Asus ROG Strix Scar 17 review

Frame rates

1920×1080, High preset, upscaling off.

Measured on this machine. Source: Notebookcheck, 2023-08-22.

In use

The short version

The Asus ROG Strix Scar 17 in this configuration is the fastest machine in every comparison our record can make for it, and the measured frame rates are not close. Against a Schenker XMG Neo 16 carrying the same RTX 4090 it is ahead by an average of 19fps across their three shared games. Against an Alienware m18 R1, also on an RTX 4090, it is ahead by 20fps across three. Against a Schenker XMG Apex 15 built around an RTX 4050 it is ahead by 167fps across seven. That last figure is the one that tells you what tier of hardware you are looking at; the first two tell you something more interesting, which is that at the very top of the RTX 40 range the chassis still matters, but only at the margins.

The specification on our record is a Ryzen 9 7945HX3D — the 3D V-Cache variant, with the stacked cache that disproportionately helps games bound by the processor rather than the graphics card — paired with an RTX 4090, 32GB of memory and a 17.3in 2560x1440 panel running at 240Hz. It weighs 2.935kg and carries a 90Wh battery. That combination of parts explains almost everything about the shape of the benchmark data below, including why the lightest titles in the set run at frame rates most laptops never approach.

Two caveats need stating before the detail, because both change how you should read this page.

The first is that our record does not hold a GPU power limit for this machine. We are stating that plainly rather than estimating, because an estimate would be worthless and a wrong figure would be worse. This is a genuine gap in what we know, and it is not a trivial one: on mobile Nvidia hardware the power limit is what separates two laptops carrying the same GPU name. The RTX 4090 in this machine and the RTX 4090 in the Schenker XMG Neo 16 are the same silicon, but our record holds a 175W figure for the Schenker and nothing for the Asus. When you see the Asus finish 19fps ahead across three shared games, you cannot attribute that gap to a power-limit difference, because we do not know whether there is one. If you are cross-shopping specific listings, the TGP is the number to hunt for in the retailer's own spec sheet, and it is worth being suspicious of any listing that will not tell you.

The second is that this page carries no score, because the benchmark suite we score on is incomplete for this machine. We average every scored laptop over the same five games so that the ranking means something; here we have three of them — Dota 2, GTA V and X-Plane 11 — and no figures for Cyberpunk 2077 or Baldur's Gate 3. There is a Cyberpunk number in the data, but it is for version 1.6, a build old enough that it is not comparable with the current one, so it cannot be used to fill the hole. A partial average against a complete average is not a comparison, it is a category error, so we do not publish one.

Buy it if you want the fastest 17in RTX 40-class laptop the benchmark data can identify, you are going to sit it on a desk, and you understand that you are paying a premium for the last few percent of performance. Look elsewhere if you play mostly esports titles at 1080p, where far cheaper machines already exceed anything your monitor can display; if you need the machine to move around with you regularly; or if you cannot get the retailer to confirm the GPU power limit, in which case you are buying a name rather than a known quantity.

Performance

The missing power limit, and why it matters more than the GPU name

Everyone shopping for a gaming laptop learns the model numbers first — 4050, 4060, 4070, 4080, 4090 — and most people stop there. That is a mistake, and it is a bigger mistake at the top of the range than at the bottom. Nvidia lets laptop manufacturers configure the same GPU across a wide band of power limits, and the difference between the bottom and the top of that band is routinely larger than the difference between two adjacent GPU tiers. A heavily restricted higher-tier card can lose to a generously fed lower-tier one. This is not an edge case, it is the normal state of the mobile GPU market, and it is why a specification that lists only "RTX 4090" is telling you which chip is inside without telling you how hard it is allowed to work.

For this machine, that is exactly the information we are missing. Our record has an RTX 4090 and no wattage. We are not going to invent one, and we are not going to reason backwards from the frame rates to a plausible figure and present that as fact, because the frame rates are influenced by the processor, the memory configuration, the cooling design and the test conditions as well as the power limit. What we can say is that the measured results are consistent with a machine that is not being held back — a 17in chassis has room for a cooling system that a thinner design does not, and the numbers at the top of the chart are extremely high. But "consistent with" is not "confirmed", and we would rather leave the gap visible than paper over it.

The practical advice that follows from this is simple. If you are choosing between this and another RTX 4090 machine, get the power limit for both in writing before you decide. Two listings for the same GPU can be separated by a substantial performance gap that no product photograph will reveal, and the retailers who quote the figure are generally the ones with a figure worth quoting.

The top of the chart: where the GPU stops being the limit

Notebookcheck's measured figures at 1920x1080 on High with upscaling switched off open with F1 2021 at 470fps, Strange Brigade at 460fps and The Witcher 3 at 414fps. Numbers of that magnitude are worth understanding properly rather than simply admiring, because they are not primarily a statement about the graphics card.

At 1080p on High settings, an older or deliberately lightweight engine does not give a top-tier GPU enough work to do. The graphics card finishes each frame long before the processor has finished preparing the next one, and the frame rate settles at whatever the CPU, the engine's own internal limits and the driver overhead will allow. That is what 470fps in a 2021 racing title represents: a ceiling imposed by everything except the GPU. It is also where the Ryzen 9 7945HX3D earns its keep. The 3D V-Cache variants carry a large stacked cache that helps precisely in this scenario — games whose bottleneck is the processor's ability to feed the pipeline rather than the GPU's ability to render. If you were wondering why this machine posts numbers at the top of the chart that other RTX 4090 laptops do not, the processor is a substantial part of the answer, and it is why the margins against the other two RTX 4090 machines below are non-zero at all.

The practical consequence of a 470fps result is not that you will play at 470fps. The panel refreshes 240 times per second. Everything above that is headroom, and headroom is not worthless — it converts directly into consistency, so that when the engine hits a demanding scene the frame rate falls from an absurd height to a merely high one rather than dropping below the refresh rate — but it is not performance you can see. Anyone whose library consists mainly of titles in this bracket is buying a great deal of GPU they will never use.

F1 23 at 374fps, Shadow of the Tomb Raider at 293fps and F1 22 at 281fps continue the pattern. There is an oddity in that sequence worth flagging honestly: F1 23 runs faster here than F1 22, which is not the direction you would expect from a newer instalment of the same series. It is not impossible — engine revisions, driver maturity and per-title optimisation all move these numbers, and the two games were not necessarily benchmarked under identical driver versions — but it is a reminder that a benchmark table assembled over time is not a controlled experiment across titles. Comparisons between machines on the same title are sound. Comparisons between titles on the same machine tell you about the games as much as the hardware.

The middle of the chart: the range most buyers will actually live in

Dota 2 at 244fps is the first of the three scoring-suite games we hold, and it lands just above the panel's 240Hz refresh rate. That is a satisfying result for a competitive title on a 240Hz screen — it means the panel is saturated with a small amount to spare — but it is also a warning about how you should read the value of this machine. Dota 2 is not a demanding game. Very much cheaper laptops clear any common panel refresh rate in it comfortably. If your play is dominated by titles of that weight, the money spent on an RTX 4090 is buying frames your eyes and your monitor cannot resolve.

Far Cry 5 at 232fps, Cyberpunk 2077 (version 1.6) at 205fps, Assassin's Creed Valhalla at 195fps, GTA V at 187fps, Far Cry 6 at 182fps and Final Fantasy XV at 180fps form a tight band between 180fps and 232fps. This is the interesting part of the data set, because these are real games with real workloads rather than synthetic-feeling engine showcases, and the machine is comfortably above the refresh rates most gaming panels sit at, while falling short of the 240Hz this particular panel offers.

That band is where the reasoning about this machine should happen. Two hundred frames per second in an open-world title at 1080p High is an enormous amount of performance in absolute terms. It is also, crucially, performance with a large safety margin: a game that averages 195fps has room to lose a third of its frame rate in a busy scene and still be a long way clear of the point where it stops feeling smooth. Machines that average a fraction of that in the same titles have no such margin, and that is what buyers are really purchasing at this tier — not a higher number, but a floor that does not collapse.

X-Plane 11 at 166fps is the third scoring-suite game, and it sits noticeably below GTA V and far below Dota 2. That is expected and it is one of the reasons the title is in the suite: flight simulation is heavily bound by the processor and by simulation work that no graphics card can accelerate, so it acts as a check on machines that would otherwise look uniformly fast. A laptop that posts 244fps in Dota 2 and 166fps in X-Plane 11 is telling you that its performance is not uniform across workloads, which is true of every machine and worth remembering when a headline figure is quoted from a single game.

The bottom of the chart: the two titles that separate this tier from the one below

Alan Wake 2 at 116fps and Star Wars Jedi Survivor at 106fps are, by a wide margin, the hardest results in the set, and they are the ones that justify the hardware.

Alan Wake 2 is a genuinely punishing title — it was built to stress modern GPU features and it does. A result of 116fps at 1080p High with upscaling switched off is very strong. It is worth restating that qualifier: upscaling off. The measured figures here are all native-resolution results, and this is a machine and a generation where DLSS is available and widely used. Turning it on would raise these numbers substantially. We do not publish estimates of by how much, because we do not have measurements of it, but the point stands that 116fps native is a floor rather than a ceiling for how this game can be made to run.

Jedi Survivor at 106fps is the lowest figure in the set. This title has a documented history of being demanding and inconsistently optimised, and a triple-digit average on it at native 1080p is a good result. It is also the number that most usefully characterises the machine: the worst case in sixteen measured games is 106fps. That is the practical definition of a top-tier gaming laptop. It does not mean every future game will behave that way, and titles released well after this data was gathered will be heavier, but a machine whose floor is 106fps across a wide and varied set has meaningful runway.

Look at the shape of the whole distribution and the picture is coherent. The set runs from 470fps down to 106fps — a spread of more than four to one. That spread is not noise and it is not inconsistency in the machine. It is the difference between games that are limited by the processor and games that are limited by the graphics card. In the first group, this machine's Ryzen 9 with 3D V-Cache is doing the heavy lifting and the GPU is idling. In the second group, the RTX 4090 is doing the work and the result is a triple-digit frame rate where a mid-range card would be in the fifties or sixties.

Why there is no score, in more detail

Our scoring suite is five games: Dota 2, GTA V, X-Plane 11, Cyberpunk 2077 and Baldur's Gate 3. Every laptop that carries a score is averaged over all five. We hold the first three for this machine and neither of the last two.

The Cyberpunk situation is the instructive one. There is a Cyberpunk figure in the measured data — 205fps — but it is tagged as version 1.6. Cyberpunk 2077 changed substantially after that build, both in what it renders and in how it performs, so a 1.6 result and a current-build result are effectively different benchmarks that happen to share a name. Substituting one for the other would inflate this machine's average against every laptop measured on the current build, which is precisely the sort of quiet error that makes a leaderboard useless. So we hold the number, label it, use it as evidence in prose, and refuse to let it into the average.

The honest consequence is that this machine cannot be ranked against the scored field, even though the frame rates below make it obvious roughly where it would land. The mean across the three suite games we do hold is 199fps, which is a high figure by any standard, but it is an average over an easier subset — Cyberpunk and Baldur's Gate 3 are both heavier than Dota 2 — so it is not directly comparable to a five-game average. Treat it as a description of this machine, not a position in a table.

The screen

A 17.3in panel at 2560x1440, 240Hz, IPS. Each of those three specifications is doing something different, and the interaction between them and the benchmark data is where the useful analysis is.

The resolution gap between the panel and the test data

Every measured figure quoted on this page was recorded at 1920x1080. The panel is 2560x1440. Those are not the same workload — 1440p asks the graphics card for roughly 78% more pixels per frame — and we do not hold 1440p measurements for this machine. That is a limitation of the data, and it needs to be stated clearly rather than glossed over, because it applies to the headline numbers on this page.

What it means in practice is that you should not read "244fps in Dota 2" as a promise of 244fps on this machine's own screen. In the CPU-limited titles at the top of the chart the drop from 1080p to 1440p will be small, because the graphics card was never the constraint in the first place — that is the defining characteristic of a CPU-bound result. In the GPU-limited titles at the bottom, the drop will be substantial, because there the extra pixels land squarely on the component that was already working hardest. Alan Wake 2 and Jedi Survivor are the two results most affected; F1 2021 and Strange Brigade the least. We are not putting figures on that because we do not have them, but the direction of the effect is not in doubt.

This is a broader point about 1440p gaming laptops generally: buying a higher-resolution panel converts GPU headroom into image quality. On a machine like this one, with a very large amount of headroom in most titles, that is a sensible trade and arguably the right one. On a mid-range machine it can be an expensive way to make your games run worse. Here the pairing makes sense — a top-tier GPU with a 1440p screen is a machine designed so the panel and the silicon are matched.

240Hz, and which games actually reach it

Taking the measured 1080p figures at face value for a moment — which, per the point above, flatters the panel's native resolution — seven of the sixteen games clear 240fps: F1 2021, Strange Brigade, The Witcher 3, F1 23, Shadow of the Tomb Raider, F1 22 and Dota 2. Seven more land between 166fps and 232fps, which is above what the more common gaming refresh rates could show but short of this one's ceiling. Two — Alan Wake 2 at 116fps and Jedi Survivor at 106fps — sit well below it.

So a 240Hz panel is genuinely useful here rather than a marketing line, but it is useful for a specific slice of the library: competitive and older titles, where a high refresh rate has a real effect on how responsive the game feels. For a single-player narrative game running at 116fps, the difference between this panel and a lower-refresh one is essentially nothing. The refresh rate earns its place because this machine is fast enough to feed it in a meaningful number of titles, which is not something you could say about most laptops sold with the same figure on the box.

IPS rather than OLED or Mini LED

Our record specifies an IPS panel. Within the top tier of 2023-era gaming laptops, IPS was one of three common choices, alongside Mini LED and OLED, and it is the pragmatic one. IPS gives you dependable viewing angles and, in high-refresh gaming implementations, none of the compromises that come with the alternatives — no risk of static-element burn-in over years of use with the same game HUD or the same desktop taskbar, and no local-dimming behaviour to manage.

What our record does not hold is any measurement of this panel's brightness, contrast or colour coverage. Those matter, especially at this end of the market, and especially against Mini LED rivals whose entire pitch is contrast and peak brightness in HDR content. We are not going to describe a display we have no measurements for. If HDR performance or colour accuracy for creative work is part of why you are considering this machine, that is a specification to verify against the specific listing you are buying, because IPS panels vary enormously and the label alone tells you very little.

Build, size and portability

A 17.3in chassis weighing 2.935kg with a 90Wh battery. The arithmetic here is unforgiving and there is no way to present it as anything other than what it is: this is a desk machine.

Under three kilogrammes is not exceptional for a 17in gaming laptop — the class routinely runs heavier — but it is roughly three times the weight of an ultraportable, and that figure excludes the power supply. Our record does not state what the charger weighs, and on a machine with an RTX 4090 it will not be small. A realistic travelling weight is the laptop plus a substantial brick plus a bag rated to carry both safely, and anyone who has carried that combination through an airport knows how quickly the difference between 2kg and 3kg stops being abstract.

The size has a compensating benefit that shows up directly in the benchmark data. A 17.3in chassis has physical room for larger fans, longer heat pipes and more exhaust area than a 14in or 16in one. That is a large part of why the highest-performing laptops in any generation tend to be the biggest ones, and why a manufacturer can configure a GPU aggressively in a machine this size without the cooling system giving up. The frame rates on this page are, in part, a consequence of the volume the designers had to work with. You cannot separate the performance from the size; they are the same decision.

The 90Wh battery deserves a note. It is close to the 100Wh limit that airlines generally impose for lithium batteries in cabin baggage, so it is about as large as a laptop battery is allowed to be. That is a good thing in principle. In practice, our record holds no battery-life measurement for this machine, so we are not going to tell you how long it runs. What can be said without any measurement at all is that no RTX 4090 laptop runs demanding games on battery for a useful length of time, and that all of them throttle heavily when unplugged, because the power draw under load exceeds anything a battery of this size can sustain. If you want to game away from a socket, this is the wrong category of machine, not merely the wrong model.

Our record also holds no dimensions beyond the screen size, no measurements of surface temperature or fan noise, and no assessment of the keyboard or the chassis materials. Those are real gaps. Surface temperature and noise in particular are where 17in gaming laptops differentiate themselves from one another, since they are all fast and they do not all achieve it quietly. We would rather point at the hole than fill it with something plausible-sounding.

How it compares

Our rival set for this machine is drawn from the same GPU generation and compared only on games both machines hold. The most striking thing about it is what is missing: there is no machine in our RTX 40 data that beats this one. Every comparison we can make, it wins. That is unusual, and it is the strongest single statement the data makes about where this configuration sits.

Against the Schenker XMG Neo 16 (Early 24) — the closest genuine rival

Same GPU, an RTX 4090, and our record holds a 175W power limit for the Schenker. Across three shared games the Asus is ahead by an average of 19fps.

This is the comparison that most repays thinking about. Two machines, the same graphics silicon, tested by the same outlet on the same methodology, and a gap of nineteen frames. Against averages that run into the hundreds, that is a difference of a few percent — real, repeatable, and small.

Where does it come from? Three candidates, and we can only be honest about which. It could be the processor: the Ryzen 9 7945HX3D's stacked cache is exactly the kind of advantage that shows up in CPU-bound titles, and at 1080p a meaningful fraction of any three-game sample is likely to be CPU-bound. It could be the power limit — but we cannot claim that, because we do not hold one for the Asus. It could be the cooling design, where a 17.3in chassis has an inherent advantage over a 16in one in sustained load. Most likely it is some combination, and the shape of the comparison — a small gap between two machines with the same GPU, against an enormous gap to a machine with a lesser GPU — points at the GPU tier being the dominant factor and everything else operating at the margins.

The buying advice that follows is straightforward. If these two are both in front of you at similar prices, nineteen frames on averages of this size should not be the deciding factor. Screen size, panel technology, keyboard, warranty and the specific price on the day all matter more. If the Schenker is meaningfully cheaper, take it. What would change the analysis is a confirmed power limit for the Asus: if it turns out to be materially above 175W, the gap is explained and you know what you are paying for. If it turns out to be at or below 175W, then the Asus is winning on processor and cooling alone, which is a more interesting result and a more durable one, because it would hold across a wider range of games than a power-limit advantage would.

Against the Alienware m18 R1 — the same silicon in a bigger box

Also an RTX 4090; across three shared games the Asus is ahead by 20fps. Our record does not hold a power limit for the Alienware either, which limits how much can be concluded.

The pattern is now clear and it is worth naming. Two independent RTX 4090 machines, from two different manufacturers, in two different chassis sizes, both land within about twenty frames of this one. Consistency of that kind across independent comparisons is evidence that the result is real rather than a quirk of one particular test session. At the top of the RTX 40 stack, the machines cluster. The GPU sets the level; the rest of the design nudges it.

The m18 R1 is an 18in machine, which brings its own trade-offs — more screen, more cooling volume, more weight, and a chassis that is harder to fit in a bag or on a shared desk. Our record does not hold a weight for it in this comparison, so we will not put a number on that trade. But the performance question between these two is close to settled: they are the same tier of machine and the choice between them should be made on price, screen preference and support, not on twenty frames.

Against the Schenker XMG Apex 15 (Late 23) — what a GPU tier is actually worth

An RTX 4050 at 140W. Across seven shared games — a much larger and therefore more reliable sample than the three-game comparisons above — the Asus is ahead by an average of 167fps.

This is the number that puts the others in perspective. One hundred and sixty-seven frames is not a margin, it is a different class of machine. And note the detail that makes it more telling, not less: the Apex 15's RTX 4050 is configured at 140W, which is a high power limit for that GPU. This is not a comparison against a throttled, thin-chassis 4050 running at half its potential. It is a comparison against a 4050 being run about as hard as a 4050 can be run, and the gap is still enormous.

That matters for how you should interpret the tight three-game comparisons above. A sceptic could look at a nineteen-frame gap between two RTX 4090 machines and conclude that the benchmark is insensitive — that it cannot tell machines apart. The Apex 15 result disproves that directly. The same methodology, on overlapping titles, produces a 167fps separation when the hardware genuinely differs. The benchmark can separate machines. When it reports a small gap between two RTX 4090 laptops, the small gap is the finding.

There is a second reason the gap widens in a seven-game comparison rather than compressing. With more shared titles, the sample is more likely to include GPU-limited games where the graphics card is the constraint. In the three-game comparisons, a larger share of the sample is titles like Dota 2 and GTA V, where at 1080p both machines are partly waiting on the processor and the genuine GPU difference is masked. Add heavier games and the mask comes off. This is the single most important thing to understand about reading 1080p benchmark tables: light titles at low resolution systematically understate the difference between graphics cards, and the more of them there are in a comparison, the more compressed the result looks.

What the spread as a whole tells you

Put the three comparisons side by side — ahead by 19, ahead by 20, ahead by 167 — and a coherent model of the market falls out of them. Moving between GPU tiers changes everything. Moving between implementations of the same GPU tier changes very little, at least at the top of the range where the chassis are all large enough to cool the part properly.

For a buyer, that has one clear implication. Decide the GPU tier first, on the basis of what you actually play and at what resolution, and then choose between implementations on price, screen, weight and support. Agonising over which RTX 4090 laptop is fastest is a way of optimising the smallest variable in the decision. Choosing between an RTX 4090 and an RTX 4050 is the decision that determines whether Alan Wake 2 runs at a hundred and sixteen frames or somewhere far below it.

Who it's for

This machine suits a fairly specific person, and the benchmark data describes them precisely.

The buyer who plays demanding single-player games and wants a floor rather than an average. Alan Wake 2 at 116fps and Jedi Survivor at 106fps are the results that matter here. A machine whose two hardest measured titles both stay in triple figures at native 1080p is one that will not embarrass itself for several years, and one where enabling upscaling gives you a large further margin to spend on resolution or settings rather than on rescuing playability.

The buyer who wants a 1440p panel driven properly. A 2560x1440 screen on a mid-range machine is a way of making games run worse in exchange for sharpness. On this one there is enough headroom in most of the measured titles that stepping up to the panel's native resolution is a reasonable use of the surplus. That is the correct pairing of GPU and screen, and it is a large part of why this configuration makes sense as a whole rather than as a list of parts.

The buyer who will use it as a desktop replacement. At 2.935kg plus charger, on a 17.3in footprint, with a battery that will not sustain gaming, this is a machine for a desk. If that is what you want — a computer that happens to fold up and can be moved between two fixed locations when you need to — the size is a feature. It is why the cooling works and why the frame rates are where they are.

The buyer who mixes gaming with processor-heavy work. The Ryzen 9 in this configuration is a sixteen-core part and the frame rates in the CPU-bound titles reflect how strong it is. Our record holds no productivity benchmarks, so we are not going to quantify that, but a machine with this processor and 32GB of memory is not solely a games console. If your use case includes compilation, rendering or simulation alongside gaming, the specification supports it.

Who should buy something else

Anyone whose library is mostly competitive esports titles. Dota 2 runs at 244fps here. It also runs at frame rates far beyond any panel's ability to display it on machines costing a fraction as much. If you play Counter-Strike, League, Valorant, Dota or Rocket League and little else, an RTX 4090 is the wrong purchase almost regardless of budget, because the frames you are paying for cannot be shown to you. Buy a mid-range machine with a high-refresh panel and spend the difference on a monitor, a chair, or nothing at all.

Anyone who needs to carry it daily. Three kilogrammes plus a large charger is a commitment. If your machine has to travel between a home, an office and a train several times a week, look at 14in and 16in designs and accept the performance you lose. The trade is real in both directions, and a laptop you resent carrying is one you will use less.

Anyone considering another RTX 4090 machine that is meaningfully cheaper. The evidence above is that same-tier machines cluster within about twenty frames of one another on averages that run to the hundreds. That is not a difference worth paying a large premium for. If a Schenker XMG Neo 16 or an Alienware m18 R1 is significantly less money on the day you buy, the frame rate data does not justify choosing this one over it.

Anyone who cannot get the power limit confirmed and is buying on performance alone. This is the honest bottom line of the gap in our record. The measured figures on this page are strong, and they were produced by a specific test unit under specific conditions. If the listing you are looking at configures the GPU differently from the unit Notebookcheck measured — which happens, particularly across regions and revisions — then the numbers here may not describe the machine you receive. Ask. A retailer who quotes a TGP is telling you they know what they are selling.

Anyone who needs a verified price-to-performance judgement before committing. We do not hold a confirmed UK price for this configuration, and we will not guess at one. That is why there is no value assessment on this page. Everything above is a statement about how fast the machine is and how it compares to its peers on measured frame rates. Whether it is worth what a given retailer is asking is a question only the price tag in front of you can answer — and at this tier, where near-identical performance is available from several manufacturers, the price tag is very often the entire argument.

What we hold, and what we don't

In the interests of being clear about the evidence base: our record for this machine holds a full specification — GPU, processor, memory, panel size, resolution, refresh rate, panel technology, weight and battery capacity — and sixteen measured game results from Notebookcheck at 1920x1080 on High with upscaling disabled.

It does not hold a GPU power limit, a confirmed UK price, any 1440p frame rate figures, any battery-life measurement, any display brightness or colour data, any thermal or acoustic measurements, and no storage capacity for this particular configuration. Three of those gaps are load-bearing. The power limit is the most important, for the reasons set out above. The absence of 1440p figures means the headline numbers describe a resolution below the panel's native one. And the absence of a price means no value judgement is possible, which is why this page carries no score.

Everything else on this page is derived from the measurements we do hold, and the comparisons are made only on games both machines have been measured in. That is a narrower basis than a full review, and it is deliberately so. What it gives you is a set of numbers that can be traced to a source, and an argument built only on those numbers.