속성:Context

하드웨어, 부하, 버전, 규모처럼 결과에 영향을 줄 수 있는 조건을 적습니다.

사용362

필터<p><a rel="nofollow" class="external text" href="https://www.semantic-mediawiki.org/wiki/Help:Property_page/Filter">검색 필터</a>를 사용하면 <code>~</code> 또는 <code>!</code> 같은 <a rel="nofollow" class="external text" href="https://www.semantic-mediawiki.org/wiki/Help:Query_expressions">쿼리 표현식</a>을 포함할 수 있습니다. 선택한 <a rel="nofollow" class="external text" href="https://www.semantic-mediawiki.org/wiki/Query_engine">쿼리 엔진</a>은 대소문자를 구분하지 않는 일치 또는 다음과 같은 기타 짧은 표현식도 지원할 수 있습니다:</p><ul><li><code>in:</code> 결과는 해당 용어를 포함해야 합니다. 예: '<code>in:Foo</code>'</li></ul><ul><li><code>not:</code> 결과는 해당 용어를 포함하지 않아야 합니다. 예: '<code>not:Bar</code>'</li></ul>

이 속성을 사용하는 문서 20개를 보여줍니다.

o

orca a distributed serving system for transformer based generative models 6bd2fa60 +

Venue: OSDI. Year: 2022. 요청 단위 배칭은 먼저 끝난 요청 때문에 GPU를 낭비하고 head-of-line blocking을 만든다. Verification: abstract_only; confidence=high. +

overcoming the memory wall with cxl enabled ssds ef2155d6 +

Venue: USENIX ATC. Year: 2023. DRAM 확장은 비싸고 플래시는 느리며 쓰기 수명 제한이 있다. Verification: abstract_only; confidence=high. +

ozz identifying kernel out of order concurrency bugs with in vivo memory access reordering 0b85f639 +

Venue: SOSP. Year: 2024. Incorrect memory-barrier use creates rare concurrency failures that ordinary schedules and hardware execution reproduce unreliably. Verification: official_abstract; confidence=high. +

p

pacar improved buffered i o locality on numa systems with page cache replication adc0c538 +

Venue: EuroSys. Year: 2026. A single page-cache copy can force remote NUMA accesses for file-heavy multithreaded workloads. Verification: official DOI/EuroSys metadata and author project full-text summary; confidence=high. +

pact a criticality first design for tiered memory 20db3445 +

Venue: ASPLOS. Year: 2026. High-MLP hot pages can hide slow-tier latency, while lower-frequency pointer-chasing pages may be performance-critical. Verification: publication metadata is preserved, but the current fetched object is a landing/cookie page; claim-bearing full text must be reattached before raising confidence; confidence=low. +

paella low latency model serving with software defined gpu scheduling 47b175ce +

Venue: SOSP. Year: 2023. 동시 모델의 커널 순서를 GPU가 불투명하게 결정해 SLO 제어가 어렵다. Verification: abstract_only; confidence=medium. +

papi exploiting dynamic parallelism in large language model decoding with a processing in memory 5f9f53cc +

Venue: ASPLOS. Year: 2025. A fixed placement wastes either GPU compute or PIM bandwidth across different inference kernels. Verification: official DOI metadata and arXiv abstract; confidence=high. +

partial failure resilient memory management system for cxl based distributed shared memory 48948c1d +

Venue: SOSP. Year: 2023. 클라이언트 장애는 refcount 누수·중복 해제·wild pointer를 만들 수 있다. Verification: abstract_only; confidence=high. +

pegasus tolerating skewed workloads in distributed storage with in network coherence directories a4e8c8ec +

Venue: OSDI. Year: 2020. 고정 파티셔닝은 소수 핫 객체가 특정 서버를 포화시키면 확장성이 무너진다. Verification: abstract_only; confidence=high. +

performance optimization of object tracking algorithms in opencv on gpus 8158d2ec +

Publication scope: international. Lab publication metadata: 1. Jaehyun Song, Hwanjin Jeong, Jinkyu Jeong, "Performance Optimization of Object Tracking Algorithms in OpenCV on GPUs," Applied Sciences, Vol. 12, Issue 15, pp. 7801, 2022, doi.org/10.3390/app12157801 Verification level: full_text. Sources checked: https://www.mdpi.com/2076-3417/12/15/7801 https://doi.org/10.3390/app12157801 +

perseus a fail slow detection framework for cloud storage system 4e24f100 +

Venue: FAST. Year: 2023. fail-slow는 정상 응답을 하면서 꼬리 지연을 키워 단순 장애 탐지로 잡기 어렵다. Verification: abstract_only; confidence=high. +

pet proactive demotion for efficient tiered memory management e2185da3 +

Venue: EuroSys. Year: 2025. Page-granular hotness can retain unused bytes in scarce fast memory and cause capacity-cliff slowdowns. Verification: official DOI metadata and accessible publisher abstract; confidence=high. +

phasedrr read reclaim scheduling without page level access counting 69578f95 +

Venue: MSST. Year: 2024. 기존 hot-page 선이동은 DRAM의 page counter가 캐시 공간을 잠식한다. Verification: full_text; confidence=high. +

pit optimization of dynamic sparse deep learning models via permutation invariant transformation 48509ad0 +

Venue: SOSP. Year: 2023. 동적 희소성은 정적 컴파일·일반 sparse kernel 모두에 불규칙성을 만든다. Verification: abstract_only; confidence=high. +

pivot-b200-20260717-immutable-runtime-a01 +

Three-policy pivot target-generation migration collector, source commit 038c9cd00c25fc83678d4d698f8780bfc5d96369, intended for sequential Qwen then Mistral collection on B200-2. +

pivot-b200-20260717-launch-lock-a01 +

All 8 B200 GPUs were idle before launch. The runner acquired experiments/runs/.weighted_slo_final.global.lock, then immediately reran its clean-tree gate before model preflight or GPU initialization. +

pivot-b200-20260717-native-extension-boundary-a01 +

Sequential Qwen then Mistral target-generation migration collection on B200-2, source commit 8ca2dbb68d202a8ba8b4ae226b64d77ec3e604ca, before any CUDA generation process was launched. +

pivot-b200-20260717-portability-preflight-a01 +

Repo dw-kv; protocol weighted_slo_final_v1 sha256=528c4a83e8dac516ab6898f9b73b652c348c14d9eb5c8a2eea1e965e50949c10; host B200-2; 8x NVIDIA B200 all 0 MiB/0% and no compute processes; runtime torch 2.11.0+cu130, CUDA 13.0, vLLM 0.21.0; pinned Qwen and Mistral snapshots available; global concurrency remains 1. +

pivot-b200-20260717-portable-pipeline-hash-a01 +

B200-to-L40S or RTX PRO migration of target-model length calibration artifacts before any final performance experiment. +

pivot-b200-20260717-tmux-argument-boundary-a01 +

B200-2 sequential migration launch at source commit 038c9cd00c25fc83678d4d698f8780bfc5d96369. +