Skip to main content
Iopex Technologies

Melbourne, Victoria

The average frame rate is the number that hides the problem.

Iopex traces frame times on the cheap Android phones most of the world actually owns, isolates the percentile spikes the mean absorbs, and costs each feature in battery drain per hour of play. Send one gameplay trace and an engineer reads it.

Name collision. iOPEX Technologies, Inc. is a separate and considerably older business working in enterprise IT and business process services. It shares no owner, no staff and no system with this company, and nothing on this site is theirs.

A stack of thin translucent layers seen at a low angle, one of them glowing hot in the middle of the stack.

FIG. 01 / THERMAL BEHAVIOUR ACROSS A LONG SESSION (CONCEPTUAL)

One

Sixty frames a second is not the same as sixty smooth frames.

A game can report an average of sixty frames per second and feel wrong, because the average absorbs the frames that arrived late. What people notice is not the mean, it is the shape of the tail: the one frame in two hundred that took four times as long as its neighbours.

So the number we work to is the 99th percentile frame time, and after that the count of frames that missed their deadline. Both are worse numbers to put on a slide and both describe what a player actually experiences.

A dense row of vertical bars of varying heights, mostly even and pale, with a cluster of much taller glowing bars where the rhythm breaks.

FIG. 02 / FRAME TIME SERIES WITH A STALL CLUSTER (CONCEPTUAL)

Two

The device that matters is not the one on your desk.

Games get built on a recent flagship and tested on a recent flagship. A large share of the people who will install them are on a four year old handset with a small thermal budget, four gigabytes of memory, and a battery that has lost a third of its capacity.

That device does not just run the same game more slowly. It runs it differently: it throttles after eight minutes, it reclaims memory mid-level, and it schedules your render thread against three background apps you cannot see. Optimising for it is a different job from making a flagship faster.

Method

What we measure, and why those four.

Each has a defined method and a target that can fail. None of them is an average.

Measurements and methods
MeasurementMethodWhy not the obvious alternative
99th percentile frame time Per-frame timestamps captured on device across a fixed 20 minute scripted session, reported as a distribution Average frame rate hides every stall. A game with a good average and a bad tail feels broken to a player and fine to a dashboard
Missed deadline count Frames whose presentation time exceeded the display interval, counted rather than averaged A count is honest about how often it happened. A percentage of a large number sounds small at any severity
Time to thermal throttle Sustained play from a cold device until the governor reduces clocks, recorded in minutes with ambient temperature noted A three minute benchmark never reaches the state a real session reaches. Most published mobile performance figures are taken before the phone gets hot
Battery drain per hour of play Measured from the device's own fuel gauge across a full session, on a device at a stated cycle count and state of health Milliwatt figures from a fresh device do not describe a four year old battery, which is the one the player has

These are the four procedures we run. Each row is a method and the target it has to clear, written out so that anyone can run it again and get their own number. We publish the procedure rather than a headline figure, because a figure detached from its method is not evidence of anything.

Send us a trace

Send one gameplay trace — engineering@iopex.co.im reads it.

One capture off the oldest handset you care about, with the device model and the operating system build written in the mail. An engineer reads it against the four measurements above and answers inside five business days: where the tail sits, how many frames missed their deadline, and the next thing worth instrumenting. No form on this site, no list to join, no account.

Legal name
IOPEX TECHNOLOGIES PTY LTD
Entity type
Proprietary company limited by shares, registered in Australia
ACN
696 561 609
ABN
54 696 561 609
ABN status
Active from 25 March 2026
GST
Registered from 25 March 2026
State
Victoria
Check it
Search the ABN on abr.business.gov.au, at no cost and with no account