What does a DeadTier score actually mean?
A DeadTier score is 0–100 against one fixed yardstick, and a CPU score is a raw count. Here is what each number is, why a phone is expected to sit low, and what the tier can and cannot tell you.
The short answer to “is 1,795 good?” is that it depends entirely on what kind of device produced it — and that the number people usually quote is not the score at all.
That particular figure is real: it is the CPU score of a budget Android phone, the first physical device this test was ever run on. Whether it is good depends on what you were expecting it to be.
There are two different numbers
The CPU score is a raw count. It comes from a maths loop running on every core your browser will give us, and it is roughly proportional to how much arithmetic your device got through. A bigger number is more work done. It has no ceiling and no units, so on its own it means very little — 1,795 is neither good nor bad until you know whether it came from a phone or a desktop tower.
The power rating is the 0–100 figure in the dial. That one is on a fixed scale: 100 means matching a strong current desktop. Every device type is measured against that same yardstick, which is the whole point — it means a phone and a laptop can be compared directly.
A phone is supposed to score low
Because the scale tops out at desktop hardware, a phone sitting at 30 or 40 is not a broken result. A phone runs on a battery, in a case, with no fan, at a fraction of the power budget. Expecting it to reach 100 would be like expecting a hatchback to match a lorry for towing.
This is why the test asks what you are testing before it will start. The same score reads as a completely different verdict depending on the answer, and guessing from the browser gets it wrong often enough to matter.
What the tier adds, and what it costs
The tier — Flagship, High-End, Mid-Range, Entry-Level, Budget — answers the second question: is this a good phone? It is the one judgement on the results screen, and it is the weakest thing there.
It is anchored on real hardware: the budget Android phone above measured 1,795 on the CPU loop and 54.6 average FPS, and the scale is built outward from that measurement. But one device can only anchor a scale, not calibrate it, so a single assumption remains about how much faster a flagship phone is. The results screen says which basis is in force, and once enough real runs of a device type have been collected the tier switches to where your device actually falls among them — at which point the assumption disappears.
Everything else on that screen is a direct measurement. The tier is the only figure carrying an estimate, and it is labelled.
The numbers worth reading instead
- 1% low FPS — the mean of the slowest 1% of frames. This is the number that matches how a device feels. An average can look healthy while the device stutters badly; the 1% low cannot.
- Frame consistency — the 1% low against the run’s own average. High means frames arrive evenly. Note that a uniformly slow device also scores high here, which is exactly why this figure is displayed but deliberately kept out of the score.
- Worst frame and hitches — the single longest frame, and how many frames took more than twice the run’s median. One 400 ms stall is invisible in an average and obvious here.
- Pass spread — each subtest runs three times, and this says how far the three differed. A large spread means the number is less repeatable than it looks, which is worth knowing before you trust it.
The most reliable comparison needs no scale at all
If you want to know whether one device is faster than another, test both and put the results side by side. That comparison is measurement against measurement — it does not depend on any reference point, any tier, or any assumption. Run the test on the device in your hand, copy the link it gives you, open that link on the other device, and run it there.
Measure the device you are holding
About a minute, nothing installed, and it works on a device you do not own.
Run the test