ASUS ROG Xbox Ally X Review: Excellent But Evolutionary Handheld PC Gaming

Before we get into the direct handheld comparisons, let's take a look at a couple of tried-and-true gaming benchmarks to see how this system compares to full-sized laptops.

F1 2022 Formula One Racing Simulation Benchmarks

Granted that racing simulator F1 2022 is getting a bit long in the tooth, but it can be surprisingly sensitive to certain system configurations, so we enjoy using it as a test as it can really tease out some peculiar performance characteristics. It helps that we have lots of performance data for it too. We tested the game on its High graphics preset at 1080p resolution with upscaling disabled.


In F1 22, the ROG Xbox Ally X takes our top slot, at least among integrated GPUs. It's not far ahead of the Legion Go S with SteamOS, though—and that machine consistently puts up a better 1% low frame rate.

Gears Tactics Unreal Engine Strategy Benchmarks

Similarly, 2020's Gears Tactics still makes an excellent benchmark for integrated graphics. It's a tightly-optimized Unreal Engine title with a highly-configurable built-in benchmark that provides a wealth of performance data after each run. We tested it on High, Medium, and Low presets at 1920×1080 resolution, just as we have done for the other systems.


For whatever reason, this benchmark loves Intel's Arc GPUs. It consistently performs well on them, especially at lower detail settings. However, the ROG Xbox Ally X certainly does well here, outpacing every other AMD system—even if only just so.

Alright, enough of all that. Here's a look at how the ASUS ROG Xbox Ally X compares against the previous Ally X, the Lenovo Legion Go S, the Steam Deck LCD, MSI's Claw 8 AI+, and a desktop version of the Phoenix silicon in the form of the Ryzen 7 8700G.

Direct Benchmark Comparisons Against Other Handhelds

First up is Armored Core VI: Fires of Rubicon. This explosive mecha action title is surprisingly lightweight considering the level of detail in the absolutely expansive environments. The ROG Xbox Ally X comes in ahead of all of the other handhelds by a nose. Ultimately all three of the Ryzen Z machines are producing essentially the same performance, while the desktop 8700G—using the same chip as the Ryzen Z1 Extreme—dominates.


In cute and colorful indie Souls-like Asterigos, we see the ASUS ROG Xbox Ally X take a front-runner position, but again, only by fractions of a frame. This benchmark can be pretty variable, so really everything here is basically margin of error, at least as far as average FPS goes. Check out the Legion Go S offering the highest 1% low FPS--that's the power of SteamOS. That's kind of a trend.


Yet again, we see the Ryzen Z2 Extreme providing essentially the same performance as the Ryzen Z1 Extreme. Zen 5 doesn't have much uplift over Zen 4 in games, and neither does RDNA 3.5. While the Ryzen Z2 Extreme has a larger GPU, it's still bound by the 35W power limit, and given that they're on the same fabrication process, Phoenix and Strix Point run about the same at 35W.


That trend more or less continues in Counter-Strike 2, although the Legion Go S pulls ahead given that this is Valve's game running on Valve's OS. The ROG Xbox Ally X puts up an unusually poor 1% low framerate here; it's not clear what the problem was but it was consistent throughout the whole test run.


Cyberpunk 2077 is demanding enough and optimized to the point that the superior architectures and stronger silicon of the Ryzen AI Z2 Extreme can flex their muscles. The performance of the newer chip is creeping up close to the power-thirsty desktop chip, which is impressive for the same reasons we just disclosed.


We don't have MSI Claw data on this game because it came out after we had to send our Claw back to MSI. However, the data we do have is pretty interesting. Elden Ring Nightreign definitely prefers AMD's newer chip, with consistently better 1% low and average frame rates versus the previous ROG Ally X. The difference isn't large, but it is real.


F1 24 uses restrictive anti-cheat that means it simply will not launch on Linux. That's why our Steam Deck and Legion Go S produced 0 FPS in this game. However, our Windows handhelds all run it very nicely at these settings, with the ROG Xbox Ally X pulling away from the pack slightly.


Same sad anti-cheat story here as with F1 24. GTA V Enhanced Edition adds a bundle of ray-tracing effects to the 12 year old game, and we've got some of them enabled here thanks to the use of the "High RT" preset. The MSI Claw takes first place thanks to the strong ray-tracing chops of the Xe2 architecture, while the ASUS ROG Xbox Ally X actually takes the last place position. It's not clear why this is the case, but we tested it several times, and also re-tested the original ROG Ally X in this game and it's consistent. Potentially there's been a regression with RDNA 3.5 here.


In the Unreal Engine 5 remaster of The Elder Scrolls IV: Oblivion, the ASUS ROG Xbox Ally X easily takes pole position among the handhelds, although it still can't topple the desktop version of Phoenix. Take careful note that this isn't averages and 1% lows like usual, but rather indoor and outdoor frame rates. The difference between this and the previous-generation parts isn't huge, yet it's enough to make this machine feel notably more fluid in this game than the other systems. You could probably get away with enabling Frame Generation here.


In indie cute 'em up Onirism—which hits 1.0 in just over a week, by the way—the ROG Xbox Ally X runs out ahead of the Z1 machines again, this time beating even the Ryzen 7 8700G, although the MSI Claw still holds first place. It's potentially the case that the Claw might perform differently if we could re-test it, because Onirism has had a couple of major technology update patches since April.


Quake II RTX at these settings is a remarkably smooth experience on the ASUS ROG Xbox Ally X. While the MSI Claw put up a higher average, the crappy 1% low score there means that the ROG Xbox Ally X and the desktop 8700G both 'feel' smoother in gameplay. This is a pretty great way to experience Quake II, all told.


Running our backup of the original Red Dead Redemption in the RPCS3 emulator, the ASUS ROG Xbox Ally X performs better than the real PlayStation 3. The Amarillo benchmark is one of the most demanding parts of the game; you can almost assuredly play the whole game like this just fine. Notably it runs considerably better at the very same settings on the Legion Go S; we think this is probably down to Linux scheduling being a bit smarter. Of course, nothing beats shoving 100-plus watts through laptop silicon.


In Street Fighter 6 World Tour, the ROG Xbox Ally X performs admirably, giving us the highest 1% low frame rate of the handhelds. Yet again, it's not a commanding lead, but a lead is a lead.


Finally, in predatory gachapon brawler Zenless Zone Zero, the 1% low FPS creeps over 30 for the ROG Xbox Ally X, making the game feel notably smoother despite the small difference in average frame rate. Similar to Onirism above, the game has had significant updates since we tested the MSI Claw 8, so it's possible that performance might not be the same now.

ASUS ROG Xbox Ally X Subjective Gaming Impressions

To tell the truth, we didn't have a lot of time with this handheld. Despite that, we spent every possible waking minute gaming on the ASUS ROG Xbox Ally X. Our conclusion: it rocks. It doesn't rock a whole lot harder than the original Ally X, as evinced in our benchmarks above, but it absolutely is better.


Integrated graphics aren't like they were. Many, perhaps even most HotHardware readers will recall the bad old days when gaming on integrated graphics was an exercise in futility when it wasn't simply outright impossible. Older integrated GPUs like the Intel GMA series were simply not capable of running the latest games. Meanwhile, you can play stuff like Dying Light: The Beast and yes, even Borderlands 4 on this machine—without even cranking the settings to the floor.


If you saw the specifications for the Ryzen Z2 Extreme and were hoping for a big performance boost over the Ryzen Z1 Extreme, that's understandable. Zen 5 versus Zen 4, a 33% larger GPU, faster memory; it should be a knockout, right? The reality is that both chips are fabbed on TSMC N4. That means they have the similar voltage/frequency curve. Architectural differences matter, but ultimately, on a given process, X power means Y performance. This machine is still capped at a maximum of 35W for sustained power limit, and that means it's basically just a little faster than the original ROG Ally X.


Speed isn't everything, though. The gaming experience on this system is great and taken in a vacuum, it's outstanding. You can play basically anything, and if you play a lightweight game like Hollow Knight: Silksong, the 80-Whr battery means you're looking at easily six or more hours of gameplay, even at 120 FPS. Exercise some restraint and enable a framerate limiter, and you might just hit ten hours.


"But Zak, what about that revised Xbox Full Screen Experience," we hear you asking. Well, there's a problem with testing that, and it's because you can't run background applications while using it. That's sort of the point, really, but it does mean that we can't use our normal frame-time capture tools in that mode. We did get it to work exactly one time, with Zenless Zone Zero, but every other time, CapFrameX simply refused to capture the application. Fortunately, some games still have built-in benchmarks, and those work just fine. Check it out:


These benchmark results compare our standard data from our tests above against games running in the Xbox Full Screen mode. The difference wasn't large, but it still surprised us. In absolute terms, the changes aren't huge, but all three games saw an improvement in minimum frame rate, and both Cyberpunk 2077 and  F1 24 saw a little bump in average FPS, too. We didn't expect this kind of change, and we're going to have to spend more time investigating this. As it stands, though, this is a start toward catching up to SteamOS for gaming performance.

Let's put this baby to bed with some thermal, acoustic, and battery data.