🔥 A curated list of DL cluster scheduling papers.
Please feel free to pull requests or open an issue to add papers.
| Scheduler | Year | Series | Paper | Objective | Heter. | Elastic | AutoML | Code |
|---|---|---|---|---|---|---|---|---|
| Acme | 2024 | NSDI | Paper | ♣ | - | - | - | Code |
| Cassini | 2024 | NSDI | Paper | ♠♣♥ | - | ✔ | - | - |
| Sia | 2023 | SOSP | Paper | ♠♣♥ | ✔ | ✔ | ✔ | Code |
| EasyScale | 2023 | SC | Paper | ♠♣ | ✔ | ✔ | - | Code |
| Hydro | 2023 | OSDI | Paper | ♠▲ | ✔ | ✔ | ✔ | Code |
| Shockwave | 2023 | NSDI | Paper | ♠♣♥ | - | ✔ | - | Code |
| ModelKeeper | 2023 | NSDI | Paper | ♦♣ | - | - | ✔ | Code |
| Lyra | 2023 | EuroSys | Paper | ♠♣ | ✔ | ✔ | - | - |
| SiloD | 2023 | EuroSys | Paper | ♠♣♥ | ✔* | - | - | - |
| FGD | 2023 | ATC | Paper | ♠♣ | - | - | - | Code |
| ElasticFlow | 2023 | ASPLOS | Paper | ♣✿ | - | ✔ | - | Code |
| Lucid | 2023 | ASPLOS | Paper | ♠♣ | - | - | - | Code |
| PowerFlow | 2023 | arxiv | Paper | ♦♣ | - | ✔ | - | - |
| EDL | 2022 | TPDS | Paper | ♠♣ | - | ✔ | - | - |
| AOnline | 2022 | TCC | Paper | ♠♣ | - | ✔ | - | - |
| Titan | 2022 | SoCC | Paper | ♠♣ | - | - | - | - |
| Muri | 2022 | SIGCOMM | Paper | ♠♣ | ✔* | - | - | Code |
| Synergy | 2022 | OSDI | Paper | ♣ | - | - | - | Code |
| Ali-MLaaS | 2022 | NSDI | Paper | ♠♣ | - | - | - | Code |
| GADGET | 2022 | INFOCOM | Paper | ♠♣ | - | ✔ | - | Code |
| CloudBrain | 2022 | ICCD | Paper | ♠ | - | - | - | Code |
| Aryl | 2022 | arxiv | Paper | ♠♣ | - | ✔ | ✔ | - |
| Singularity | 2022 | arxiv | Paper | ♠♣♦ | - | ✔ | - | - |
| 2021 | TPDS | Paper | ♣ | - | ✔ | - | Code | |
| Astraea | 2021 | TPDS | Paper | ♥ | - | - | - | Code |
| Horus | 2021 | TPDS | Paper | ♠♣ | - | - | - | - |
| Liquid | 2021 | TPDS | Paper | ♣ | - | - | - | Code |
| POP | 2021 | SOSP | Paper | ♥♣ | ✔ | - | - | Code |
| Chronus | 2021 | SoCC | Paper | ✿ | - | - | - | Code |
| SEER | 2021 | SoCC | Paper | ▲ | - | ✔ | ✔ | - |
| Helios | 2021 | SC | Paper | ♣♦ | - | - | - | Code |
| ONES | 2021 | SC | Paper | ♠♣ | - | ✔ | - | Code |
| Pollux | 2021 | OSDI | Paper | ♠♣♥ | - | ✔ | ✔ | Code |
| AFS | 2021 | NSDI | Paper | ♠♣ | - | ✔ | - | - |
| SMD | 2021 | INFOCOM | Paper | ♣ | - | - | - | - |
| ANDREAS | 2021 | FCloud | Paper | ♦ | - | - | - | - |
| RubberBand | 2021 | EuroSys | Paper | ♦ | - | ✔ | ✔ | - |
| Hermes | 2021 | Electronics | Paper | ♣ | - | - | ✔ | - |
| Jigsaw | 2021 | DistributedML | Paper | ♣ | - | - | - | - |
| DynamoML | 2021 | CLOSER | Paper | ♠♣ | - | ✔ | - | - |
| GENIE | 2020 | TPDS | Paper | ✿ | - | ✔ | - | - |
| Parrot | 2020 | TCC | Paper | ♣ | - | - | - | - |
| Non-Intrusive | 2020 | SC | Paper | ♠♣ | - | ✔ | - | - |
| Antman | 2020 | OSDI | Paper | ♠♣ | - | ✔ | - | Code |
| Gavel | 2020 | OSDI | Paper | ♣♥ | ✔ | - | - | Code |
| HiveD | 2020 | OSDI | Paper | ♣ | - | - | - | Code |
| Themis | 2020 | NSDI | Paper | ♥ | - | - | - | - |
| Salus | 2020 | MLSys | Paper | ♠♣ | - | - | - | Code |
| Vaibhav et al. | 2020 | MASCOTS | Paper | ♠♣ | - | ✔ | - | - |
| SPIN | 2020 | INFOCOM | Paper | ♣ | - | - | - | - |
| E-LAS | 2020 | ICPP | Paper | ♣ | - | - | - | - |
| CODA | 2020 | ICDCS | Paper | ♣ | ✔* | - | - | - |
| Elan | 2020 | ICDCS | Paper | ♠♣ | - | ✔ | - | - |
| Yeung | 2020 | HotCloud | Paper | ♠ | - | - | - | - |
| 2020 | EuroSys | Paper | ♥♣ | ✔ | - | - | - | |
| MLCloudPrice | 2020 | DISPA | Paper | ♣♦ | - | - | - | Code |
| MLFS | 2020 | CoNext | Paper | ♣✿ | - | - | - | Code |
| MARBLE | 2020 | CCGRID | Paper | ♠♣ | - | ✔ | - | - |
| Ada-SRSF | 2020 | arxiv | Paper | ♣ | ✔* | - | - | - |
| Co-scheML | 2020 | ACSOS | Paper | ♣ | - | - | - | - |
| HyperSched | 2019 | SoCC | Paper | ✿ ▲ | - | ✔ | ✔ | - |
| Tiresias | 2019 | NSDI | Paper | ♣ | - | - | - | Code |
| FfDL | 2019 | Middleware | Paper | ♣ | - | - | - | Code |
| JPAS | 2019 | JNCA | Paper | ♣▲ | - | - | ✔ | - |
| Harmony | 2019 | INFOCOM | Paper | ♣ | - | - | - | - |
| Cynthia | 2019 | ICPP | Paper | ♦ | - | ✔ | - | - |
| Jahani | 2019 | ICCCS | Paper | ♦ | ✔ | ✔ | - | - |
| 2019 | GLOBECOM | Paper | ♣ | - | - | - | - | |
| Dragon | 2019 | CLOSER | Paper | ♠♣ | - | ✔ | - | - |
| 2019 | CC | Paper | ♦ | ✔* | ✔ | - | - | |
| Philly | 2019 | ATC | Paper | ♣ | - | - | - | Code |
| Gandiva | 2018 | OSDI | Paper | ♠♣ | - | ✔ | ✔ | - |
| OASiS | 2018 | INFOCOM | Paper | ♠♣ | - | ✔ | - | - |
| Optimus | 2018 | EuroSys | Paper | ♣ | - | ✔ | - | Code |
| Dorm | 2017 | SMARTCOMP | Paper | ♥ | - | - | - | - |
| Topology-Aware | 2017 | SC | Paper | ♣ | - | - | - | Code |
| HyperDrive | 2017 | Middleware | Paper | ♣▲ | - | - | ✔ | - |
Symbols of Training Schedulers:
| JCT | Utilization | Cost | Fairness | DDL | Accuracy |
|---|---|---|---|---|---|
| ♣ | ♠ | ♦ | ♥ | ✿ | ▲ |
| Scheduler | Year | Series | Paper | Objective | Batch | Share | Cloud | Source Code |
|---|---|---|---|---|---|---|---|---|
| SpotServe | 2024 | ASPLOS | Paper | ♦♥ | ✔ | - | ✔ | Code |
| DeltaZip | 2023 | arxiv | Paper | ♥♠ | ✔ | - | - | Code |
| MOSEL | 2023 | arxiv | Paper | ♦♠ | ✔ | - | - | - |
| Punica | 2023 | arxiv | Paper | ♥♠ | - | ✔* | - | Code |
| S-LoRA | 2023 | arxiv | Paper | ♥♠ | - | ✔* | - | Code |
| Symphony | 2023 | arxiv | Paper | ✿♠♦ | ✔ | - | - | - |
| DeepPlan | 2023 | EuroSys | Paper | ♦♠ | ✔ | - | - | Code |
| Tabi | 2023 | EuroSys | Paper | ♦♣ | - | - | ✔ | - |
| Kairos | 2023 | HPDC | Paper | ♦♥♠ | ✔ | - | ✔ | Code |
| Shepherd | 2023 | NSDI | Paper | ✿♠♦ | ✔ | - | - | - |
| AlpaServe | 2023 | OSDI | Paper | ♦♠ | ✔ | ✔ | - | Code |
| Clover | 2023 | SC | Paper | ♥ | ✔ | ✔ | - | Code |
| iGniter | 2023 | TPDS | Paper | ✿♥ | ✔ | ✔ | ✔ | Code |
| Gpulet | 2022 | ATC | Paper | ✿♠♥ | ✔ | ✔ | - | Code |
| Cocktail | 2022 | NSDI | Paper | ♣♦♥ | - | - | ✔ | Code |
| INFaaS | 2021 | ATC | Paper | ♦♥♠ | - | ✔ | ✔ | Code |
| MIG-SERVING | 2021 | CoRR | Paper | ♦♥ | ✔ | ✔ | - | - |
| Mendoza et al. | 2021 | EuroMLSys | Paper | ♦ | - | ✔ | - | - |
| Abacus | 2021 | SC | Paper | ♦♠ | - | ✔ | - | Code |
| Morphling | 2021 | SoCC | Paper | ♥♠ | ✔ | ✔ | ✔ | Code |
| Irina | 2020 | APNet | Paper | ♦♠✿ | ✔ | ✔ | - | - |
| DyBatch | 2020 | CCGrid | Paper | ♦♠ | ✔ | ✔ | - | - |
| CMS | 2020 | Future Internet | Paper | ♣✿ | - | - | - | - |
| PERSEUS | 2020 | IC2E | Paper | ♦♥♠ | ✔ | - | ✔ | Code |
| AutoDeep | 2020 | Infocom | Paper | ♦♥♠ | - | ✔ | ✔ | - |
| Clockwork | 2020 | OSDI | Paper | ♦♠ | ✔ | - | - | Code |
| GSLICE | 2020 | SoCC | Paper | ♠✿ | ✔ | ✔ | - | - |
| Inferline | 2020 | SoCC | Paper | ♦♥ | ✔ | - | ✔ | Code |
| MArk | 2019 | ATC | Paper | ♦♥ | ✔ | - | ✔ | Code |
| TrIMS | 2019 | CLOUD | Paper | ♦♠✿ | ✔ | ✔ | ✔ | Code |
| Kube-Knots | 2019 | CLUSTER | Paper | ♦✿ | ✔ | ✔ | - | - |
| Gilman et al. | 2019 | DIDL | Paper | ♦♠ | - | ✔ | - | - |
| Nanily | 2019 | HPCC | Paper | ♦♠ | ✔ | - | - | - |
| Ebird | 2019 | ICCD | Paper | ♦♠✿ | ✔ | ✔ | - | Code |
| Tolerance Tiers | 2019 | ISPASS | Paper | ♣♦♥ | - | - | ✔ | - |
| RRL | 2019 | SC | Paper | ♦ | ✔ | ✔ | - | Code |
| ParM | 2019 | SOSP | Paper | ♦ | ✔ | - | - | Code |
| HiveMind | 2018 | NIPS | Paper | ♠ | ✔ | ✔ | - | - |
| Space-Time | 2018 | NIPS | Paper | ♠✿ | ✔ | ✔ | - | - |
| Ease.ml | 2018 | VLDB | Paper | ♣ | - | - | - | Code |
| Rafiki | 2018 | VLDB | Paper | ♣♦ | ✔ | - | - | Code |
| Clipper | 2017 | NSDI | Paper | ♣♦♠ | ✔ | - | - | Code |
Symbols of Inference Schedulers:
| Accuracy | Throughput | Latency | Cost | Utilization |
|---|---|---|---|---|
| ♣ | ♠ | ♦ | ♥ | ✿ |
| Terminology | Definition |
|---|---|
| JCT | Job Completion Time (Job Finish Time - Job Submission Time) |
| Fairness | a metric to assess whether resources are fairly shared among users or jobs |
| QoS | Quality of Service |
| DDL | Deadline, a time point where DL job must be completed |
| SLO | Service Level Objective |