Who can analyze a crash dump, and what does it cost?
Four ways to get a crash dump analyzed: yourself, free tools, an AI analyser or a consultant. What each needs, what it costs you, and where DumpTriage fits.
Short answer: Anyone with a debugger, symbols and the dump can analyze one, and the tools for that are free. What a crash dump costs you is time and skill, so the choice is between your hours, a consultant's hours and a service that runs the debugger for you. Whichever you pick, the dump goes to whoever reads it, so settle what it may contain first.
Four ways to get a dump analyzed
| Route | What you need | What it costs you | Fits when |
|---|---|---|---|
| Yourself, in WinDbg or gdb | The debugger, symbols, the dump and the skill to read the output. | Your time. The first dump takes longest. | A common stop code or one application crash, and the dump can stay on your machine. |
| A free viewer | The minidump file. Tools such as NirSoft's BlueScreenView list the drivers a minidump points at. | Depth: a list of suspects is not a root cause, and a viewer does not run a debugger analysis. | A first look at a Windows minidump to see which driver is named. |
| An AI analyser | An account, or the tool's API, and the dump or the debugger's output. | The dump, or text from it, goes to the tool's servers and often to a model provider. | You have many dumps, or nobody on the team reads a debugger. |
| A consultant or colleague | Someone who reads dumps, and a way to get the file to them. | Their hours, and the dump goes to another person. | A hard case, such as memory corruption, where you want a person to take responsibility for the answer. |
The choice between the first route and the rest is covered in analyze a crash dump yourself or hand it off. If you start on your own, how to analyze a Windows minidump and Linux core dumps with gdb and crash are free to follow.
Questions to ask any service or person before you send a dump
- What does the dump hold? A kernel or full memory dump can contain far more than a minidump. What is inside a dump lists each type.
- Does a model see the dump itself, or only the debugger's text output, and is that text redacted first?
- Where is the dump stored, for how long, and who can read it?
- Is it used to train or fine-tune a model?
- Who is the subprocessor behind the service? A service should list them.
Our answers are on the AI and data page and the subprocessors list, and is it safe to send a customer's crash dump walks through the same questions for a dump that belongs to someone else.
When the size decides it
A Windows full or kernel memory dump can be as large as the machine's RAM, which rules out routes that need you to copy it by email. Crash dump too big, or no WinDbg on this machine covers smaller dump types and opening a dump on another machine.
Where DumpTriage fits
DumpTriage is the third route: you upload a Windows crash dump or a Linux core dump (or send it with the API, or let the agents from Downloads send new dumps). It runs the debugger on our own worker, reads the output and sends back the faulting module, the stack and a plain-language root cause; anything the automated passes can't settle goes to an engineer. How the analysis works has the three tiers.
What happens to the dump, in the words of our AI and data page:
- No AI model sees your dump. The debugger runs on our own worker. In the automated analysis, a model only reads the debugger's text output, after we redact it.
- We don't train or fine-tune any model on your dumps, transcripts or reports, and our AI provider's terms say it won't either.
Redaction is pattern-based and best-effort, not a guarantee: a secret in a shape we don't recognise can get through. For the most sensitive systems, keep them out of scope or send minidumps rather than full memory dumps.
Common questions
- Who can analyze a crash dump?
- You, with a debugger and symbols; a colleague or consultant who reads dumps; or an analysis service that runs the debugger for you. Which fits depends on how many dumps you have, how large they are, and whether the dump may leave your network.
- Is there a free way to analyze a crash dump?
- Yes. WinDbg and the Microsoft symbol server are free, gdb and crash are free on Linux, and a viewer such as NirSoft's BlueScreenView lists the driver a minidump points at. They cost your time and skill, and a viewer does not run a full debugger analysis.
- Can an AI tool analyze a BSOD dump file?
- Some can. Tools differ in what they send where: some paste the debugger's output into a chat model, some upload the dump. Ask what leaves your network, who reads it, how long it is kept and whether it trains models before you send a customer's dump.
Rather not read the dump yourself?
Upload it to DumpTriage. It runs the debugger, reads the output and sends back the faulting module, the stack and a plain-language root cause (how the analysis works). Cases the automated passes can't settle go to an engineer. Free during the pilot.
Create an account