Hacker News 熱門53

Show HN: PantheonGPU – GPU health testing and AI workload benchmarking

2026/08/19 02:47· 56天前

AI 摘要

Hi HN, I built PantheonGPU because I wanted a better way to answer a simple question: is this GPU actually healthy and performing the way it should? A GPU can show normal temperatures and utilization and still be underperforming, unstable under certain workloads, or have memory, PCIe, or configuration issues. PantheonGPU actively tests the GPU instead of only monitoring telemetry. It currently includes 45+ tests covering compute, tensor workloads, memory, cache, PCIe, thermals, stability, and AI/LLM inference. It supports both NVIDIA CUDA and AMD ROCm. I’m also exploring a larger use case: running Pantheon across GPU fleets to identify individual GPUs that behave differently from the rest of a server or cluster. I’d especially appreciate feedback from people running AI infrastructure, multi-GPU systems, local LLMs, or GPU clouds. Comments URL: https://news.ycombinator.com/item?id=49350637 Points: 5 # Comments: 0

正文

GPU stress testing and diagnostics

Pantheon tests GPU compute, memory, cache, interconnect, and power behavior. Run focused workloads, capture telemetry, and keep the results for comparison.

Download Pantheon View benchmarks Get help or contribute

45 workloadsfocused stress tests

CUDA + ROCmNVIDIA and AMD support

Local reportsexportable telemetry

Quick Start

The Debian package is the simplest installation path for Ubuntu and Debian systems.

1. Install prerequisites

Install the basic build tools:

sudo apt-get update
sudo apt-get install -y make g++

Then install the compiler for your GPU platform. You only need one:

NVIDIA CUDAAMD ROCm/HIP

sudo apt-get install -y nvidia-cuda-toolkit

sudo apt-get install -y hipcc

2. Install Pantheon

Download and install the latest Debian package:

VERSION=1.0.14
wget &quot;https://github.com/saqibkh/pantheongpu_website/releases/download/v${VERSION}/pantheongpu_${VERSION}_amd64.deb&quot;
sudo apt install &quot;./pantheongpu_${VERSION}_amd64.deb&quot;

To uninstall the Debian package later:

sudo apt-get remove pantheongpu

3. Verify the installation

Run a short hardware inventory test:

pantheon --test baseline_metrics --duration 10

Then run a targeted stress test on GPU 0:

pantheon --test fp64_virus --duration 30 --gpu 0

Note

Pantheon automatically detects CUDA, ROCm/HIP, or mock mode. Run the pantheon command directly; you do not need to pass --platform cuda.

Completely remove Pantheon

The native package command above removes Pantheon's package-managed files. To also remove runtime-created files and the current user's compiled workload cache, or to remove a portable installation on RHEL, Fedora, Rocky Linux, AlmaLinux, or another Linux distribution, run:

curl -fsSL https://pantheongpu.com/uninstall.sh | sudo sh

This leaves CUDA, ROCm, system compilers, and benchmark reports stored outside Pantheon's installation and cache directories untouched.

Alternative: install from the release bundle

The release bundle contains the Debian package and an install.sh helper for RHEL-family and other Linux distributions.

VERSION=1.0.14
wget &quot;https://github.com/saqibkh/pantheongpu_website/releases/download/v${VERSION}/pantheongpu_${VERSION}_amd64.tar.gz&quot;
tar -xzf &quot;pantheongpu_${VERSION}_amd64.tar.gz&quot;
cd &quot;pantheongpu_${VERSION}_amd64&quot;
sudo apt install &quot;./packages/pantheongpu_${VERSION}_amd64.deb&quot;

Uninstall a Debian package installation with:

sudo apt-get remove pantheongpu

On RHEL-family and other Linux systems, install the portable bundle with sudo ./install.sh. Remove that installation with:

sudo rm -f /usr/local/bin/pantheon &amp;&amp; sudo rm -rf /opt/pantheongpu

Use the complete-removal command above if you also want to clear the current user's compiled workload cache.

Build cache

First-run workload builds are cached under ${XDG_CACHE_HOME:-$HOME/.cache}/pantheongpu/builds/. Set PANTHEON_BUILD_CACHE_DIR to choose another writable cache directory.

查看原文（Hacker News 熱門）