Test Your Ping Under Pressure
A fast connection can still feel laggy when uploads, downloads, cloud backups, streams, or other devices fill your router's queues. This test measures how responsive your connection remains while it is busy.
Test server: DQT Edge (Cloudflare) · same origin as this site · …
⚠ This test transfers a lot of data — avoid running it on a metered mobile connection.
For the clearest router/ISP result: use Ethernet, pause VPNs and large downloads, keep this tab active, and run the test more than once if results vary.
Latency over time
RTT samples · phase bands · median & p95DQT grades the increase in median latency while your connection is saturated. The grade is a practical diagnostic indicator, not a universal networking certification.
Latency
Speed
Connection readiness
DQT practical guidelines — not official universal standards.Latency over time
RTT samples · phase bands · median & p95Detailed measurements
| Phase | Min | p25 | Median | Mean | p75 | p95 | Max | Jitter | Failed | Samples |
|---|
Diagnosis & recommendations
Router options vary by model and ISP. Record existing settings before making changes. Windows tweaks cannot fix queueing that happens inside your router or modem.
Still Getting Lag or Ping Spikes?
A clean speed test does not rule out driver, adapter, Wi-Fi, routing, background-process, or game-configuration problems. DQT can inspect the full path from Windows and your network adapter to your router and game settings.
What is bufferbloat?
Bufferbloat is excess latency caused by oversized, unmanaged queues in your router or modem. When bulk traffic — a game download, a cloud backup, a 4K stream — fills the line, packets pile up in those buffers. Your time-sensitive traffic (game packets, voice, video calls) then waits behind the bulk data, and your ping climbs from a few milliseconds to hundreds.
Why can fast Internet still feel laggy?
Speed and responsiveness are different properties. A speed test measures how much data can move per second; responsiveness is how quickly a single small packet makes the round trip while the line is busy. Deep buffers help raw throughput benchmarks but poison latency — which is why a gigabit connection can feel worse in a firefight than a well-managed 100 Mbps line.
How this DQT test works
The test measures round-trip latency to the same endpoint this website is served from, first while your connection is idle, then while it saturates your download with parallel streams, and again while it saturates your upload. Latency probes continue through every phase on a kept-alive connection. Your grade is based on how much the median latency rises under load — the DQT scale runs from A+ (≤5 ms increase) to F (>120 ms). Raw percentiles, jitter and failed probes are shown in the detailed table; unreliable or unsaturated runs are marked inconclusive instead of being graded.
Test data is generated on the fly and discarded — uploads are never stored, and no account or personal information is required.
Why upload bufferbloat hurts gaming and calls
Upload capacity is usually a fraction of download, so it saturates first — one phone backing up photos can fill it completely. Because your game actions, voice audio and webcam feed all travel upstream, a bloated upload queue delays exactly the packets you feel most. Upload bufferbloat is the most common cause of "my ping spikes when someone starts streaming."
How to reduce bufferbloat
The reliable fix is Smart Queue Management (SQM) on your router — algorithms like CAKE and FQ-CoDel keep queues short by design. Set the shaper slightly below your real-world throughput (start around 90% of what this test measures, then retest and adjust). If your ISP box has no SQM, a small router running OpenWrt or similar firmware in front of it is a common solution. Note that Windows-side tweaks cannot fix queues that live inside the router or modem.
Ethernet versus Wi-Fi testing
Wi-Fi adds its own variable latency (airtime contention, interference, retransmissions) that can look identical to bufferbloat. Test on Ethernet to grade your router and ISP cleanly; then repeat on Wi-Fi — the difference between the two runs is your wireless overhead, which is fixed with placement, channels or better access points rather than queue management.
Frequently asked questions
What is a good bufferbloat grade?
On the DQT scale, A+ or A means your median latency rises 15 ms or less while the connection is saturated — gaming and calls should stay responsive. C or below means queues are adding noticeable delay under load.
Why does my ping increase while uploading?
Upload bandwidth is usually much smaller than download, so a single backup or stream can fill the router's upload queue. Your game packets then wait behind bulk data, which appears as ping spikes.
Can bufferbloat cause gaming lag?
Yes. Bufferbloat raises latency exactly when the connection is busy — someone streaming or downloading on your network can add tens or hundreds of milliseconds to your ping even though your speed test looks fast.
Does faster Internet eliminate bufferbloat?
No. Bufferbloat is about queue management, not capacity. A gigabit line with deep unmanaged buffers can still spike badly under load, while a slower line with Smart Queue Management can stay flat.
Should I test over Ethernet or Wi-Fi?
Test over Ethernet to measure your router and ISP cleanly. Then test over Wi-Fi: the difference between the two tells you how much of your latency comes from the wireless link itself.
What are SQM, CAKE, and FQ-CoDel?
Smart Queue Management (SQM) is a router feature that keeps latency low under load. CAKE and FQ-CoDel are the modern queueing algorithms behind it — they interleave small time-sensitive packets ahead of bulk transfers instead of letting queues grow.
Why are different speed tests giving different results?
Different tests use different servers, paths, durations and parallelism, and some are affected by CDN caching or server load. Focus on consistency across repeated runs of the same test rather than comparing absolute numbers between tools. This test's throughput can also be limited by the shared edge platform it runs on — its main purpose is the latency behaviour, which stays accurate.