-
Notifications
You must be signed in to change notification settings - Fork 14
Description
I would like to request improved cluster info guidance as provided in the file cluster_info.yaml
The readme also refers to:
https://github.com/GreenAlgorithms/green-algorithms-tool/tree/master/data
which does not exist, imagining this is the correct link:
https://github.com/GreenAlgorithms/GA-data
The main question is if TDP values, in whether these are taken from specs as the source code seems to elude to, or need to be supplied as documented.
TDP values are described as 'per core', are equal to large TDP values, or if the TDP values entered in the yaml commented by "TDP of the processor, in W, per core" need to be divided by the number of cores. There is a discrepancy between CPU and GPU sections, with [TDP] being per core for CPU, but not for GPU:
CPU:
type: # [CPU or GPU]
model: "" # [str] the model of the processing core on this partition. Not actually used by the code but useful for reference for others.
TDP: <10> # [number] TDP of the processor, in W, per core
GPU:
type: # [CPU or GPU]
model: "" # [str] the model of the processing core on this partition. Not actually used by the code but useful for reference for others.
TDP: <250> # [number] For GPUs, the TDP is for the entire GPU
# For GPU partitions, we also need info about the CPUs available for support.
model_CPU: "" # [str] Not actually used by the code but useful for reference for others.
TDP_CPU: <10> # [number] TDP of the processor, in W, per core
Minor: the links in the readme appear to be old, e.g.: https://github.com/Llannelongue/GreenAlgorithms4HPC.git