I’m working on a project to generate text from a 1.2B parameter full precision LLM (5gb)
Unfortunately I’m limited in the infrastructure I can use to deploy this model. There is no batch inference supported. The infrastructure I have allows me to deploy a copy of the model on a single A100, 1 per process with up to 9 processes supported (these are called “replicas”). I understand that this makes little sense given my model is memory bound, and each process will fight for memory bandwidth to read in the same weights, but I can’t change that for now.
My average input and output tokens are roughly 1000 each. I estimate the kv cache per token is roughly 400kB using full precision.
I have benchmarks of the latency of the model using various “replicas” as described above. I wanted to compare this to the theoretical performance of the A100. For my use case time to first token is negligible (<200ms), and generation is memory bound.
I find that with 5 or more replicas, the math works out and my model is roughly as fast as I expect. For example, with 1000 output tokens, 6 replicas, it’s like I’m generating using a batch of 6 requests from a 30gb model + 5gb for the kv cache. At a memory bandwidth around 1-1.3tbps that translates to ~30s per request, which is not far from what I see. The same goes for other replica numbers, 5, 7, 8 and 9.
However, when I run with a single replica, I expect generation to hover around the 5-6s mark on average. Instead, I see > 20s. I need to add 4 more replicas before the number starts to make sense. It almost seems like the model takes up too little memory to be allocated the entire memory bandwidth.
Does anyone know where this extra latency could be coming from? Do models have to reach a certain amount of used memory for A100 memory bandwidth to hit their available memory bandwidth?
Think about this: 9 women take 9 months to have 9 babies, so 1 woman should take 1 month. That is basically what you are saying.
What is basically happening is that one request takes >20seconds but your GPU utilization is far from 100%. Basically you are not using the full parallelism potential of the GPU because that is how the model works. There is very little you can do there. But when you have 6 requests at the same time, the GPU can make those requests parallel with 100% usage of its resources.