---
title: "LM TEK 4U8GPU10KW validation"
description: "Platform ratings and measured evidence answer different questions. Ratings define the envelope; measurements describe a run with stated conditions and source."
url: "https://www.lmtek.com/ai-hpc-servers/4u8gpu10kw/validation"
locale: "en"
---

# LM TEK 4U8GPU10KW validation

Platform ratings and measured evidence answer different questions. Ratings define the envelope; measurements describe a run with stated conditions and source.

## The rated envelope.

| Specification | Value |
| --- | --- |
| GPU power envelope | Up to 700 W per slot sustained |
| Power envelope | 10 kW installed (5 × 2 kW); 8 kW sustained, N+1 |
| Max chassis power | Up to 8000 W sustained |
| Operating temperature range | 5–30°C ambient · 5–60°C coolant |
| Working pressure | 2500 Pa |
| Max pressure | 30 000 Pa (with EK-Pro PRM) · 75 000 Pa otherwise |

Rated. The 700 W per-slot sustained envelope and the 8000 W sustained chassis figure are platform maxima; achievable sustained power depends on the chosen components and ambient conditions.

Source: LM TEK 4U8GPU10KW published specifications.

## Two-hour torture test

| Specification | Value |
| --- | --- |
| Condition: Duration | 2 h |
| Condition: Ambient | 29°C measured; air-conditioning set point 30°C |
| Condition: Load programme | Maximum-power torture test |
| Condition: GPU slots populated | 8 |
| GPU package | 76 to 81°C |
| CPU | 68°C |
| DRAM | 36 to 39°C |
| Coolant, radiator outlet / inlet | 44 / 58.4°C |
| System flow | 273 l/h |
| Dual D5 pumps | 4500 RPM at 100% PWM |

Measured. Source: LM TEK test report 631, MAX POWER run.

These results describe that run, not validation of a 700 W-per-slot configuration.

## 80-hour continuous run

| Specification | Value |
| --- | --- |
| Condition: Duration | 80 h continuous |
| Condition: Ambient | 21.4 to 23.2°C (22°C nominal) |
| Condition: Load programme | Synthetic full load on the CPU and all eight GPU slots |
| Condition: Pumps and fans | 100% PWM |
| Condition: GPU slots populated | 8 |
| GPU temperature, all eight slots | 65 to 73°C |
| Coolest slot / warmest slot | 66 to 67°C / 68 to 73°C |
| CPU | 58 to 59°C |
| Coolant into the radiator | 47.9 to 49.4°C |
| Coolant out of the radiator | 34.7 to 36.5°C |
| System flow | 269 to 273 l/h |

Measured. Source: LM TEK test report 631, long-term run.

Ranges are the minimum and maximum of the 80 hourly readings.

## Power delivery in the same run

| Specification | Value |
| --- | --- |
| Power distribution board, maximum | 37.7°C |
| Power supplies, maximum | 35°C |
| GPU power connectors, maximum | 40.4°C |

Source: LM TEK test report 631, long-term run.

GPU power connectors under load: 40.4°C at the connectors. The hotter spot in frame is a water block, not a connector.

Power distribution board under load: 37.7°C maximum.

## Sustained draw, from a separate test

| Specification | Value |
| --- | --- |
| Sustained draw | 6800 to 7000 W |

[Plan for the heat](/ai-hpc-servers/deployment#plan-for-the-heat)

Measured, separate test. Source: LM TEK long-term endurance testing.

LM TEK long-term endurance testing recorded 6800 to 7000 W sustained draw. This is a separate test from both Report 631 runs above. It has no written report, so its ambient, load programme, duration and populated slots are not published. Further testing is being documented.

## Further evidence, on request.

Acoustic measurements under sustained load are available on request.

Third-party independent measurement reports are available on request under NDA.

## The measured conditions do not promise the same result for every configuration.

[Deployment planning](/ai-hpc-servers/deployment)

The platform ships with published validation specs: supported board formats and sockets, the per-slot GPU envelope, block availability, and the operating limits. A configuration inside them is quotable as is. For a board, GPU, or cooling requirement outside them, send LM TEK engineering the intended configuration.