← Back to velocitry.app
Velocitry

The drive Windows called “Healthy”

Case study · London, May–August 2026

The short version: a PC froze at random for months. Every health-check tool said the hardware was fine. The Windows event log said the storage controller had stopped responding 82 times. The fix was a £100 drive. This is the fault that Velocitry was built around.

Whose PC this was

Mine. I am not going to dress up my own machine as a client job — Velocitry has just launched and I would rather show you a real diagnosis I can prove than a testimonial I cannot. Every number on this page comes from logs and SMART dumps I still have.

The symptom

Random freezes. Not a crash, not a blue screen — the whole machine would simply stop for anywhere between two seconds and half a minute, then carry on as if nothing had happened. Games stuttered. Large file copies crawled. Windows updates occasionally failed for no stated reason.

I did what everybody does. New drivers. Fresh GPU drivers. A clean Windows install. Three different “PC health check” utilities. Every one of them reported no problem.

What the health tools said

This is the SMART report from the drive, taken while the machine was freezing several times a day:

SMART attributeReadingVerdict
Critical Warning0x00No warning
Percentage Used2%Practically new
Media and Data Integrity Errors0Clean
Error Information Log Entries0Clean

By every number the drive volunteered about itself, it was a healthy, nearly-new SSD. Any tool that reads SMART — which is nearly all of them — would tell you to look elsewhere. That is exactly what they all did.

What the evidence said

SMART is the drive grading its own homework. So I stopped asking the drive for its opinion and asked Windows what had actually happened, using the storage controller reset events the operating system records whether the drive likes it or not.

MeasurementReading
Controller resets logged82, across 25 separate days
Period covered27 May – 9 August 2026
Worst single day7 resets (22 July)
Worst episode6 resets in 8 minutes (9 August)
Unsafe shutdowns recorded47
Time spent at critical temperature14 minutes
Time spent at warning temperature252 minutes

Each of those resets is the drive dropping off the bus and Windows re-establishing the connection. That is the freeze. The machine is not slow — it is waiting for a disk that has briefly ceased to exist.

The measurements that made it undeniable

The detail that ruled out the obvious explanation

The tempting conclusion is “the drive is just slow, or overloaded”. It was not. Under a deep queue the same drive sustained 350,000 IOPS at 0.1 ms — genuinely fast. A drive that can do that is not struggling with load. It was failing intermittently and recovering, which is precisely why every average-based benchmark and every health label missed it.

The fix

A replacement NVMe drive. About £100 in parts. Months of “unfixable” freezes stopped and have not returned.

No amount of optimisation would have fixed this. That matters, because it is the case where an honest diagnosis costs the person offering it money: the correct answer was “buy a part”, not “pay me to tune it”.

Why this is the whole point of Velocitry

The free health scan reads the same 30 days of error history this diagnosis came from, on your machine, and sends you a plain-English report. It is read-only, it changes nothing, and there is no obligation attached to it.

An honest limit on this page. This was one fault on one machine, and it is presented as a worked example of a method, not as a typical result. Your PC may have nothing wrong with it at all — in which case the scan will say so and you will not hear a sales pitch from me.