Python script for GPU assigning to tf processes -- 2
Budget: $250 – $750 USD
Hi there:
We need a python script (scripts) that is (are) capable of assign a "fraction" of the memory (one or more Nvidia GPU hardware) to a tensorflow processes.
The ultimate objective is to keep alive many (at least two) processes with access to the own amounts of GPU memory respectively. For example, in the case of a 16 GB Nvidia Tesla Graphics Card, we could have 2 processes of training with 4GB each one, 1 process of inference with 2GB, and 2GB free.
We're going to test the script(s) in windows and linux environment before the project acceptation.
Feel free to ask for more information.
We need a python script (scripts) that is (are) capable of assign a "fraction" of the memory (one or more Nvidia GPU hardware) to a tensorflow processes.
The ultimate objective is to keep alive many (at least two) processes with access to the own amounts of GPU memory respectively. For example, in the case of a 16 GB Nvidia Tesla Graphics Card, we could have 2 processes of training with 4GB each one, 1 process of inference with 2GB, and 2GB free.
We're going to test the script(s) in windows and linux environment before the project acceptation.
Feel free to ask for more information.