Asus Strix Scar 18 review

Frame rates

1920×1080, High preset, upscaling off.

Measured on this machine. Source: Notebookcheck, 2025-04-02.

In use

The short version

The Strix Scar 18 is Asus at the top of its gaming range: an RTX 5090 running at a 175W power limit, a Core Ultra 9 275HX, 64GB of memory, a 2TB SSD and an 18in 2560x1600 240Hz panel, in a 3.307kg chassis. It is a desktop replacement in the fullest sense, and on specification there is very little above it.

The complication is that the published benchmark data for this machine is unusually thin. Notebookcheck's figures cover six games, and only three of them belong to our five-game scoring suite. There are no Cyberpunk 2077 or Baldur's Gate 3 results, which is why this machine carries no score: every ranked laptop on this site is averaged over the same five titles, and a three-game average cannot honestly be placed against five-game averages. There is also no confirmed UK price, which independently rules out any value judgement.

That thinness is not just a bookkeeping problem, it actively distorts the comparisons. The three scoring games this machine does hold are Dota 2 at 200fps, GTA V at 162fps and X-Plane 11 at 122fps, and those happen to be the three least graphics-limited titles in the whole suite. Dota 2 is an esports engine, GTA V is from 2013, and X-Plane 11 is a flight simulator whose frame rate is famously dictated by the processor. Comparing a 5090 to anything else on that particular trio is close to the worst possible way to demonstrate what a 5090 does.

The results reflect exactly that. Against the Acer Predator Helios Neo 16S AI, an RTX 5070 Ti machine at £2,679, this laptop averages level across three shared games. Against the Lenovo Legion 5 Pro, also a 5070 Ti, it averages 1fps behind. Neither of those results means a 5090 at 175W performs like a 5070 Ti. They mean the three games available for comparison do not ask enough of a graphics card to separate them.

The comparison that does find a difference is against Schenker's XMG Neo 16 A25, another RTX 5090 machine, which averages 37fps ahead across the same three shared games. Since both carry the same class of GPU and the shared titles are heavily processor-dependent, that gap is much more likely to be a CPU-side story than a graphics one. It also proves the methodology can find real differences when they exist, which is what makes the level results against the 5070 Ti machines meaningful rather than merely inconclusive.

Buy this if you want a large, fully-specified 18in desktop replacement and you are comfortable buying on specification rather than on a complete benchmark record. Do not buy it expecting our frame rate data to justify it over cheaper machines, because on the games we hold, it does not.

Performance

What 175W means in an 18in chassis

The most informative number on any modern gaming laptop is the GPU power limit rather than the GPU name. Nvidia ships its mobile parts in a wide configurable range, and where a manufacturer sets that limit determines most of the outcome. Two laptops with identical GPU badges can be far apart; two with different badges can be indistinguishable.

The Strix Scar 18 runs its RTX 5090 at 175W, which is at the upper end of how these parts are configured, and an 18in chassis is the correct place to put that. Cooling capacity in a laptop scales with surface area, fan volume and heatsink real estate, and an 18in machine has substantially more of each than a 16in one. The 3.307kg weight and the large footprint are the price of the power limit; they are also what makes it viable.

What we cannot tell you is whether the machine sustains 175W over long sessions, because we hold no sustained power logs, thermal measurements or noise figures for it. We also cannot demonstrate the benefit of that power limit from the frame rate data, for the reason set out below: the games available do not lean on the graphics card hard enough to show it. This is the central frustration with this particular laptop's record. The specification says top tier and the benchmark set is not equipped to confirm or deny it.

Six games is a thin set, and it matters

Notebookcheck's data for this configuration covers six titles: Strange Brigade at 418fps, Dota 2 at 200fps, Final Fantasy XV at 199fps, GTA V at 162fps, X-Plane 11 at 122fps and Alan Wake 2 at 116fps.

Look at the release dates implied by that list. Strange Brigade, Final Fantasy XV, GTA V and X-Plane 11 are all long-standing benchmark fixtures from earlier hardware eras. Dota 2 is a competitive title whose engine has never been graphically demanding. That leaves Alan Wake 2 as the only genuinely modern, GPU-punishing game in the set.

The consequence is important and easy to misread. The lowest figure here is 116fps, and it would be natural to conclude that this machine never drops below 116fps. That conclusion is wrong. It never drops below 116fps in these six games, and this set contains almost none of the current releases that bring high-end hardware to its knees. There are no results for the heaviest recent Unreal Engine 5 open-world titles, nothing from the current wave of cinematic single-player releases beyond Alan Wake 2, and no ray-tracing-mandatory games. A broader suite on a machine of this class would produce results well below 116fps in the hardest titles, and we simply do not have them.

Read the numbers below with that in mind. They are accurate and they are sourced, but they describe the easier half of the modern gaming landscape.

Working through the numbers

Strange Brigade at 418fps is the ceiling measurement. It is a well-optimised Vulkan title that has stayed in benchmark suites precisely because it scales cleanly and puts no unusual load on a system. At 418fps the graphics card has long since stopped being the limiting factor, and the number should carry no weight in a purchase decision beyond confirming that nothing is obviously wrong.

Dota 2 at 200fps and Final Fantasy XV at 199fps sit almost exactly level, which is a useful coincidence because they arrive there by completely different routes. Dota 2 is limited by its engine and the processor; Final Fantasy XV is an older console-era port that a current GPU handles without effort. Neither is telling you much about the RTX 5090.

GTA V at 162fps is the most CPU-bound result in the set by some margin. That game shipped in 2013, its rendering demands were modest even then by current standards, and its frame rate on modern hardware is essentially a measure of how fast the processor can feed it. A 5090 is not the reason this number is 162 rather than something else, and no graphics card upgrade would move it meaningfully.

X-Plane 11 at 122fps is the result that will matter most to a specific audience. Flight simulation is one of the most processor-dependent workloads in gaming, and a three-figure frame rate in it is a genuine indicator of the Core Ultra 9 275HX's capability. If you are buying an 18in machine for simulation work, this is the number in the set with real predictive value for you.

Alan Wake 2 is the only modern-heavy result

Alan Wake 2 at 116fps carries a disproportionate share of the useful information here, because it is the only title in the set built to current rendering standards.

Remedy's engine is expensive per pixel by design, with heavy volumetric lighting and a rendering approach that has made hardware look slow since the game's release. Recording 116fps at 1080p on High with upscaling switched off is a strong result. It puts the machine well above the 60fps threshold with room to raise settings or resolution, and it is the one data point that genuinely demonstrates the 175W power limit doing work.

It is also the one number in this set where the comparison to lower-tier machines would have been informative, and unfortunately it is not among the three games shared with the 5070 Ti rivals. That is the crux of this machine's benchmark problem: the games that could distinguish it are missing from the overlap, and the games in the overlap cannot distinguish it.

If you want a single figure to represent what this laptop does in a demanding modern game, 116fps in Alan Wake 2 is it. Extrapolate cautiously. Other current releases are harder still, and this machine will produce lower numbers in them, at 1080p and considerably lower at the panel's native resolution.

Where the spread comes from

The six results run from 418fps down to 116fps, a factor of about three and a half. In a fuller set that ratio would be much wider. Three separate mechanisms are at work.

The first is engine era. Strange Brigade, GTA V and Final Fantasy XV were built when the target was filling many pixels cheaply. Alan Wake 2 was built when the target became per-pixel quality, with lighting and geometry pipelines costing several times as much per frame. No amount of graphics power collapses that gap because the games are not doing comparable amounts of work.

The second is processor dependence. Dota 2, GTA V and X-Plane 11 all have their frame rates substantially determined by the CPU, and at 1080p that dependence is at its strongest. When the resolution is low the graphics card finishes each frame quickly, so the processor becomes the thing everything waits on. Three of the six games here fall into that category, and all three of them are the scoring games this machine holds.

The third is resolution. Every figure quoted is at 1920x1080, while the panel is 2560x1600. That is close to twice the pixel count, and the games most affected are the ones already limited by per-pixel work rather than by the processor. Alan Wake 2's 116fps will fall furthest at native resolution; GTA V's 162fps will barely move, because the processor is already the constraint. We do not hold native-resolution measurements and will not estimate them, but the direction is not in question.

The processor is doing more of this than the GPU

Of the six measured games, four are substantially processor-influenced, and the Core Ultra 9 275HX is a large part of why the numbers look the way they do. That is not a flaw in the machine; it is a description of the benchmark set.

The practical implication is for a specific type of buyer. If you play strategy, simulation, management or CRPG titles, where the processor rather than the graphics card sets the pace, this machine's specification is genuinely well matched to your use, and the X-Plane 11 and GTA V figures are the relevant evidence. Civilization-style turn processing, large simulation ticks and dense city scenes all live in the same performance domain as those results.

If you play modern graphically demanding single-player games, the specification is also well matched, but the evidence for that in our data reduces almost entirely to Alan Wake 2. You would be buying on the strength of the 175W power limit and the 24GB of VRAM rather than on a measured record.

Memory, VRAM and storage

The configuration carries 64GB of system memory, 24GB of VRAM and a 2TB SSD, and all three are generous.

64GB is more than any current game requires. Its value is in everything else: virtual machines, video timelines, large project files, local AI work, or the ordinary reality of leaving a browser full of tabs open behind a running game. For gaming alone it is surplus, and if gaming is genuinely all you do, it is money that could have gone elsewhere.

The 24GB of VRAM is the more consequential number and one of the strongest arguments for this GPU tier. Texture memory demands in recent open-world releases have risen sharply, and when a card runs short the symptom is not a graceful frame rate decline but stuttering and texture pop-in that make a game feel broken regardless of the average. A 24GB allocation removes that concern for the practical lifetime of the machine at these resolutions. It also makes the laptop useful for local machine learning, where VRAM capacity rather than throughput is usually the binding constraint.

2TB is the right amount rather than a luxurious one. Modern installs routinely exceed 100GB and a handful will fill half the drive, but it is enough that you will not be uninstalling something every time a new release arrives.

The screen

240Hz against these frame rates

The panel is 18in, 2560x1600 and 240Hz. Of the six measured games, one clears 240fps at 1080p: Strange Brigade at 418fps. Dota 2 and Final Fantasy XV land in the high 190s to 200s, GTA V at 162fps, X-Plane 11 at 122fps and Alan Wake 2 at 116fps.

On this evidence the 240Hz refresh is headroom rather than a capability you will regularly saturate, at least in these titles at 1080p, and less so at the panel's native resolution. That said, the benchmark set contains only one competitive game, and competitive titles are exactly where high refresh rates earn their keep. Dota 2 at 200fps suggests the machine has the processor speed for that category. If your games are esports shooters not represented in this set, the 240Hz panel may well be doing real work for you; we simply do not have measurements to confirm it.

Where the refresh rate is unambiguously useful is that it costs nothing when unused. A high refresh panel makes the desktop and menus smoother regardless of in-game frame rate, and it gives the machine somewhere to grow if you play older titles or turn settings down.

Native 2560x1600

16:10 at 2560x1600 on an 18in panel is a sensible specification. The extra vertical space over 16:9 is genuinely useful for reading, writing and any work involving timelines or spreadsheets, and on a panel this large the pixel density is comfortable without pushing scaling into awkward territory.

All the measured frame rates are at 1080p, roughly half the native pixel count. The games with headroom to absorb that increase are obvious from the list: Strange Brigade, Dota 2, Final Fantasy XV and GTA V all have a great deal of room. X-Plane 11 at 122fps will lose less than it looks like it should, because its bottleneck is the processor rather than the pixels. Alan Wake 2 at 116fps is the one where native resolution will cost the most, and where the value of the 175W power limit and 24GB of VRAM becomes real.

The panel technology isn't on record

Our specification data does not record the panel technology for this machine. We know the size, resolution and refresh rate; we do not know whether it is IPS, OLED or something else, and we hold no response time, contrast, brightness or colour gamut measurements for it.

That is a meaningful gap on an 18in machine at this level, because panel choice is one of the few things that genuinely differentiates otherwise similar high-end laptops, and because the difference between a good IPS panel and an OLED one is immediately visible in a way that a 2fps frame rate difference is not. If you are close to buying, this is the specification line worth confirming directly with the retailer for the exact variant in front of you. Panel options frequently differ between regional configurations of the same model, so a general answer about the model line is not a reliable answer about the unit you would receive.

Build, size and portability

At 3.307kg with an 18in screen, this is not a portable computer in any meaningful sense, and it is not trying to be. That figure excludes the power supply, and a machine feeding a 175W GPU alongside a Core Ultra 9 275HX needs a substantial adapter, so realistic travel weight is well above three and a half kilograms.

The footprint matters as much as the mass. An 18in laptop will not fit comfortably on an aircraft tray table, it overhangs desks sized for 15in machines, and a large share of laptop bags will not accept it at all. Moving it is a deliberate act rather than something you do without thinking.

Framed correctly, none of that is a criticism. This is a desktop replacement, and its portability is the ability to move between two fixed locations occasionally rather than to work from anywhere. On that basis the size is an advantage: the chassis volume is what makes 175W viable, the 18in panel provides a genuinely large working surface, and a deck that size has room for full key spacing and a numeric keypad.

The buyers this suits are people replacing a desktop entirely, people in shared or rented accommodation where a tower and separate monitor are impractical, and people who move between two homes or between home and an office and want their whole environment to come with them. For everyone else, 3.307kg in an 18in footprint should be read as a hard filter. If you commute with a laptop, work on trains, or carry a machine across a campus, this is the wrong shape and no amount of specification compensates.

It is worth noting that the two 5070 Ti rivals it draws level with are 16in machines. Whatever else the comparison shows, it shows that the frame rates in these particular games do not require an 18in chassis. If size is a cost to you rather than a benefit, that is a relevant piece of information.

The 90Wh battery

The battery is 90Wh. That is large, though slightly below the effective 100Wh ceiling that keeps a laptop straightforwardly acceptable as aircraft cabin baggage, and it is a little smaller than the largest cells fitted to comparable 18in machines.

We do not hold a measured runtime figure and will not state one. The general dynamic applies as it does across this class: a GPU configured to draw up to 175W will empty a battery of any size quickly when gaming, and most machines in this category also reduce their power limits substantially when unplugged, so unplugged gaming performance is typically well below the mains figures quoted here. Treat this as a mains-powered machine with enough battery to move between rooms and handle some light work, and consult independent runtime measurements if that matters to you.

How it compares

Before the individual comparisons, one point governs all of them. Every comparison here runs over three shared games, and this machine holds exactly three scoring-suite results: Dota 2, GTA V and X-Plane 11. Our comparisons prefer the scoring suite where it overlaps, so those three are almost certainly the games being compared in each case. They are also, as established above, the three least graphics-limited titles in the suite. Keep that in view throughout.

Acer Predator Helios Neo 16S AI: level with this machine, over three shared games

The Helios Neo 16S AI carries an RTX 5070 Ti and has a confirmed UK price of £2,679. Across three shared games, the Strix Scar 18 averages exactly level with it.

Taken at face value that is a startling result: a 5090 at 175W producing the same frame rates as a 5070 Ti. Taken in context, it is close to inevitable. If the shared games are the three CPU-bound scoring titles, then all three are contests the graphics card is barely participating in. Dota 2, GTA V and X-Plane 11 at 1080p are races between processors, and both machines have processors fast enough to run them well.

The correct conclusion is not that the two machines perform identically. It is that on these three games they perform identically, and that a buyer whose library consists of esports titles, older games and simulators would genuinely see no benefit from the more expensive machine. That is a real and useful finding for that specific buyer. It is not a general verdict on the 5090.

What this comparison cannot tell you is what happens in Alan Wake 2, Cyberpunk 2077, or any of the demanding modern releases where a 175W 5090 and a 5070 Ti would be expected to diverge sharply. The overlap does not include them. The £2,679 figure is worth holding onto as the one confirmed price in this comparison set, but without a price for the Strix Scar 18 there is no value comparison to draw from it.

Lenovo Legion 5 Pro: 1fps ahead of this machine, over three shared games

The Legion 5 Pro also carries a 5070 Ti, and averages 1fps ahead of the Strix Scar 18 across three shared games. The analysis is the same as above, with the same caveats, and the 1fps figure is well inside the variation you would expect from repeating a benchmark on the same hardware.

Two independent 5070 Ti machines landing level with this one on the same three games is worth something, though. It rules out the possibility that one of them was an outlier or had an unusually strong configuration. The consistency of the result reinforces the interpretation: these three games do not separate graphics cards at this level, and two different manufacturers' 5070 Ti implementations both reach the same ceiling because the ceiling is set by something other than the GPU.

Our data does not record a power limit for either 5070 Ti machine, so we cannot say how they are configured. It does not matter much for this particular comparison, because in CPU-limited games the GPU power limit is not the operative variable either.

Schenker XMG Neo 16 A25: 37fps ahead of this machine, over three shared games

Schenker's XMG Neo 16 A25 carries an RTX 5090, and averages 37fps ahead of the Strix Scar 18 over the same three shared games. Against a three-game scoring average of 161fps here, that is a substantial margin, and it is the single most informative comparison in the set.

Its importance is partly methodological. When a benchmark reports two consecutive ties against lower-tier machines, a sceptical reader might reasonably suspect the test cannot separate hardware at this level at all. This result disproves that. The same three-game comparison, run the same way at the same resolution, finds a 37fps difference when a difference exists. A level result is therefore a finding rather than a measurement failure.

Its importance is also diagnostic. Both machines carry the same class of graphics card, and the three shared games are predominantly processor-limited. A 37fps gap on that basis is very unlikely to be a GPU story. Our data does not record the A25's processor or its GPU power limit, so we cannot name the cause, and we are not going to guess at one we cannot support. What we can say is that whatever advantage the Schenker holds, it is showing up in exactly the games where the graphics card matters least, which points away from the GPU and towards the rest of the platform.

For a buyer, the practical reading is this: if your games are the CPU-limited kind, the A25 is measurably faster and the gap is not marginal. If your games are the GPU-limited kind, this comparison tells you nothing at all, because none of those games are in the overlap.

Schenker XMG Core 16 VE M25: 64fps behind this machine, over three shared games

The XMG Core 16 VE M25 carries an RTX 5060, and the Strix Scar 18 averages 64fps ahead of it across three shared games.

This is the comparison that shows the money doing something, and it is worth pausing on why the gap opens up here when it did not against the 5070 Ti machines. Even in processor-limited games, there is a floor below which the graphics card starts to matter again. A 5060 has substantially less capability than a 5070 Ti or a 5090, and in games like GTA V and X-Plane 11 at 1080p it is far enough down the stack that it becomes the constraint rather than the processor. Once the GPU is the bottleneck, the tier difference reappears immediately.

That gives a rough map of where the transition sits: somewhere between a 5060 and a 5070 Ti, these three games stop being GPU tests and start being CPU tests. Everything above that point measures the same; everything below it falls away.

Our data has no confirmed UK price for the Schenker either, so this is a performance comparison and not a value one.

What all four comparisons mean together

Read as a set, the four results describe the shape of the data rather than the shape of the machine.

• Two 5070 Ti laptops: level and 1fps ahead. The shared games cannot separate high-end GPUs.
• Another 5090 laptop: 37fps ahead. The comparison method works, and something other than the GPU is responsible.
• A 5060 laptop: 64fps behind. Below a certain tier, the GPU becomes the constraint again and the gap reappears.

The honest summary is that our frame rate record cannot currently demonstrate what a 175W RTX 5090 does in an 18in chassis, because the games available for comparison do not ask it to. That is a limitation of the data, not a criticism of the laptop, and it deserves saying plainly rather than dressing up as a verdict. Anyone using this site's numbers to justify spending flagship money on this machine over a 5070 Ti will not find the justification in these six games. It may well exist in games we do not have results for.

Why there's no score, and what else we don't have

Every scored machine here is ranked on the same five games so that any two laptops can be compared like for like. This configuration holds three of them: Dota 2, GTA V and X-Plane 11, with Cyberpunk 2077 and Baldur's Gate 3 missing. Averaging three and comparing that to other machines' five-game averages would produce a number that looks precise and is not, so the score stays blank.

The missing pair matters more than the count suggests. Cyberpunk 2077 in particular is the most graphics-limited title in the scoring suite, and its absence is exactly why the comparisons above compress. A five-game set including it would very likely separate this machine from the 5070 Ti rivals. We cannot claim that as a result, because it has not been measured, but it is worth understanding why the record looks the way it does.

The other gaps in what we can tell you:

• Price. No confirmed UK figure. Without it there is no value assessment, and value is the largest single component of how we normally judge a machine.
• Native-resolution frame rates. Everything is at 1920x1080; nothing at the panel's 2560x1600.
• Battery life. The cell is 90Wh. There is no measured runtime.
• Noise, temperatures and sustained power. No fan noise, no surface or component temperatures, no evidence about whether the 175W limit holds under long sessions.
• Panel technology and measurements. The technology is not on record, and there are no response time, contrast, brightness or gamut figures.
• Build quality, keyboard and ports. We have not handled the machine and will not describe how it feels.

Who it's for

The Strix Scar 18 makes most sense for a buyer who wants a genuine desktop replacement and knows why. You want an 18in screen because you want the working area and you are not carrying it anywhere. You want 1600p rather than 1080p so that desktop work is pleasant rather than merely tolerable. You want the memory and storage sorted from the start, and 64GB with 2TB does that. And you want a GPU with enough VRAM that texture budgets will not force you into compromises for years, which 24GB at this tier addresses about as well as anything currently available.

Two groups fit particularly well. The first is people replacing a desktop entirely, where the memory configuration means the machine handles the non-gaming work that used to justify a second computer. The second is simulation, strategy and CRPG players, where the processor rather than the graphics card sets the pace. X-Plane 11 at 122fps, GTA V at 162fps and Dota 2 at 200fps are all processor-led results, and they are the strongest evidence in this dataset that the platform underneath the GPU is a fast one.

There is a third group worth naming: people whose use extends beyond games. 24GB of VRAM with 64GB of system memory is a capable configuration for local machine learning, video work and anything else where capacity rather than raw speed is the binding constraint. Those workloads are not measured here, but the specification is well suited to them and it is a legitimate part of the case for the machine.

Who should buy something else

Anyone who moves their laptop. 3.307kg plus a large adapter, in an 18in footprint, punishes portability. Both 5070 Ti rivals in this comparison set are 16in machines that matched it on the games they share.

Anyone whose library is esports, older titles and simulators. This is the clearest actionable finding in the review. On the three games we can compare, a 5070 Ti machine at £2,679 performs identically. If Dota 2, GTA V and X-Plane 11 are representative of what you play, the extra money buys you nothing in frame rate, and you should spend it on a screen, a chassis or a warranty instead.

Anyone who needs the benchmark record to justify the purchase. Six games, three of them in the scoring suite, and no results for the current heavyweight releases. If you buy on measured evidence, this machine's evidence is thin, and that is a legitimate reason to wait for a fuller dataset or choose a machine that has one.

Anyone who needs to know the price to decide. Without a confirmed UK figure we cannot tell you whether this is well-judged or badly judged. Given that the one rival with a confirmed price matched it on the games they share, the price is doing more of the work in this decision than usual. If the Strix Scar 18 is close to £2,679 it becomes an argument about screen size, memory and VRAM, all of which favour it. If it is far above, the frame rate data available will not carry the difference.