---
title: "Match the platform to the workload"
description: "The workload determines the populated configuration. Reference configurations illustrate a scale; they do not promise application performance."
url: "https://www.lmtek.com/ai-hpc-servers/use-cases"
locale: "en"
---

# Match the platform to the workload

The workload determines the populated configuration. Reference configurations illustrate a scale; they do not promise application performance.

[Configure a system](/configure)

A barebone platform: LM TEK supplies the chassis, the integrated cooling loop, and platform-level power. You specify the board, CPUs, GPUs, memory, and storage.

## Compute-heavy work, kept in the building.

### Medical imaging

Volumetric reconstruction and AI-assisted diagnostics, next to the data.

### Scientific visualisation

Simulation output rendered where the dataset lives.

### CAD & CAM

Solver and simulation runs behind the design team.

### Video post-production

Shared GPU capacity for rendering, encoding and grading.

### Computational research

Dense single-node HPC and training workloads.

## Four workloads, one platform.

[Private LLM inference: Private AI for sensitive data](/ai-solutions/private-ai-for-sensitive-data)

[Fine-tuning: AI workload patterns](/ai-solutions/foundations/ai-workload-patterns)

[Simulation and HPC: Configure a system](/configure)

[Rendering: Configure a system](/configure)

### Private LLM inference

Size local inference around the model, concurrency and data requirements. Reference scale: Modest start · Departmental AI / single-team inference. Reference scale: Mid-tier · Research group / multi-user inference / light fine-tuning. Reference scale: Production maximum · Enterprise inference / production fine-tuning.

### Fine-tuning

Discuss memory capacity, accelerator support and the intended training workflow. Reference scale: Mid-tier · Research group / multi-user inference / light fine-tuning. Reference scale: Production maximum · Enterprise inference / production fine-tuning.

### Simulation and HPC

Start with the solver, data movement and compute requirements.

### Rendering

Check renderer support and the balance between GPU capacity, CPU work and storage.

## Three reference scales.

| Specification | Value |
| --- | --- |
| Modest start | 2× RTX 6000 Ada · 1× EPYC 9354 · 256 GB DDR5 ECC |
| Mid-tier | 4× RTX PRO 6000 Blackwell · 2× EPYC 9474F · 512 GB DDR5 ECC |
| Production maximum | 8× NVIDIA H200 NVL · 2× Xeon Platinum 8580 · 1 TB DDR5 ECC |

[Read the buying guide](/ai-solutions/ai-infrastructure-buying-guide)

[Explore the flagship platform](/ai-hpc-servers/4u8gpu10kw)

The chassis, the integrated cooling loop, and platform-level power arrive engineered as one system; the board, CPUs, GPUs, memory, and storage are specified per deployment. One chassis carries a configuration from a two-GPU entry build to eight accelerators at up to 700 W per slot.

Reference configurations are examples drawn from real customer builds, not a fixed bill of materials. The final BOM is specified per deployment; a configuration inside the published validation specs is quotable as is, and LM TEK engineering is available for anything outside them.

Sizing a first deployment? The buying guide covers how to start small without throwing early spend away.

- Modest start: Departmental AI / single-team inference
- Mid-tier: Research group / multi-user inference / light fine-tuning
- Production maximum: Enterprise inference / production fine-tuning

## The workload determines the populated configuration.

[Understand the application with AI Solutions](/ai-solutions)

Describe the model, solver or renderer and the data it works on; an engineer sizes the configuration with you.