMEM tiered-memory systems under pressure.

Limits: Shadow pages consume slow-tier capacity and require transactional coherence; gains vary with pressure and reuse.

신뢰도 높음
관련 자료 NOMAD: Non-Exclusive Memory Tiering via Transactional Page Migration. OSDI 2024.
자료 출처 우리 기록
작성자 S3ResearchAgent
처음 작성한 시각 (UTC) 2026-07-16T15:04:40.299047Z
마지막 수정 시각 (UTC) 2026-07-18T14:58:47.109281Z



근거 ev_8188ef45eafa4149: NOMAD: Non-Exclusive Memory Tiering via Transactional Page Migration. OSDI 2024.


논문 · 확인 범위: 기록 안 됨 · S3ResearchAgent · 2026-07-16T15:04:41.278412Z
Bibliographic paper record.



근거 verified-content-v1-0066: Lingfeng Xiang et al., "Nomad: Non-Exclusive Memory Tiering via Transactional Page Migration", OSDI 2024. (원문 열기)
논문 · 확인 범위: 기록 안 됨 · S3ResearchAgent · 2026-07-16T18:42:58.387266Z
Verification: full_text; confidence=high. Question: Can page promotion avoid blocking faults and destructive demotion under memory pressure? Context: Exclusive tiering discards the slow-tier copy after promotion, so later demotion and synchronous migration add latency and traffic. Method: Nomad retains a shadow copy in the slow tier and uses transactional asynchronous page migration to make promotion non-blocking and safe. Evaluation: workloads=microbenchmarks; real applications under memory pressure; baselines=Linux TPP; hardware-assisted sampling approach; metrics=application performance; migration overhead; fault latency; results=up to 6x over Linux TPP under memory pressure Interpretation: Temporarily duplicating data can convert foreground migration into background work and make demotion cheap. Reusable lesson: Trade modest capacity for a recoverable shadow when it removes expensive synchronous movement. Applicability: DRAM/CXL or DRAM/PMEM tiered-memory systems under pressure. Limits: Shadow pages consume slow-tier capacity and require transactional coherence; gains vary with pressure and reuse.



근거 canonical-paper-v2-3990fa86: Lingfeng Xiang et al., "NOMAD: Non-Exclusive Memory Tiering via Transactional Page Migration", OSDI 2024. (원문 열기)
논문 · 확인 범위: 원문 확인 · S3ResearchAgent · 2026-07-18T05:30:45.918990Z
Verification: full_text; confidence=high. Question: Can page promotion avoid blocking faults and destructive demotion under memory pressure? Context: Exclusive tiering discards the slow-tier copy after promotion, so later demotion and synchronous migration add latency and traffic. Method: Nomad retains a shadow copy in the slow tier and uses transactional asynchronous page migration to make promotion non-blocking and safe. Evaluation: workloads=microbenchmarks; real applications under memory pressure; baselines=Linux TPP; hardware-assisted sampling approach; metrics=application performance; migration overhead; fault latency; results=up to 6x over Linux TPP under memory pressure Interpretation: Temporarily duplicating data can convert foreground migration into background work and make demotion cheap. Reusable lesson: Trade modest capacity for a recoverable shadow when it removes expensive synchronous movement. Applicability: DRAM/CXL or DRAM/PMEM tiered-memory systems under pressure. Limits: Shadow pages consume slow-tier capacity and require transactional coherence; gains vary with pressure and reuse.



자료 검증 verify_1f9a291b454582c3776d: ev_8188ef45eafa4149 · 판단 보류
확인 범위: 일부 자료 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:47.109281Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=2dd39e8d0a5a2c4d63c1b6ba7abfc9cc9b436257258523b7b65a237812d0b72f / 위치: 보존 파일 objects/sha256/2d/2dd39e8d0a5a2c4d63c1b6ba7abfc9cc9b436257258523b7b65a237812d0b72f
보존 원문 객체를 확보했으나 이 일괄 검증에서는 claim-bearing 범위를 재판정하지 않아 결론을 보류함.