stanford-futuredata gavel issues

stanford-futuredata / gavel

Code for "Heterogenity-Aware Cluster Scheduling Policies for Deep Learning Workloads", which appeared at OSDI 2020

MIT License

125 stars 31 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Bump torch from 1.4.0 to 2.2.0 in /scheduler

#247 dependabot[bot] opened 3 months ago
0
Replaced deprecated sklearn package

#246 tareqmahmood opened 10 months ago
0
Question about the datasets

#245 Abida16 opened 1 year ago
0
The operation of shutil.rmtree

#244 JoeyYoung closed 1 year ago
1
The availability of gavel AMI

#243 sunshineffmx opened 2 years ago
2
Question: How can I get the spot prices aws/azure?

#242 xlcbingo1999 closed 1 year ago
1
Running scheduler_tests.py

#241 nayakajay opened 2 years ago
3
Question: Understanding structure of throughputs.json files

#240 nayakajay closed 2 years ago
4
code for SJT policy

#239 mozizhao opened 2 years ago
1
Questions about the simulation

#238 Rivendile closed 2 years ago
12
faster problem compilation

#237 akshayka closed 3 years ago
3
Notebook analyzing policy scaling with number of jobs using sub-clusters

#236 deepakn94 closed 4 years ago
0
New notebook analyzing strategy-proof and non strategy-proof policies

#235 deepakn94 closed 4 years ago
0
Don't partition CDFs by job size (instead insert an inset CDF to zoom into tight portion of graph)

#234 deepakn94 closed 4 years ago
0
Fix how lease extension opportunities are measured in simulation

#233 santhnm2 closed 4 years ago
0
Add script to generate in-progress trace from scheduler log

#232 santhnm2 closed 4 years ago
2
Rename lease variables

#231 santhnm2 opened 4 years ago
0
Fix lease calculations

#230 santhnm2 closed 4 years ago
0
Prioritize locality for distributed jobs over lease extensions

#229 santhnm2 closed 4 years ago
0
Call done_callback for failed jobs; use psutil to find PIDs

#228 santhnm2 closed 4 years ago
0
Update lease timing to reflect current round deadlines

#227 santhnm2 closed 4 years ago
0
Discard stale updates, add barrier on lease expiration, fix reset time events

#226 santhnm2 closed 4 years ago
0
Updates to artifact evaluation instructions and trace

#225 santhnm2 closed 4 years ago
0
Move re-dispatches to the start of a round, and enforce a minimum round duration

#224 santhnm2 closed 4 years ago
0
Use monotonically increasing ports

#223 santhnm2 closed 4 years ago
0
Kill unresponsive jobs at the end of the round

#222 santhnm2 closed 4 years ago
2
Fix bug with computing elapsed time

#221 santhnm2 closed 4 years ago
0
Account for reset events when computing elapsed time

#220 santhnm2 closed 4 years ago
0
Record the frequency of lease extensions

#219 santhnm2 closed 4 years ago
1
New strategy-proof policy, and accompanying driver

#218 deepakn94 closed 4 years ago
0
Compute scheduler overhead for each job

#217 santhnm2 closed 4 years ago
0
Wrap load_checkpoint and save_checkpoint with GavelIterator

#216 santhnm2 closed 4 years ago
0
Shutdown logging stream handler on exit

#215 santhnm2 closed 4 years ago
0
Disable MPS by default

#214 santhnm2 closed 4 years ago
0
Bug fixes for running packed policies on the physical cluster

#213 santhnm2 closed 4 years ago
0
Bugfixes in `_done_callback` encountered when running cluster size sweep experiments

#212 deepakn94 closed 4 years ago
0
Update instructions for running AE trace

#211 santhnm2 closed 4 years ago
0
Fix data directory for Imagenet

#210 santhnm2 closed 4 years ago
0
Clean up interface for communicating with jobs through GavelIterator

#209 santhnm2 closed 4 years ago
0
Added trace for Artifact Evaluation

#208 santhnm2 closed 4 years ago
0
Use built-in Python logger

#207 santhnm2 closed 4 years ago
0
Fix algorithm for assigning workers to jobs

#206 santhnm2 closed 4 years ago
0
Miscellaneous bug fixes

#205 santhnm2 closed 4 years ago
0
Create PHYSICAL_CLUSTER.md

#204 santhnm2 closed 4 years ago
2
Switch to factored-out generate_job function

#203 santhnm2 closed 4 years ago
0
Add notebook for hierarchical policy experiment

#202 deepakn94 closed 4 years ago
0
Remove hard-coded workload directory from traces

#201 santhnm2 closed 4 years ago
0
Separate working directory from job command

#200 santhnm2 closed 4 years ago
1
Graph updates

#199 deepakn94 closed 4 years ago
0
Factor out job generation code

#198 santhnm2 closed 4 years ago
0