Fine-grained Profiling of Online AI Workloads on Kubernetes Using ACK AI Profiling

This article demonstrates how to use ACK AI Profiling, built on eBPF and dynamic process injection, to perform non-intrusive, low‑overhead profiling of Kubernetes‑deployed large‑language‑model inference services, identify GPU memory growth causes, and apply optimization recommendations to prevent OOM issues.

AI profilingGPU MemoryKubernetes

0 likes · 10 min read

Fine-grained Profiling of Online AI Workloads on Kubernetes Using ACK AI Profiling

Linux Kernel Journey

Nov 5, 2024 · Artificial Intelligence

Understanding AI Flame Graphs: Insights from Brendan Gregg

The article introduces Intel's AI Flame Graph, a low‑overhead profiling tool that visualizes AI accelerator and GPU workloads across the full software stack, explains its design, demonstrates SYCL matrix‑multiply benchmarks, discusses challenges of AI instruction analysis, and outlines future adoption and impact.

AI profilingGPUIntel

0 likes · 16 min read

Understanding AI Flame Graphs: Insights from Brendan Gregg

Linux Code Review Hub

Nov 2, 2024 · Artificial Intelligence

Inside Intel’s AI Flame Graph: Low‑Overhead Profiling for Faster, Greener AI

The article introduces Intel’s AI Flame Graph, a low‑overhead profiling tool that visualizes AI accelerator and GPU execution alongside the full software stack, explains its design, shows SYCL matrix‑multiply examples, discusses challenges of AI workload analysis, and outlines future adoption and impact on performance and energy savings.

AI profilingGPUIntel

0 likes · 16 min read

Inside Intel’s AI Flame Graph: Low‑Overhead Profiling for Faster, Greener AI