Benchmark charts are how we compare graphics cards, CPUs, and settings — but a single bar labeled “FPS” hides most of the story. Learning to read the numbers behind the bars helps you spot marketing spin, understand whether a part suits your monitor, and predict how a game will actually feel. Here’s what the figures mean and which ones to trust.
Average FPS isn’t enough
Averages vs percentiles
Average FPS is the mean frame rate over a test run. It tells you overall speed but nothing about smoothness. Two cards can average 100 FPS while one feels buttery and the other stutters. That’s why good reviewers report 1% low and sometimes 0.1% low figures — the frame rate during the worst moments. Big gaps between average and 1% low mean stutter.
Frame time
Frame time (in milliseconds) is the flip side of FPS: it measures how long each frame takes. Consistent frame times feel smooth; spikes feel like hitches even at a high average. A frame time graph that looks like a flat line is ideal; one full of tall spikes reveals stutter an FPS average would hide.
| Metric | What it tells you | Why it matters |
|---|---|---|
| Average FPS | Overall speed | Headline comparison |
| 1% low FPS | Worst frequent dips | Perceived smoothness |
| 0.1% low FPS | Rare worst spikes | Bad stutter/hitches |
| Frame time (ms) | Per-frame consistency | Reveals hidden stutter |
| 99th percentile | Same idea as 1% low | Alternate label |
Context the bars don’t show
Test settings and resolution
Always check the fine print: resolution (1080p, 1440p, 4K), quality preset, and whether upscaling (DLSS/FSR) or frame generation is on. A card that looks dominant at 1080p may be CPU-limited there and only stretch its legs at 4K. Comparing numbers from different settings is meaningless.
CPU vs GPU bound
If several GPUs post nearly identical results, the test is likely CPU-bound — the processor, not the graphics card, is the limit. That’s common at 1080p with fast cards. At 4K the GPU usually becomes the bottleneck and differences widen. Match the benchmark’s resolution to the monitor you actually own.
Reading it for your purchase
Decide what “enough” means for your display first. For a 1440p 144 Hz monitor you want a card whose 1% lows stay near 144 in your games, not just the average. Chasing a higher average that still dips below your refresh rate buys you numbers you won’t feel.
FAQ
Why do reviewers show 1% lows instead of just average FPS?
Because smoothness depends on the bad moments, not the mean. A high average with poor 1% lows feels stuttery, while a slightly lower average with strong 1% lows feels smooth. The 1% low number captures the experience your eyes actually notice during heavy scenes.
Should I trust one benchmark or several?
Several, across multiple games. A single title can favor one architecture or have a quirk. Looking at a spread of games — and multiple outlets — averages out anomalies and driver-specific wins, giving you a truer picture of how a part performs across your library.
Spotting misleading charts
Marketing slides and even some reviews present data in ways that flatter a product. Watch for a truncated Y-axis — a bar chart that starts at 90 FPS instead of 0 makes a 5% difference look like a landslide. Check whether “up to” performance claims come from a single cherry-picked title or a broad average. And be wary of comparisons that quietly change variables: enabling upscaling or frame generation on one product but not the other inflates the gap dramatically, since generated frames aren’t equivalent to natively rendered ones for latency and image quality.
Use benchmarks to plan, not to obsess
Numbers guide a purchase, but they aren’t the lived experience. A card that averages 130 FPS versus one at 138 FPS is indistinguishable in play; a 10% price difference for that gap rarely makes sense. Focus on the threshold that matters for your monitor — can this part keep 1% lows above my refresh rate in the games I actually play? — and stop there. It’s also worth checking the test system: a benchmark run on a top-tier CPU and fast RAM may not reflect the frame rates you’ll see on a more modest build, so match the review conditions to your own setup as closely as you can before drawing conclusions.
Bottom line
Read past the average: check 1% lows and frame-time consistency for smoothness, confirm the test resolution and settings match your setup, and watch for CPU-bound results that flatten GPU differences. Once you read charts this way, you’ll buy the right part for your monitor instead of the one with the tallest bar.
