-
Feature
-
Resolution: Done
-
Critical
-
None
-
Product / Portfolio Work
-
-
0% To Do, 0% In Progress, 100% Done
-
False
-
-
False
-
None
-
None
-
-
None
-
-
None
-
None
-
None
-
None
Feature Overview (aka. Goal Summary)Â Â
As an OpenShift administrator looking to run AI workloads on the platform, efficient GPU utilization is crucial given the high cost of these resources. While NVIDIA GPUs offer a method to pre-slice the GPU for multiple workloads, this approach can lead to resource wastage if the slicing does not align with the actual workload demands.
Therefore, I want to dynamically slice the GPU based on the specific requirements of each workload, ensuring optimal utilization and minimizing resource waste.
Â
GoalÂ
TP in 4.19
Acceptance criteriaÂ
- operator can run on infra/worker node
- Operator should not modify Machine config
- is build and tested via Konflux
- Â
Â
GA goalÂ
- FIPS complient -> depends on nvidiaÂ
- should work in disconnected mode
- can be installed in non *openshift NS
Non-Goals:
- Share MIG slices among multiple containers
- Achieve scheduling latency below 5s (need help from RH team)
- SNO/MicroShift testing is out of scopeÂ
- clones
-
OCPSTRAT-1591 Dev P: Dynamic Accelerator slicer Operator (fka: InstaSlice)
-
- Closed
-
- is related to
-
RFE-3180 MIG Support for PODS
-
- Backlog
-
- links to
(29 links to)