Focus on MSPT, TPS, and real tick pressure
TickVisor uses Spark profile timing to estimate performance pressure and help identify whether the problem is entities, chunks, block entities, players, event handling, or runtime overhead.
Spark profiles
.sparkprofile
Lag triage
Turn Spark profile output into a readable fix list.
Spark gives you the data. TickVisor makes that data easier to act on by grouping timing noise into server systems, likely causes, and performance recommendations.
Why it matters
Spark profile analyzer for actual production servers.
Built around the files server teams already collect when TPS drops, MSPT spikes, or players start reporting lag in the least helpful way possible.
Convert profiler noise into owner-friendly language
Not every server owner wants to read raw method names for an hour. TickVisor summarizes what matters and preserves deeper details for technical review.
Best results come from real load
Capture your Spark profile while the issue is happening. Machines, players, mobs, chunks, and automation should be doing normal work during the capture.
What you get
Reports that turn noisy diagnostics into decisions.
Top systems
See grouped timing by server system instead of staring at isolated Java frames.
Likely causes
Prioritize the most plausible causes behind spikes, low TPS, or heavy tick time.
Report comparison
Compare before and after captures to confirm whether a fix helped.
FAQ
Common questions
Does TickVisor replace Spark?
No. Spark captures the profile. TickVisor analyzes the exported profile and turns it into a clearer report for decision making.
Which Spark file type should I upload?
A .sparkprofile export is the best starting point for tick performance issues. Logs can be added when errors or crashes need context.
How long should I profile for?
Profile long enough to capture the problem under normal load. A short clean capture is less useful than a capture taken while the lag is actually happening.
Start clean
Use one real capture and fix one thing at a time.
Upload useful diagnostics, read the report, apply one change, then compare another capture. Primitive? Yes. Effective? Also yes, annoyingly.