Y2KDASH vs Waveform: An Honest Comparison
After DSLReports Speedtest went offline in March 2025, two free browser-based tools remain for measuring bufferbloat: Y2KDASH and Waveform. Both are legitimate, both use the same measurement principle, and both produce comparable numbers. The difference is how long they run, what they show, and what you do with the result.
Summary table
| dimension | Y2KDASH | Waveform |
|---|---|---|
| test duration | continuous, 60s sampling cadence | single-shot, ~30 seconds total |
| test backend | speed.cloudflare.com (Anycast) | Waveform's own servers |
| primary output | rolling time-series charts + AIM grade (A–F) | letter grade A+ to F + before/after latency chart |
| packet loss measurement | yes, 20 parallel probes per burst | no (loaded/unloaded latency only) |
| jitter measurement | yes, per-sample stdev | inferred from latency variance |
| throughput measurement | yes, continuous | yes, within the test window |
| shareable result | live URL, updates | shareable result page snapshot |
| price | free, no sign-up, no ads | free, no sign-up |
| data logged server-side | none (all client-side via Cloudflare) | anonymized aggregates per Waveform policy |
| best for | ongoing monitoring, before/after router tuning, tail-latency distribution | quick one-time grade, screenshot for ISP complaint, short sessions |
Methodology: both test the same thing
Both tools measure loaded latency: round-trip time during saturated download and upload. Both do it the same way — they start a large download, fire parallel ping probes during the transfer, and measure the difference between the unloaded baseline and the loaded measurement. The difference is the bufferbloat delta. See the loaded-latency and bufferbloat glossary entries for the definitions.
Waveform runs the cycle once. Y2KDASH runs the cycle every 60 seconds while the tab is open. Neither approach is more "correct" — they answer different questions.
Output format: single grade vs. distribution
Waveform returns a letter grade (A+ to F) based on the peak loaded-latency delta observed during its test window. The grade is easy to read and easy to share. Connection quality rarely behaves as a single point — it varies across minutes and hours — and a single shot captures whatever moment the test happened to sample.
Y2KDASH returns a time-series distribution. You see every sample on a rolling chart. The AIM composite grade aggregates the distribution into a letter, the same way Waveform does, but the underlying data shows you the tail: the 99th-percentile latency under load, the hour-of-day variation, the correlation between throughput and jitter. For diagnosing intermittent issues, the distribution is more useful than the grade.
Accuracy: comparable within measurement noise
Running both tools on the same connection produces loaded-latency numbers within 5–10 ms of each other. The difference is noise, not methodology. Either tool can be relied upon for an honest bufferbloat verdict.
One edge case: Waveform's shorter test window is more susceptible to transient conditions. A 30-second test that happens during a quiet moment on a shared cable segment may report better numbers than reality. A 10-minute Y2KDASH run captures both the good moments and the bad.
When to use Waveform
- Screenshot for an ISP support ticket. Waveform's grade format is recognizable and immediately readable by a non-technical support agent.
- Quick sanity check. 30 seconds and you have a number. Useful when you want to verify a hypothesis (e.g., "did my cable line just degrade?") without leaving a tab open.
- One-time baseline before a test. Run Waveform to confirm your network is in a known state, then do whatever test you actually care about.
When to use Y2KDASH
- Tuning a router. Changes to CAKE / SQM / FQ-Codel show up in the distribution, not the mean. A 10-minute Y2KDASH run before and after gives you the full before/after picture across percentiles.
- Diagnosing intermittent issues. "Video calls drop every day around 8 PM" is invisible to a one-shot test. The continuous monitor captures the pattern.
- Comparing two ISPs or two connections. Run Y2KDASH on each for an hour. Compare the distributions. Much more informative than comparing two single-shot grades.
- Knowing when the connection is normal vs. degraded. The rolling chart shows your personal baseline. Deviations become obvious.
FAQ
Which is more accurate, Y2KDASH or Waveform?
Both use the same measurement principle (parallel ping probes during saturated download/upload) and produce comparable numbers within measurement noise. Waveform's single-shot test runs a fixed-duration stress cycle. Y2KDASH samples repeatedly over time and reports the distribution. For a definitive one-number answer either tool works; for understanding how a connection behaves across hours, only Y2KDASH provides the data.
Do Y2KDASH and Waveform use the same test servers?
Y2KDASH uses Cloudflare's speed.cloudflare.com endpoints and Cloudflare's global Anycast edge. Waveform uses their own servers. Both are client-side measurements, so the network path between the user and either test endpoint determines the minimum latency floor; the bufferbloat delta is independent of which server is used.
Can I use both tools? Do results agree?
Yes to both. Running Waveform and Y2KDASH on the same connection produces closely matching loaded-latency numbers (typically within 5–10 ms of each other). Running both is a useful way to confirm a measurement is not a single-tool artifact.
Why did DSLReports shut down, and does that affect these tests?
DSLReports Speedtest went offline in March 2025 after the operator wound down the site. It had been the canonical bufferbloat test for a decade. Waveform and Y2KDASH are the two browser-based replacements for most users. Neither shares DSLReports' infrastructure; both were built independently.
Is Y2KDASH or Waveform better for before/after a router change?
Y2KDASH is better because it captures the distribution over time, not just a single sample. A router change often affects tail latency more than average latency, which a one-shot test misses. A 10-minute Y2KDASH sample before the change and a 10-minute sample after gives a clear delta across the full distribution.
Try both
Both are free, both are in-browser, both start instantly. Running both on the same connection is the fastest way to build an accurate picture of where you stand.
Related reading: The Speed Test Lie · How to Fix Bufferbloat · Networking Glossary