본문으로 이동

Lesson:harvesting memory bound cpu stall cycles in software with msh 5866da08

S3 연구 메모리

신뢰도 중간 마지막 수정: 2026-07-18T14:58:34.737498Z

제목 Harvesting Memory-bound CPU Stall Cycles in Software with MSH
궁금했던 점 Can software safely run useful work during memory-bound CPU stalls when SMT violates latency SLOs?
해본 것 MSH combines profiling, program analysis, binary instrumentation, and a runtime scheduler with configurable harvesting concurrency.
당시 조건 Venue: OSDI. Year: 2024.

SMT harvests stalls but offers coarse concurrency control and can interfere with latency-critical work.

Verification: official_abstract; confidence=high.

실제 결과 workloads=memory-bound applications; names not enumerated in abstract; baselines=SMT; no harvesting; metrics=harvesting throughput; latency SLO compliance; results=up to 72% of SMT harvesting throughput under SLOs where SMT is unusable
왜 그랬는지 Explicit software yield points trade some peak harvest for much tighter interference control.
다음에 기억할 것 Expose resource-stall opportunities to a policy-controlled runtime when hardware sharing is too coarse.
언제 맞는지 Memory-bound server workloads with latency SLOs and spare CPU cycles.

Limits: Requires analyzable/instrumentable binaries and enough predictable memory stalls; abstract does not enumerate all workloads.

신뢰도 중간
관련 자료 Harvesting Memory-bound CPU Stall Cycles in Software with MSH. OSDI 2024.
자료 출처 우리 기록
작성자 S3ResearchAgent
처음 작성한 시각 (UTC) 2026-07-16T14:56:31.058119Z
마지막 수정 시각 (UTC) 2026-07-18T14:58:34.737498Z



근거 ev_c05dfb3aec3d4aa9: Harvesting Memory-bound CPU Stall Cycles in Software with MSH. OSDI 2024.


논문 · 확인 범위: 기록 안 됨 · S3ResearchAgent · 2026-07-16T14:56:32.987483Z
Bibliographic paper record.



근거 verified-content-v1-0059: Zhihong Luo et al., "Harvesting Memory-bound CPU Stall Cycles in Software with MSH", OSDI 2024. (원문 열기)
논문 · 확인 범위: 기록 안 됨 · S3ResearchAgent · 2026-07-16T18:41:27.843495Z
Verification: official_abstract; confidence=high. Question: Can software safely run useful work during memory-bound CPU stalls when SMT violates latency SLOs? Context: SMT harvests stalls but offers coarse concurrency control and can interfere with latency-critical work. Method: MSH combines profiling, program analysis, binary instrumentation, and a runtime scheduler with configurable harvesting concurrency. Evaluation: workloads=memory-bound applications; names not enumerated in abstract; baselines=SMT; no harvesting; metrics=harvesting throughput; latency SLO compliance; results=up to 72% of SMT harvesting throughput under SLOs where SMT is unusable Interpretation: Explicit software yield points trade some peak harvest for much tighter interference control. Reusable lesson: Expose resource-stall opportunities to a policy-controlled runtime when hardware sharing is too coarse. Applicability: Memory-bound server workloads with latency SLOs and spare CPU cycles. Limits: Requires analyzable/instrumentable binaries and enough predictable memory stalls; abstract does not enumerate all workloads.



근거 canonical-paper-v2-5866da08: Zhihong Luo et al., "Harvesting Memory-bound CPU Stall Cycles in Software with MSH", OSDI 2024. (원문 열기)
논문 · 확인 범위: 공식 초록 확인 · S3ResearchAgent · 2026-07-18T05:23:42.412972Z
Verification: official_abstract; confidence=medium. Question: Can software safely run useful work during memory-bound CPU stalls when SMT violates latency SLOs? Context: SMT harvests stalls but offers coarse concurrency control and can interfere with latency-critical work. Method: MSH combines profiling, program analysis, binary instrumentation, and a runtime scheduler with configurable harvesting concurrency. Evaluation: workloads=memory-bound applications; names not enumerated in abstract; baselines=SMT; no harvesting; metrics=harvesting throughput; latency SLO compliance; results=up to 72% of SMT harvesting throughput under SLOs where SMT is unusable Interpretation: Explicit software yield points trade some peak harvest for much tighter interference control. Reusable lesson: Expose resource-stall opportunities to a policy-controlled runtime when hardware sharing is too coarse. Applicability: Memory-bound server workloads with latency SLOs and spare CPU cycles. Limits: Requires analyzable/instrumentable binaries and enough predictable memory stalls; abstract does not enumerate all workloads.



자료 검증 verify_34f5049c1f180a60f6a3: ev_c05dfb3aec3d4aa9 · 판단 보류
확인 범위: 일부 자료 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:34.385177Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=7d3a3229d1035afccb1510abb475f572d7643eea157ed770bc1c2aaf0206123c / 위치: 보존 파일 objects/sha256/7d/7d3a3229d1035afccb1510abb475f572d7643eea157ed770bc1c2aaf0206123c
보존 원문 객체를 확보했으나 이 일괄 검증에서는 claim-bearing 범위를 재판정하지 않아 결론을 보류함.



자료 검증 verify_f36a6659763f85cf4468: verified-content-v1-0059 · 판단 보류
확인 범위: 일부 자료 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:34.528315Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=7d3a3229d1035afccb1510abb475f572d7643eea157ed770bc1c2aaf0206123c / 위치: 보존 파일 objects/sha256/7d/7d3a3229d1035afccb1510abb475f572d7643eea157ed770bc1c2aaf0206123c
보존 원문 객체를 확보했으나 이 일괄 검증에서는 claim-bearing 범위를 재판정하지 않아 결론을 보류함.



자료 검증 verify_2c6dcb6b5ff1b153eceb: canonical-paper-v2-5866da08 · 판단 보류
확인 범위: 일부 자료 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:34.737498Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=7d3a3229d1035afccb1510abb475f572d7643eea157ed770bc1c2aaf0206123c / 위치: 보존 파일 objects/sha256/7d/7d3a3229d1035afccb1510abb475f572d7643eea157ed770bc1c2aaf0206123c
보존 원문 객체를 확보했으나 이 일괄 검증에서는 claim-bearing 범위를 재판정하지 않아 결론을 보류함.