ONE QUESTION. TWO PLATFORMS.
How fast is
a real answer?
Choose an option. The backend checks the answer, saves your result in a database, and sends it back. We time the whole round trip.
MULTIPLE CHOICE
Loading question…
The answer key stays on the backend.
Submit → grade → save → response
Measured in this browser, including the network and a committed database write.
03 / INSIDE THE REQUEST
The code, with a stopwatch.
Loading backend code…
Submit an answer to see its measured steps.
How to read these timings
- These are the latest successful clicks for this case in this browser. AWS and Playcademy are separate requests, not a paired benchmark.
- Each duration covers the named backend timer. Database steps include Drizzle, communication with the database, and the database work.
- Parallel queries overlap. Their group duration counts that time once; do not add a group to its children.
- “Outside handler” is round trip minus handler time. It includes network, routing, middleware, response encoding and browser processing; it is not a pure network measurement.
- On Playcademy, the Worker clock advances after I/O. A zero-duration step is shown as “Unresolved”; it can still do real computation.
- These timers and their response fields already existed. Code loads before you answer, and this view renders after the stopwatch stops. Database readback is a separate request.
Your recent attempts
Your measured requests will appear here.
| Backend | Case | Result | Round trip | Backend | Database |
|---|
Published benchmark
Measured comparison results will appear here when published.
| Transport | Platform | Saved / requests | Median | p95 | p99 |
|---|
The benchmark alternates platforms, sends one request at a time, and checks every saved result afterward. Warm connections show repeated use; new connections include connection setup. New connections do not establish a cold backend. These timings reflect the benchmark client’s location.