Commit b7a6f903 authored by rashadul.kabir's avatar rashadul.kabir
Browse files

Commit for merging with main branch.

parents af78b6d4 45cfbdce

README.md

+9 −2
Original line number Diff line number Diff line
@@ -32,11 +32,18 @@ Note: Requires python3.9 or greater. 
    DPATH=~/data/frontier-sample-2024-01-18

    python main.py -f $DPATH/slurm/joblive/$DATEDIR $DPATH/jobprofile/$DATEDIR



## Open Telemetry dataset



Download `job_table.parquet` from https://zenodo.org/records/10127767



    # Marconi100

    python main.py --system marconi100 -f ~/data/job_table.parquet 



    Note, once the data has been processed, it will be saved as an NPZ file, which

    can be more quickly started in subsequent simulations

## Snapshot of extracted workload data



To reduce the expense of extracting the needed data from the telemetry parquet files,

RAPS saves a snapshot of the extracted data in NPZ format. The NPZ file can be

given instead of the parquet files for more quickly running subsequent simulations, e.g.:



    python main.py -f jobs_2024-02-20_12-20-39.npz

config/frontier/power.json

+2 −2
Original line number Diff line number Diff line
@@ -4,10 +4,10 @@ 
    "POWER_CPU_IDLE": 90,

    "POWER_CPU_MAX": 280,

    "POWER_MEM": 74.26,

    "POWER_NIC": 20,

    "POWER_NVME": 30,

    "POWER_NICS": 80,

    "POWER_CDUS": 211836.75,

    "POWER_SWITCH": 250,

    "POWER_CDU": 8473.47,

    "POWER_UPDATE_FREQ": 15,

    "RECTIFIER_PEAK_THRESHOLD": 13670,

    "SIVOC_LOSS_CONSTANT": 13,

config/frontier/system.json

+1 −0
Original line number Diff line number Diff line
@@ -6,6 +6,7 @@ 
    "CHASSIS_PER_RACK": 8,

    "NODES_PER_BLADE": 2,

    "SWITCHES_PER_CHASSIS": 4,

    "NICS_PER_NODE": 4,

    "RECTIFIERS_PER_CHASSIS": 4,

    "NODES_PER_RECTIFIER": 4,

    "MISSING_RACKS": [41],

config/frontier/uq.json

+1 −1
Original line number Diff line number Diff line
@@ -2,8 +2,8 @@ 
    "POWER_GPU_UNCERTAINTY": 0.05 ,

    "POWER_CPU_UNCERTAINTY": 0.05 ,

    "POWER_MEM_UNCERTAINTY": 0.05 ,

    "POWER_NIC_UNCERTAINTY": 0.05 ,

    "POWER_NVME_UNCERTAINTY": 0.05 ,

    "POWER_NICS_UNCERTAINTY": 0.05 ,

    "POWER_CDUS_UNCERTAINTY": 0.05 ,

    "POWER_NODE_UNCERTAINTY": 0.002,

    "POWER_SWITCH_UNCERTAINTY": 0.05 ,

config/marconi100/power.json

+2 −2
Original line number Diff line number Diff line
@@ -4,10 +4,10 @@ 
    "POWER_CPU_IDLE": 47.25,

    "POWER_CPU_MAX": 252,

    "POWER_MEM": 74.26,

    "POWER_NIC": 21,

    "POWER_NVME": 45,

    "POWER_NICS": 21,

    "POWER_CDUS": 0,

    "POWER_SWITCH": 250,

    "POWER_CDU": 0,

    "POWER_UPDATE_FREQ": 20,

    "RECTIFIER_PEAK_THRESHOLD": 13670,

    "SIVOC_LOSS_CONSTANT": 0,