Lason Arena Gaming See Reviews
Mechanical keyboard with subtle green backlighting on a desk, moody gaming-room atmosphere

Hands-on looks at headsets, keyboards, mice and headphones, plus gaming and tech industry news.

Build your own FPS graphing calculator with OSD and frame time tools

If you have ever watched your on-screen counter drop from a steady 120 to 80 and wondered whether it was a real hiccup or just a glitch in the overlay, you already know that raw FPS is a poor measure of smoothness. A graphing calculator built from frame time data tells a much richer story, showing exactly how each frame was spaced and where the engine stumbled. For Australian PC enthusiasts who spend good money through retailers like PCCasegear, Mwave and JB Hi-Fi, that detail matters when weighing whether a new GPU, faster SSD or higher refresh monitor will actually be felt in game.

The good news is that the same tools used by hardware reviewers are free, lightweight and well supported on Windows. With a bit of wiring between an on-screen display, a capture backend and a spreadsheet, you can produce professional looking frame time graphs from the comfort of your loungeroom. The rest of this guide walks through the metrics worth plotting, the software stack that does the heavy lifting, and how to turn those curves into smarter upgrade decisions.

Why frame time graphs tell the truth raw counters hide

FPS counters are averages, and averages are flattering. They report the number of frames rendered each second but collapse every spike, dip and microsecond stall into a single tidy number. A card pushing 100 FPS might feel silky or it might feel like wading through treacle, depending on how those frames are spaced. Frame time is the missing variable: it is the actual duration, measured in milliseconds, between the start of one frame and the next. Plot frame time as a line and the shape of your gameplay experience appears.

Smooth gameplay shows a tight, almost flat line. A second-long stutter shows as a single tall spike. A persistent microstutter, the kind that makes a high-end system feel cheap, shows as a regularly repeating saw-tooth pattern. None of these patterns are visible in a counter that simply reads 60 or 144. Once you have a graph, you stop guessing and start diagnosing, which is why seasoned reviewers from Sydney to Perth lean on frame time captures instead of headline numbers when they put a new build through its paces.

The metrics worth putting on your graph

Before you reach for the software, decide what you actually want to plot. Average frame rate is the obvious one but it is the least useful. The metrics that genuinely separate a good build from a great one are percentile lows. The 1% low tells you the worst frame time across the slowest one per cent of frames, while the 0.1% low drills even deeper into the rare but noticeable judders. Plot both as horizontal lines alongside your frame time curve and you instantly see whether a stutter is rare or constant.

Frame time variance and frame pacing deserve their own graph too. Variance measures how far each frame time strays from the average, and frame pacing tracks the consistency of delivery between frames. A 60 FPS experience with poor frame pacing feels worse than a locked 55 FPS experience with even pacing, because the human eye is remarkably good at picking up irregular motion. Many Aussie builders chasing high refresh rates for their new 240 Hz panels overlook pacing entirely and end up with monitors that look worse than they should, simply because the data behind the panel was never examined.

The tool stack for on-screen display and capture

The backbone of any home graphing calculator is a tight toolchain. RivaTuner Statistics Server, paired with MSI Afterburner, remains the gold standard for the on-screen display. It hooks into DirectX, Vulkan and many older APIs, lets you choose which values appear on screen, and adds negligible overhead when configured properly. For capture, two free utilities do the heavy lifting: PresentMon, which records raw frame data with timestamps, and CapFrameX, which wraps PresentMon in a friendly interface and exports CSV files ready for plotting.

For users who prefer an even more polished workflow, NVIDIA FrameView is a solid alternative that bakes the same capture approach into a tidy dashboard, and it works regardless of which brand of GPU sits in your case. On the visualisation side, Google Sheets or Microsoft Excel handle the heavy maths, while tools such as FrameView's built-in charts or specialised helpers like FrameTime Analyser provide ready-made graph templates. Every one of these tools is free, regularly updated, and small enough to keep on a secondary drive without crowding out your game library.

Setting up a capture rig that produces clean data

A graph is only as honest as the conditions in which it was captured. Before pressing record, close browsers, Discord, RGB control suites and anything else that might inject background load. Set Windows power plan to High Performance, lock your refresh rate to match the in-game target, and make sure VSync is disabled unless you are specifically testing how it behaves. Australian summers push cases hard, especially in Adelaide, Brisbane and inland parts of Western Australia, so verify your cooling before recording. A graph that drops every twenty minutes because the CPU throttled at 95 degrees tells you about your airflow, not your graphics card.

Drivers matter too. Use a clean install of the latest stable branch rather than the bleeding-edge optional releases when benchmarking for a buying decision, since the hotfix variants can shift performance by several per cent. Where possible, run benchmarks at the same time of day across captures so thermals start from a similar baseline. The NBN connection you use for online testing should remain consistent, but for offline benchmarks you can simply leave the router alone. A repeatable rig is the difference between a graph that informs a purchase and one that just decorates a forum post.

Running in-game benchmarks you can trust

Built-in benchmarks are the easiest starting point. Cyberpunk 2077, Shadow of the Tomb Raider, Far Cry 6, Forza Horizon 5 and Assassin's Creed Valhalla all ship with repeatable tests, and most Australian gamers already own at least one of them through Steam or the Epic Store. Run the same test three times, export each capture, and average the runs to wash out random variance. A short, scripted walkthrough in a game without a benchmark works just as well, provided you follow the same route each time and trigger the same scripted events.

For multiplayer or open-world titles, pick a quiet stretch of map, walk in a straight line for sixty seconds, and capture the slice. Record overlays rather than averages using CapFrameX or FrameView, and save the CSV alongside a screenshot of the OSD so you can verify later which run was which. Many local reviewers at sites like this one keep a shared folder of standardised routes, and adopting one of those is a quick way to make your results comparable to builds that other Australians are testing on the same hardware.

Reading the graphs like a hardware reviewer

Open the CSV in your spreadsheet of choice, calculate the moving average, drop in percentile markers and you will see the personality of each scene reveal itself. A flat curve with a single tall spike is usually a streaming asset hitch or a shader compile, common in modern open-world games and harmless most of the time. A curve that ramps steadily upward as the scene gets busier points to a CPU bottleneck, often visible in dense cities like Night City or Novigrad. A curve that stays flat until a sudden cliff drop suggests a thermal limit or a power supply issue, both worth chasing down before spending on a new GPU.

Comparing graphs across hardware is where the calculator earns its keep. Plot the same scene on two GPUs at the same settings and the shape of the curve often tells a more honest story than the average FPS headline. The cheaper card might post 90 FPS to the more expensive card's 110, but if its 1% lows are noticeably worse, the more expensive card is genuinely smoother, not just faster. That is the kind of insight worth its weight when comparing listings between Centre Com, Umart and a Perth-based system integrator before deciding where to spend.

Turning frame time data into smarter upgrades

Once you can read a graph, the upgrade path tends to write itself. A CPU-bound curve points at a faster processor, a faster memory kit or both, particularly on AMD Ryzen platforms where memory tuning has a real impact on frame pacing. A GPU-bound curve suggests a more capable graphics card, a resolution drop, or the addition of upscaling technologies like DLSS, FSR or XeSS, all supported by the latest drivers on shelves in Australian stores right now. A jagged curve that improves with VSync enabled but worsens with it off points at a refresh rate mismatch, and a new 165 Hz or 240 Hz monitor might do more for feel than a generational GPU leap.

Storage surprises are common too. Modern DirectStorage titles show up as periodic spikes when the game streams textures, and swapping a SATA SSD for a quality NVMe drive from a brand stocked at JB Hi-Fi can smooth those spikes dramatically. RAM shows up as percentile lows rather than averages, which is why two systems with identical headline FPS can feel completely different. None of this is guesswork once the graph is in front of you, and that is the real value of building a graphing calculator rather than trusting a single on-screen counter.

Practical recommendations for clean FPS graphing

Take the leap this arvo, retire the misleading counter on the corner of your screen and find out what your rig is really doing in every scene. Subscribe to Lason Arena Gaming for hands-on guides, deep dives and Australian-flavoured hardware coverage that helps you spend smarter and play smoother.