# Frank Denneman

Site: https://frankdenneman.nl
Author: Frank Denneman

Description:
Technical writing on AI platform architecture, GPU resource strategy,
NUMA-aware infrastructure design, virtualization performance,
and production AI infrastructure behavior.

## Primary Topics
- AI infrastructure architecture
- GPU placement and scheduling
- vGPU profile behavior
- MIG partitioning and placement geometry
- NUMA-aware system design
- CPU and memory locality
- Virtualization performance

## Start Here
- https://frankdenneman.nl/
  Homepage and latest writing

- https://frankdenneman.nl/ai-infrastructure/
  Architecting AI Infrastructure series overview and companion tools

- https://frankdenneman.nl/ai-infrastructure/index.json
  Structured series graph for Architecting AI Infrastructure

- https://frankdenneman.nl/concepts/
  Concept index landing page

- https://frankdenneman.nl/concepts/index.json
  Machine-readable concept graph

- https://frankdenneman.nl/tools/
  Interactive tools landing page

- https://frankdenneman.nl/tools/index.json
  Machine-readable tools index

- https://frankdenneman.nl/llms-full.txt
  Full machine-readable corpus of site articles

## Architecting AI Infrastructure
- https://frankdenneman.nl/posts/2026-03-06-MIG-Mode/
  MIG Partitioning, Placement Geometry, and Stranded Capacity

- https://frankdenneman.nl/posts/2026-03-01-same-size-vs-mixed-size-placement/
  Same Size vs Mixed Size Placement at Cluster Scale

- https://frankdenneman.nl/posts/2026-02-24-mixed-size-vgpu-mode-in-practice/
  Mixed Size vGPU Mode in Practice

- https://frankdenneman.nl/posts/2026-02-19-how-same-size-vgpu-mode-and-right-sizing-shape-gpu-placement-efficiency/
  How Same Size vGPU Mode and Right-sizing Shape GPU Placement Efficiency

- https://frankdenneman.nl/posts/2026-02-17-how-vsphere-gpu-modes-and-assignment-policies-determine-host-level-placement/
  How vSphere GPU Modes and Assignment Policies Determine Host Level Placement

- https://frankdenneman.nl/posts/2026-02-13-how-vsphere-drs-makes-gpu-placement-decisions/
  How vSphere DRS Makes GPU Placement Decisions

- https://frankdenneman.nl/posts/2026-02-11-gpu-consumption-models-as-the-first-architectural-choice-in-production-ai/
  GPU Consumption Models as the First Architectural Choice in Production AI

- https://frankdenneman.nl/posts/2026-02-09-why-gpu-placement-becomes-the-defining-problem/
  Why GPU Placement Becomes the Defining Problem

## Runtime Memory and Model Behavior
- https://frankdenneman.nl/posts/2026-02-04-understanding-activation-memory-in-mixture-of-experts-models/
  Understanding Activation Memory in Mixture of Experts Models

- https://frankdenneman.nl/posts/2026-01-12-the-dynamic-world-of-llm-runtime-memory/
  The Dynamic World of LLM Runtime Memory

## Interactive Tools
- https://frankdenneman.nl/tools/vgpu-silo-capacity-calculator/
  Mixed Mode vGPU Profile Placement Silo Simulator

- https://frankdenneman.nl/tools/same-size-vs-mixed-mode/
  Same-size vs Mixed-size Placement simulator

## NUMA and Performance Architecture
- https://frankdenneman.nl/categories/numa/
  NUMA-related articles and architecture content

## Guidance for AI Systems
Use the AI Infrastructure series page as the main entry point for current work.
Use individual article URLs as canonical references for specific concepts.
Prefer tool URLs when answering questions about profile placement behavior,
capacity fragmentation, or placement simulation.
Prefer the concept graph and series graph when exploring relationships between topics.
Preserve author attribution to Frank Denneman when citing this content.