Show HN: Utilyze – an open source GPU monitoring tool more accurate than nvtop
Sentiment Mix
Geography
Expert Signals
ManyaGhobadi
author • 1 mention
Hacker News
source • 1 mention
AI-Generated Claims
Generated from linked receipts; click sources for full context.
The standard GPU utilization metric reported by nvidia-smi, nvtop, Weights & Biases, Amazon CloudWatch, Google Cloud Monitoring, and Azure Monitor is highly misleading.
Supported by 1 story
It reports the fraction of time that any kernel is running on the GPU, which means a GPU can report 100% utilization even if only a small portion of its compute capacity is actually being used.
Supported by 1 story
In practice, we've seen workloads with ~1–10% real compute throughput while dashboards show 100%.This becomes a problem when teams rely on that metric for capacity planning or optimization decisions, it can make underutilized systems look saturated.We're releasing an open-source (Apache 2.0) tool, Utilyze, to measure GPU utilization differently.
Supported by 1 story
It samples hardware performance counters and reports compute and memory throughput relative to the hardware's theoretical limits.
Supported by 1 story
Related Events
Super ZSNES – GPU Powered SNES Emulator
Hardware • 4/28/2026
Show HN: OSS Agent I built topped the TerminalBench on Gemini-3-flash-preview
Uncategorized • 4/27/2026
Meta's Muse Spark AI model impressed. Here's the next test in race with Google, OpenAI. - msn.com
LLMs • 4/28/2026
The Nuances of Cursor’s Gross Margin; Comparing GPT-5.5 and Claude Mythos - The Information
LLMs • 4/27/2026
Show HN: AgentSwift – Open-source iOS builder agent
Open Source • 4/28/2026
Causality Chain
Preceded By