algorithm simulations and metrics

[olderTaoist]

i use hived on product environment with 20+ nivida V100， so i want to get some reporters on reducing fragmentation of GPU and job wait time like hived paper in fifth chapter，how can i simulations. meantime, Is there a plan to add some metrics of the hived scheduler???

[olderTaoist]

@fanyangCS

[fanyangCS]

hived is designed to remove external fragmentation, it relies on Intra-VC scheduler to reduce internal fragmentation. we don't have a plan expose specific metrics, but we have APIs to query the current scheduling status, which can be used to derive metrics like fragmentation.

[olderTaoist]

hived is designed to remove external fragmentation, it relies on Intra-VC scheduler to reduce internal fragmentation. we don't have a plan expose specific metrics, but we have APIs to query the current scheduling status, which can be used to derive metrics like fragmentation.

the APIs of scheduling status is /v1/inspect/clusterstatus/physicalcluster and /v1/inspect/clusterstatus/virtualclusters? the external fragmentation is mean that virutal cluster mapping to physical cluster？the Intra-VC scheduler is somewhere like Binpack scheduling one virtual cell by one。