S-Lab-System-Group/Awesome-DL-Scheduling-Papers

335

14 commits

updated Jan 22, 2024

See the code

README

Awesome-DL-Scheduling-Papers

πŸ”₯ A curated list of DL cluster scheduling papers.

Please feel free to pull requests or open an issue to add papers.

Table of Content

Schedulers for DL Training

SchedulerYearSeriesPaperObjectiveHeter.ElasticAutoMLCode
Acme2024NSDIPaper♣---Code
Cassini2024NSDIPaper♠♣β™₯-βœ”--
Sia2023SOSPPaper♠♣β™₯βœ”βœ”βœ”Code
EasyScale2023SCPaperβ™ β™£βœ”βœ”-Code
Hydro2023OSDIPaperβ™ β–²βœ”βœ”βœ”Code
Shockwave2023NSDIPaper♠♣β™₯-βœ”-Code
ModelKeeper2023NSDIPaper♦♣--βœ”Code
Lyra2023EuroSysPaperβ™ β™£βœ”βœ”--
SiloD2023EuroSysPaper♠♣β™₯βœ”*---
FGD2023ATCPaper♠♣---Code
ElasticFlow2023ASPLOSPaperβ™£βœΏ-βœ”-Code
Lucid2023ASPLOSPaper♠♣---Code
PowerFlow2023arxivPaper♦♣-βœ”--
EDL2022TPDSPaper♠♣-βœ”--
AOnline2022TCCPaper♠♣-βœ”--
Titan2022SoCCPaper♠♣----
Muri2022SIGCOMMPaperβ™ β™£βœ”*--Code
Synergy2022OSDIPaper♣---Code
Ali-MLaaS2022NSDIPaper♠♣---Code
GADGET2022INFOCOMPaper♠♣-βœ”-Code
CloudBrain2022ICCDPaperβ™ ---Code
Aryl2022arxivPaper♠♣-βœ”βœ”-
Singularity2022arxivPaper♠♣♦-βœ”--
$DL^2$2021TPDSPaper♣-βœ”-Code
Astraea2021TPDSPaperβ™₯---Code
Horus2021TPDSPaper♠♣----
Liquid2021TPDSPaper♣---Code
POP2021SOSPPaperβ™₯β™£βœ”--Code
Chronus2021SoCCPaper✿---Code
SEER2021SoCCPaperβ–²-βœ”βœ”-
Helios2021SCPaper♣♦---Code
ONES2021SCPaper♠♣-βœ”-Code
Pollux2021OSDIPaper♠♣β™₯-βœ”βœ”Code
AFS2021NSDIPaper♠♣-βœ”--
SMD2021INFOCOMPaper♣----
ANDREAS2021FCloudPaper♦----
RubberBand2021EuroSysPaper♦-βœ”βœ”-
Hermes2021ElectronicsPaper♣--βœ”-
Jigsaw2021DistributedMLPaper♣----
DynamoML2021CLOSERPaper♠♣-βœ”--
GENIE2020TPDSPaper✿-βœ”--
Parrot2020TCCPaper♣----
Non-Intrusive2020SCPaper♠♣-βœ”--
Antman2020OSDIPaper♠♣-βœ”-Code
Gavel2020OSDIPaper♣β™₯βœ”--Code
HiveD2020OSDIPaper♣---Code
Themis2020NSDIPaperβ™₯----
Salus2020MLSysPaper♠♣---Code
Vaibhav et al.2020MASCOTSPaper♠♣-βœ”--
SPIN2020INFOCOMPaper♣----
E-LAS2020ICPPPaper♣----
CODA2020ICDCSPaperβ™£βœ”*---
Elan2020ICDCSPaper♠♣-βœ”--
Yeung2020HotCloudPaperβ™ ----
$Gandiva_{fair}$2020EuroSysPaperβ™₯β™£βœ”---
MLCloudPrice2020DISPAPaper♣♦---Code
MLFS2020CoNextPaperβ™£βœΏ---Code
MARBLE2020CCGRIDPaper♠♣-βœ”--
Ada-SRSF2020arxivPaperβ™£βœ”*---
Co-scheML2020ACSOSPaper♣----
HyperSched2019SoCCPaper✿ β–²-βœ”βœ”-
Tiresias2019NSDIPaper♣---Code
FfDL2019MiddlewarePaper♣---Code
JPAS2019JNCAPaper♣▲--βœ”-
Harmony2019INFOCOMPaper♣----
Cynthia2019ICPPPaper♦-βœ”--
Jahani2019ICCCSPaperβ™¦βœ”βœ”--
$Sched^2$2019GLOBECOMPaper♣----
Dragon2019CLOSERPaper♠♣-βœ”--
$FC^2$2019CCPaperβ™¦βœ”*βœ”--
Philly2019ATCPaper♣---Code
Gandiva2018OSDIPaper♠♣-βœ”βœ”-
OASiS2018INFOCOMPaper♠♣-βœ”--
Optimus2018EuroSysPaper♣-βœ”-Code
Dorm2017SMARTCOMPPaperβ™₯----
Topology-Aware2017SCPaper♣---Code
HyperDrive2017MiddlewarePaper♣▲--βœ”-

Symbols of Training Schedulers:

JCTUtilizationCostFairnessDDLAccuracy
♣♠♦β™₯βœΏβ–²

Schedulers for DL Inference

SchedulerYearSeriesPaperObjectiveBatchShareCloudSource Code
SpotServe2024ASPLOSPaper♦β™₯βœ”-βœ”Code
DeltaZip2023arxivPaperβ™₯β™ βœ”--Code
MOSEL2023arxivPaperβ™¦β™ βœ”---
Punica2023arxivPaperβ™₯β™ -βœ”*-Code
S-LoRA2023arxivPaperβ™₯β™ -βœ”*-Code
Symphony2023arxivPaperβœΏβ™ β™¦βœ”---
DeepPlan2023EuroSysPaperβ™¦β™ βœ”--Code
Tabi2023EuroSysPaper♦♣--βœ”-
Kairos2023HPDCPaper♦β™₯β™ βœ”-βœ”Code
Shepherd2023NSDIPaperβœΏβ™ β™¦βœ”---
AlpaServe2023OSDIPaperβ™¦β™ βœ”βœ”-Code
Clover2023SCPaperβ™₯βœ”βœ”-Code
iGniter2023TPDSPaper✿β™₯βœ”βœ”βœ”Code
Gpulet2022ATCPaperβœΏβ™ β™₯βœ”βœ”-Code
Cocktail2022NSDIPaper♣♦β™₯--βœ”Code
INFaaS2021ATCPaper♦β™₯β™ -βœ”βœ”Code
MIG-SERVING2021CoRRPaper♦β™₯βœ”βœ”--
Mendoza et al.2021EuroMLSysPaper♦-βœ”--
Abacus2021SCPaper♦♠-βœ”-Code
Morphling2021SoCCPaperβ™₯β™ βœ”βœ”βœ”Code
Irina2020APNetPaperβ™¦β™ βœΏβœ”βœ”--
DyBatch2020CCGridPaperβ™¦β™ βœ”βœ”--
CMS2020Future InternetPaperβ™£βœΏ----
PERSEUS2020IC2EPaper♦β™₯β™ βœ”-βœ”Code
AutoDeep2020InfocomPaper♦β™₯β™ -βœ”βœ”-
Clockwork2020OSDIPaperβ™¦β™ βœ”--Code
GSLICE2020SoCCPaperβ™ βœΏβœ”βœ”--
Inferline2020SoCCPaper♦β™₯βœ”-βœ”Code
MArk2019ATCPaper♦β™₯βœ”-βœ”Code
TrIMS2019CLOUDPaperβ™¦β™ βœΏβœ”βœ”βœ”Code
Kube-Knots2019CLUSTERPaperβ™¦βœΏβœ”βœ”--
Gilman et al.2019DIDLPaper♦♠-βœ”--
Nanily2019HPCCPaperβ™¦β™ βœ”---
Ebird2019ICCDPaperβ™¦β™ βœΏβœ”βœ”-Code
Tolerance Tiers2019ISPASSPaper♣♦β™₯--βœ”-
RRL2019SCPaperβ™¦βœ”βœ”-Code
ParM2019SOSPPaperβ™¦βœ”--Code
HiveMind2018NIPSPaperβ™ βœ”βœ”--
Space-Time2018NIPSPaperβ™ βœΏβœ”βœ”--
Ease.ml2018VLDBPaper♣---Code
Rafiki2018VLDBPaperβ™£β™¦βœ”--Code
Clipper2017NSDIPaperβ™£β™¦β™ βœ”--Code

Symbols of Inference Schedulers:

AccuracyThroughputLatencyCostUtilization
♣♠♦β™₯✿

Glossary of Terms

TerminologyDefinition
JCTJob Completion Time (Job Finish Time - Job Submission Time)
Fairnessa metric to assess whether resources are fairly shared among users or jobs
QoSQuality of Service
DDLDeadline, a time point where DL job must be completed
SLOService Level Objective

Contributors

Qinghao-Hu

10 commits

gaow0007

3 commits

smile-luobin

1 commits

S-Lab-System-Group/Awesome-DL-Scheduling-Papers

335

14 commits

updated Jan 22, 2024

See the code

README

Awesome-DL-Scheduling-Papers

πŸ”₯ A curated list of DL cluster scheduling papers.

Please feel free to pull requests or open an issue to add papers.

Table of Content

Schedulers for DL Training

SchedulerYearSeriesPaperObjectiveHeter.ElasticAutoMLCode
Acme2024NSDIPaper♣---Code
Cassini2024NSDIPaper♠♣β™₯-βœ”--
Sia2023SOSPPaper♠♣β™₯βœ”βœ”βœ”Code
EasyScale2023SCPaperβ™ β™£βœ”βœ”-Code
Hydro2023OSDIPaperβ™ β–²βœ”βœ”βœ”Code
Shockwave2023NSDIPaper♠♣β™₯-βœ”-Code
ModelKeeper2023NSDIPaper♦♣--βœ”Code
Lyra2023EuroSysPaperβ™ β™£βœ”βœ”--
SiloD2023EuroSysPaper♠♣β™₯βœ”*---
FGD2023ATCPaper♠♣---Code
ElasticFlow2023ASPLOSPaperβ™£βœΏ-βœ”-Code
Lucid2023ASPLOSPaper♠♣---Code
PowerFlow2023arxivPaper♦♣-βœ”--
EDL2022TPDSPaper♠♣-βœ”--
AOnline2022TCCPaper♠♣-βœ”--
Titan2022SoCCPaper♠♣----
Muri2022SIGCOMMPaperβ™ β™£βœ”*--Code
Synergy2022OSDIPaper♣---Code
Ali-MLaaS2022NSDIPaper♠♣---Code
GADGET2022INFOCOMPaper♠♣-βœ”-Code
CloudBrain2022ICCDPaperβ™ ---Code
Aryl2022arxivPaper♠♣-βœ”βœ”-
Singularity2022arxivPaper♠♣♦-βœ”--
$DL^2$2021TPDSPaper♣-βœ”-Code
Astraea2021TPDSPaperβ™₯---Code
Horus2021TPDSPaper♠♣----
Liquid2021TPDSPaper♣---Code
POP2021SOSPPaperβ™₯β™£βœ”--Code
Chronus2021SoCCPaper✿---Code
SEER2021SoCCPaperβ–²-βœ”βœ”-
Helios2021SCPaper♣♦---Code
ONES2021SCPaper♠♣-βœ”-Code
Pollux2021OSDIPaper♠♣β™₯-βœ”βœ”Code
AFS2021NSDIPaper♠♣-βœ”--
SMD2021INFOCOMPaper♣----
ANDREAS2021FCloudPaper♦----
RubberBand2021EuroSysPaper♦-βœ”βœ”-
Hermes2021ElectronicsPaper♣--βœ”-
Jigsaw2021DistributedMLPaper♣----
DynamoML2021CLOSERPaper♠♣-βœ”--
GENIE2020TPDSPaper✿-βœ”--
Parrot2020TCCPaper♣----
Non-Intrusive2020SCPaper♠♣-βœ”--
Antman2020OSDIPaper♠♣-βœ”-Code
Gavel2020OSDIPaper♣β™₯βœ”--Code
HiveD2020OSDIPaper♣---Code
Themis2020NSDIPaperβ™₯----
Salus2020MLSysPaper♠♣---Code
Vaibhav et al.2020MASCOTSPaper♠♣-βœ”--
SPIN2020INFOCOMPaper♣----
E-LAS2020ICPPPaper♣----
CODA2020ICDCSPaperβ™£βœ”*---
Elan2020ICDCSPaper♠♣-βœ”--
Yeung2020HotCloudPaperβ™ ----
$Gandiva_{fair}$2020EuroSysPaperβ™₯β™£βœ”---
MLCloudPrice2020DISPAPaper♣♦---Code
MLFS2020CoNextPaperβ™£βœΏ---Code
MARBLE2020CCGRIDPaper♠♣-βœ”--
Ada-SRSF2020arxivPaperβ™£βœ”*---
Co-scheML2020ACSOSPaper♣----
HyperSched2019SoCCPaper✿ β–²-βœ”βœ”-
Tiresias2019NSDIPaper♣---Code
FfDL2019MiddlewarePaper♣---Code
JPAS2019JNCAPaper♣▲--βœ”-
Harmony2019INFOCOMPaper♣----
Cynthia2019ICPPPaper♦-βœ”--
Jahani2019ICCCSPaperβ™¦βœ”βœ”--
$Sched^2$2019GLOBECOMPaper♣----
Dragon2019CLOSERPaper♠♣-βœ”--
$FC^2$2019CCPaperβ™¦βœ”*βœ”--
Philly2019ATCPaper♣---Code
Gandiva2018OSDIPaper♠♣-βœ”βœ”-
OASiS2018INFOCOMPaper♠♣-βœ”--
Optimus2018EuroSysPaper♣-βœ”-Code
Dorm2017SMARTCOMPPaperβ™₯----
Topology-Aware2017SCPaper♣---Code
HyperDrive2017MiddlewarePaper♣▲--βœ”-

Symbols of Training Schedulers:

JCTUtilizationCostFairnessDDLAccuracy
♣♠♦β™₯βœΏβ–²

Schedulers for DL Inference

SchedulerYearSeriesPaperObjectiveBatchShareCloudSource Code
SpotServe2024ASPLOSPaper♦β™₯βœ”-βœ”Code
DeltaZip2023arxivPaperβ™₯β™ βœ”--Code
MOSEL2023arxivPaperβ™¦β™ βœ”---
Punica2023arxivPaperβ™₯β™ -βœ”*-Code
S-LoRA2023arxivPaperβ™₯β™ -βœ”*-Code
Symphony2023arxivPaperβœΏβ™ β™¦βœ”---
DeepPlan2023EuroSysPaperβ™¦β™ βœ”--Code
Tabi2023EuroSysPaper♦♣--βœ”-
Kairos2023HPDCPaper♦β™₯β™ βœ”-βœ”Code
Shepherd2023NSDIPaperβœΏβ™ β™¦βœ”---
AlpaServe2023OSDIPaperβ™¦β™ βœ”βœ”-Code
Clover2023SCPaperβ™₯βœ”βœ”-Code
iGniter2023TPDSPaper✿β™₯βœ”βœ”βœ”Code
Gpulet2022ATCPaperβœΏβ™ β™₯βœ”βœ”-Code
Cocktail2022NSDIPaper♣♦β™₯--βœ”Code
INFaaS2021ATCPaper♦β™₯β™ -βœ”βœ”Code
MIG-SERVING2021CoRRPaper♦β™₯βœ”βœ”--
Mendoza et al.2021EuroMLSysPaper♦-βœ”--
Abacus2021SCPaper♦♠-βœ”-Code
Morphling2021SoCCPaperβ™₯β™ βœ”βœ”βœ”Code
Irina2020APNetPaperβ™¦β™ βœΏβœ”βœ”--
DyBatch2020CCGridPaperβ™¦β™ βœ”βœ”--
CMS2020Future InternetPaperβ™£βœΏ----
PERSEUS2020IC2EPaper♦β™₯β™ βœ”-βœ”Code
AutoDeep2020InfocomPaper♦β™₯β™ -βœ”βœ”-
Clockwork2020OSDIPaperβ™¦β™ βœ”--Code
GSLICE2020SoCCPaperβ™ βœΏβœ”βœ”--
Inferline2020SoCCPaper♦β™₯βœ”-βœ”Code
MArk2019ATCPaper♦β™₯βœ”-βœ”Code
TrIMS2019CLOUDPaperβ™¦β™ βœΏβœ”βœ”βœ”Code
Kube-Knots2019CLUSTERPaperβ™¦βœΏβœ”βœ”--
Gilman et al.2019DIDLPaper♦♠-βœ”--
Nanily2019HPCCPaperβ™¦β™ βœ”---
Ebird2019ICCDPaperβ™¦β™ βœΏβœ”βœ”-Code
Tolerance Tiers2019ISPASSPaper♣♦β™₯--βœ”-
RRL2019SCPaperβ™¦βœ”βœ”-Code
ParM2019SOSPPaperβ™¦βœ”--Code
HiveMind2018NIPSPaperβ™ βœ”βœ”--
Space-Time2018NIPSPaperβ™ βœΏβœ”βœ”--
Ease.ml2018VLDBPaper♣---Code
Rafiki2018VLDBPaperβ™£β™¦βœ”--Code
Clipper2017NSDIPaperβ™£β™¦β™ βœ”--Code

Symbols of Inference Schedulers:

AccuracyThroughputLatencyCostUtilization
♣♠♦β™₯✿

Glossary of Terms

TerminologyDefinition
JCTJob Completion Time (Job Finish Time - Job Submission Time)
Fairnessa metric to assess whether resources are fairly shared among users or jobs
QoSQuality of Service
DDLDeadline, a time point where DL job must be completed
SLOService Level Objective

Contributors

Qinghao-Hu

10 commits

gaow0007

3 commits

smile-luobin

1 commits