Live metrics
CPU, working set and private memory, I/O, threads and handles, each with a graph and an explanation.
The problem and the solution
The problem
Profiling tools show plenty of numbers but don't explain what they mean, and finding out why a program was slow took time and several tools.
The solution
It attaches to a program that is already running and shows its CPU, memory, I/O, threads and locks, with an explanation for every metric and a guide to finding the bottleneck.
01 — Overview
What it does
Argus attaches to a program that is already running, without modifying it, and shows how it uses CPU, memory, I/O, threads and locks. Every metric has a short explanation of what it measures and what an unusual value means.
Flame graph
Where the process spends CPU time, from ETW samples, with zoom and search.
Thread timeline
Thread states over time and what each thread is waiting for: locks, I/O or nothing.
Memory and disk
Hard page faults and allocations; disk operations with median and p99 latency.
Sessions
Recordings saved as .argus files, to replay later or to compare before and after a change.
Export
Data to CSV, the flame graph to SVG, the whole session to JSON.
02 — Details
Requirements and stack
Platform Windows 10/11
Download 5.1 MB
Installation Not needed
License MIT
Requirements
Windows 10 or 11, 64-bit. Argus is a single portable .exe; the zip also contains a script that installs it for the current user.
Flame graph, timeline, disk and memory use ETW and need Argus to run as administrator; the basic metrics work without.
Built with
Projects
Orbit 
Pulse 
Prism 
Cove 
Inlay