Actually i am very new to DevOps and client requirement is to server SDXL model to generate images i already created APIs which are required for this project in Django Rest framework. it working good but my Issue started when multiple request are getting hit, actually i am currently using GCP compute engine with L4 which is better then T4 so but my limitation is it take around 1 min and 20 sec for generating 4 images and also single request at a time because i have implemented Queue to not over exhausted GPU ram.
client requirements is to server multiple request at a time like at least 5-7 at single time and also need to overcome the generation time to 20-30 sec for generation 4 images , see I am newbie I don't know much about DevOps and GPUs so please help me out share some reference or share your idea what I can do in minimum budget. because it a initial state of project we don't have much funds. I really appreciate your help, Thanks in advance.