LLM Performance Optimizations

Wallaroo provides more than LLM deployment, but multiple methods to optimize LLM performance and efficiency. The following guides and benchmarks demonstrate different ways to improve LLM performance and improve resource use.

LLM Inference with Qualcomm QAIC

Qualcomm QIAC provides AI acceleration for Large Language Models (LLMs) at low power with high performance for x86 architectures.For access to these sample models and a demonstration on using LLMs with Wallaroo:

LLM Inference with vLLM and llama.cpp

Large Language Models (LLMs) are the go-to solution in terms of Neuro-linguistic programming (NLP), promoting the the need for efficient and scalable deployment solutions. Llama.cpp and Virtual LLM (vLLM) are two versatile tools for optimizing LLM deployments with innovative solutions to different pitfalls of LLMs.

  • Llama.cpp is known for its portability and efficiency designed to run optimally on CPUs and GPUs without requiring specialized hardware.
  • vLLM shines with its emphasis on user-friendliness, rapid inference speeds, and high throughput.
For access to these sample models and a demonstration on using LLMs with Wallaroo:
Continuous Batching for LLMs

Continuous batching provides a method for increased performance for serving LLMs in realtime GenAI applications (e.g AI agents) for scaled usage. Wallaroo leverages vLLM as a runtime to maximize LLM performance on GPUs for such applications.
For additional information and a demonstration on using LLMs with Wallaroo:

Dynamic Batching for LLMs

Dynamic batching improves inference result performance at scale in high traffic scenarios.
For access to these sample models and a demonstration on using LLMs with Wallaroo:

Autoscaling for LLMs

Autoscale triggers reduces latency for LLM inference requests by adding additional resources and scaling them down based on scale up and scale down settings.
For access to these sample models and a demonstration on using LLMs with Wallaroo:

Performance Benchmarks