Why this page has no score
Every other review in this section covers a product. This one covers a skill.
CoreForge does not run a benchmark lab. We are not going to publish frame rate charts, because the frame rate numbers you see online are the product of expensive equipment, controlled test benches and a great deal of work by people who do this professionally. Republishing their tables as if they were ours would be dishonest, and it would take traffic away from the people who earned it.
What we can do is teach you to read their work properly. That is what this page is. There is no score at the bottom because a score would be meaningless here.
Part 1: how to read a benchmark
The average FPS is the headline. The 1% lows are the experience.
Most charts lead with average frames per second. It is the easiest number to compare and the least useful on its own.
What you actually feel while playing is consistency. A game averaging 90 FPS that regularly dips to 40 feels worse than one locked at a steady 70. That is what 1% lows and 0.1% lows measure: the worst moments, not the typical ones. When a reviewer publishes both, read the lows first. If a chart only shows averages, it is telling you half the story.
The test system is part of the result
A frame rate number is meaningless without its conditions. Before you compare two charts, check that they share the same:
- Resolution and settings preset
- CPU, RAM speed and motherboard
- Driver version and game patch
- Scene or benchmark run used
Two reviewers testing the same GPU can legitimately publish different numbers because they tested different things. That is not a contradiction, it is methodology. This is also why comparing a number from one site against a number from another site is usually invalid. Compare within a single reviewer's chart, where everything else was held constant.
Why CPU reviews test at 1080p with a flagship GPU
This confuses a lot of people, and it looks absurd at first: why would anyone test a $600 processor at the lowest resolution with the most expensive graphics card available?
Because the goal is to isolate the CPU. At low resolution, the graphics card finishes its work quickly and the processor becomes the limiting factor, so differences between CPUs become visible. Raise the resolution and the GPU becomes the bottleneck, and every CPU starts to look the same.
The practical consequence matters for your buying decision. The large gaps you see between processors in 1080p charts shrink at 1440p and often nearly disappear at 4K. That does not make the test wrong. It means you have to ask which scenario matches your setup. We flagged exactly this in our Ryzen 7 9800X3D coverage: it is the fastest gaming chip of its generation, and the size of that lead depends heavily on your resolution and your GPU.
Upscaling and frame generation break comparisons
DLSS, FSR and XeSS render the game at a lower internal resolution and reconstruct it. Frame generation inserts frames that were never rendered. Both raise the number on the chart. Neither is the same thing as native performance.
So a native 1440p result and a DLSS Quality result are not comparable, and a frame-generated result is not comparable to either. Reviewers who do this well label it clearly. Marketing material frequently does not, or buries it in a footnote. When you see a spectacular generational leap advertised, check whether both bars used the same rendering path.
A handful of games is not a verdict
GPU performance varies significantly by engine and by title. A chart covering five games tells you about those five games. If none of them is a game you play, the chart is weakly relevant to you.
Look for reviewers who test a broad suite, and pay attention to the outliers rather than only the average of averages. Better still, look for results in the specific games you spend your time in.
Manufacturer charts are marketing
This is not a moral judgment, it is just what they are. First-party performance claims are produced by the company selling the product, using conditions it chose. The footnotes underneath usually disclose those conditions, and they are worth reading precisely because so few people do.
Use first-party material for specifications, which are verifiable, and independent testing for performance, which is contestable.
One reviewer is a data point, not a consensus
Before you act on a benchmark, check whether two or three independent outlets reached similar conclusions. When they disagree, the reason is usually methodological, and finding that reason teaches you more than the numbers did.
Part 2: which resolution should you target?
Start with what people actually run
It is worth grounding this in real data rather than enthusiast assumptions. According to The FPS Review's analysis of Valve's April 2026 Steam Hardware Survey, 1920x1080 remained the dominant primary display resolution at 52.21%, with 2560x1440 at 21.41% and climbing steadily, and 4K at 5.09%.
Read that carefully. Half of PC gaming still happens at 1080p. 1440p is the clear growth resolution but is still roughly a fifth of the market. 4K remains a small minority. If you feel behind for gaming at 1080p, you are not.
The rule: buy the pair, not the part
The most common expensive mistake is buying one half of the equation. A 240Hz monitor fed 80 frames per second wastes most of what you paid for. A card capable of 160 frames connected to a 60Hz display shows you 60 of them.
So decide the pair together:
1080p high refresh. Right if you mainly play competitive shooters, if your budget is tight, or if you already own a fast 1080p panel. High frame rates are far easier to reach here, and a mid-range card can drive a 144Hz to 240Hz panel comfortably.
1440p at 144Hz to 180Hz. The sensible default for a new mid-range build in 2026. Meaningfully sharper than 1080p at 27 inches, and a current mid-range card can drive it at high frame rates. This is the pairing behind the parts we have covered in this section.
4K. Justified when image quality matters more to you than frame rate, when you sit close to a large display, or when you also do creative work. It demands a substantially more expensive GPU to hold high frame rates, which is why its share remains small.
The VRAM question
One detail deserves attention because it constrains real decisions. Reporting on the June 2026 Steam survey noted that 8GB remains the most common VRAM configuration, held by roughly a quarter of players, while the commonly cited comfort level for 1440p at high settings sits higher than that.
The honest reading is not that 8GB cards cannot play at 1440p. It is that they require settings discipline: texture quality is usually the first thing to lower, and it often costs less visual quality than people fear. Buying a card with more VRAM buys you fewer compromises later, which is exactly why we flagged 12GB as a point of discussion in our RTX 5070 coverage.
Settings are a dial, not a switch
Ultra presets frequently cost a large share of your frame rate for a difference you would struggle to see in motion. Dropping shadows, volumetrics and ambient occlusion by one notch is often the single highest-value performance change available, and it is free.
Before concluding that your hardware is inadequate for a resolution, try the resolution with sensible settings rather than maximum ones.
Part 3: a worked example
The parts covered elsewhere in this section form one coherent build, and the logic connecting them is exactly the logic above.
A Ryzen 7 9800X3D is a top-tier gaming processor. An RTX 5070 is a card positioned for 1440p. An LG UltraGear 27GP850-B is a 27-inch 1440p panel running 165Hz with a 180Hz overclock. Those three agree with each other: the resolution target matches the card's class, and the refresh rate is within reach rather than aspirational.
Now apply the reading skills. If you compared that CPU against cheaper processors using 1080p charts, you would see large gaps. At the 1440p this build actually targets, those gaps narrow considerably, because the GPU is doing more of the work. That is not an argument against the processor. It is an argument for knowing which chart applies to you, and for being honest that a cheaper CPU would have been defensible in this specific pairing.
That is the whole point of this page. The chart is not wrong. The question is whether it describes your situation.
Part 4: the Canadian angle
Two local considerations change the calculation here.
First, component pricing pressure through 2026 has made buying the correct tier unusually important. Memory and storage pricing in particular has swung dramatically, as we documented in our DDR5 and SSD coverage. When prices are volatile, overbuying one component to chase benchmark numbers costs more than it used to, and mismatched pairs are more expensive to correct.
Second, prices quoted in benchmark reviews are almost always in US dollars. A performance-per-dollar chart built on US pricing does not describe the Canadian market once you add conversion and GST plus QST. Recalculate value using the price you would actually pay on Amazon.ca or at a Canadian retailer, not the figure in the review.
What we will not do
We will not republish other outlets' frame rate tables. Their benchmark data is the product they invest in, and lifting it would be both a copyright problem and a dishonest one.
When we reference performance findings elsewhere on this site, we summarize conclusions in our own words, attribute them, and link to the original so you can read the full test. If you want the numbers, the sources below are where to get them, and they deserve the visit.