When I reviewed the OnePlus 3T, I lightly touched upon its gaming performance, noting that it was similar to what I found on the OnePlus 3. Since then, a few factors have led me to properly test the gaming performance of this device, with the results of my analysis published below.
With 2017 flagships fast approaching, I needed a new baseline to compare and contrast both the new implementations of old chipsets, as we'll find on the LG G6 and HTC U Ultra, and the upcoming devices running new processors, including the Snapdragon 835 and Exynos 9 series. I've written positively about the OnePlus 3T's real-world performance in the past, explaining some of the under-the-hood changes and tweaks that make the result user experience so smooth and sleek. When it comes to gaming, one would expect the OnePlus 3T to lead Android as it packs the most powerful GPU on a phone, at the highest clockspeed we've seen on such GPU, flanked by copious amounts of RAM and running a relatively light-weight ROM. However, a recent original report by XDA further strengthened the idea of the OnePlus 3T as a great phone for mobile gaming, even if that meant making a big mistake in other areas.
When Cheating is Oddly Beneficial
In the interest of keeping the introduction of this review short, I will keep this section brief though I do suggest you check out the full investigation. In short, I found that OnePlus had been cheating on benchmarks through a mechanism introduced into Oxygen OS in the community builds, possibly as a result of the merging of the disparate development teams the company had up until late last year. The OnePlus 3 and OnePlus 3T's stock ROMs (Oxygen 4.0.1) would detect a specific package, which we found in a manifest through a ROM dump, and then configure the processor scaling and thermal throttling threshold to maximize the chance of a high score on any given benchmark. This is different than previous cheating mechanisms by other OEMs, which essentially amounted to setting a 'performance' governor to max out clockspeeds for the duration of the test — OnePlus' cheating mostly minimized variance by effectively having the processors stay at mid-to-high frequencies regardless of the load at hand, which especially bumped less-intensive tasks that didn't call for quick and short bursts of performance.
Furthermore, such mechanism wasn't limited to benchmarks only — some games were also affected and benefit from the performance changes. I've confirmed so in my testing, as some games such as Asphalt 8 do display similar behavior to what I found on benchmarks, even while idle or under minimal load. While the changes made to bump benchmark scores are duly considered cheating, the changes made to the phone's behavior under games does have the benefit of increased framerate, something tangible and useful to the user. As a result, the OnePlus 3T is a better phone as far as gaming is concerned, although problems do arise from such a setup. For one, only games chosen by OnePlus get this added benefit, and this list might become outdated. Moreover, it might result in inconsistent performance, where one game might be underperforming while other similarly-demanding games get the privilege of relaxed throttling and higher clockspeeds on average. Nevertheless, having such boost is a net positive in my view, even if it's not universal — it's not like the device suffers from botched frequency scaling that would necessitate specifically targeting games with a performance mode in the first place.
Suggested Reading: OnePlus 3 and 3T Storage Speed Differences Under F2FS
Finally, one last factor specific to the OnePlus 3 and OnePlus 3T is the adoption of F2FS as a filesystem for users running Nougat. We've detailed what F2FS is, how it is enabled and what difference it makes in a previous article, but for those short of time, the change to F2FS enables the OnePlus 3T to reach higher read/write speeds on its UFS 2.0 storage chip. In turn, this increases the loading time of applications, and the result is most notable in heavy games with long loading times — it can cut the opening speed of games like Asphalt 8 by as much as 50%, and it ultimately puts it ahead of competitors while loading games or levels within games.
A Quick Word on Methodology and Understanding Results
The data for these articles has been gathered using Gamebench, a real-world performance measuring tool. The samples range from 10 to 30 minutes depending on the game and the variations I saw within testing at my discretion. Every run began at an outside temperature of 28°C | 82.4°F at the location of the chipset, from idle, with minimal background services (every other app disabled). Some games have a framerate cap of 30, while others have a framerate cap of 60 — in the case of Asphalt 8, the game is capped at 30 but the menus run at 60FPS. All games were tested at their maximum possible settings for both resolution (1080 x 1920) and textures/effects, looping the first level (or act) of the game. Finally, do note that loading times often introduce huge drops in framerate manifested in the graphs, which do weigh down the resulting average (though not by much) — when you see a drastic momentary drop to a near halt in these graphs, you are most certainly looking at a loading screen.
Asphalt 8 — [AVG: 30FPS]
This game is probably most famous for its graphics, something that almost every reviewer has noted at some point. You likely saw this game being played on phones within YouTube videos, or read it was listed in a "Top 10 Graphics on Android" article — that said, Asphalt 8 is not quite as impressive – nor demanding – as it used to be. It was a few years back when Gamebench itself crowned the Note 4 the king of Android gaming for its ability to nearly-sustain 60FPS on Asphalt 8. My Note 4, months later, could only top 30 as they later introduced a framerate cap (for a while, it wasn't even a universal one). Truth be told, Asphalt 8 is still a good looking game, but most devices with a GPU equivalent to an Adreno 420 or better have no trouble hitting that ceiling. The OnePlus 3 is no exception, but it does surprise in other ways.
The device manages to consistently average 30 frames per second with very, very few dips along the road. The graphics quality of Asphalt 8 has indeed increased slightly over the years, and it does feature a significant number of great-looking levels, some with even more demanding environmental effects such as rain. I played the famous first level for a total of 30 minutes, and found the experience to be as pleasant as it can be. It is true that Asphalt 8 is specifically targeted by OnePlus (in fact, it's the one game I know for OnePlus tests at their lab) for higher framerates, and while the lift of thermal restrictions could easily have turned a more-demanding game into a reminder of why we disliked the Snapdragon 810, I was pleasantly surprised by the thermal consistency and ultimately-decent temperature as well.
Throughout this session, the device hit a peak of 41°C | 105.8°F near the housing of the processor — this is hardly a problem, and while the device starts feeling warm at that point, it doesn't quite reach the 44°C | 111.2°F to 46°C | 114.8°F I experienced with the OnePlus 2, a far worse offender. The sides of the OnePlus 3T did feel slightly hotter than I anticipated, but surprisingly, the overall heat was intangible when I repeated this test with a case, at no loss in performance.
As shown above, the game utilizes a lot of filters, effects and bloom to mask its bland textures and some boxy models; Asphalt 8 is impressive in motion, but looks shoddy and muddy in still frames.The most extensive frame dips never put it under 24 frames per second while in game, and these often came as a result of extreme collisions. GPU utilization never quite reached 90 percent, either, but it ranged from 50 to 85 while the CPU took a back place, hovering between 10 and 35. This will become more telling as we examine the results obtained by other devices.
Overall, though, Asphalt 8 plays beautifully on the OnePlus 3T, though that's not exactly a spectacular endorsement given that this game has no trouble running properly on less-powerful devices (even if not for such long periods of time).
Dead Trigger 2 — [AVG: 55FPS]
This is also a very popular and great-looking game, and a sequel to yet another graphics powerhouse. Built on top of the Unity Engine, Dead Trigger 2 offers a myriad of effects, sharp textures, spectacular lighting for a mobile game, and good looking character and weapon models. While the game itself might be bland, its graphics are definitely quite taxing on smartphones, and while it does offer enhanced graphics for specific devices, the ultra settings on the OnePlus 3T were taxing enough to actually see some frequent dips I didn't find on other titles.
Looping 30 minutes of the first fetch mission after the tutorial, I got to experience some patterns that are telling of when the game runs best, and where the OnePlus 3T fails most. Interestingly enough, the beginning of the level was the most unsavory in every one of the 15-some continuous runs, with the framerate dipping as low as 40 very consistently. I presume this is due to the unnatural amount of bloom and the many reflections on the puddles; going inside the first buildings sees a much better framerate almost immediately. In general, the entire interior area of the level sees great framerate, with dips only happening whenever the many zombies accumulate in areas with various environmental effects.
One area in particular was by far the worst of every run, a long hallway with intense lighting effects and half a dozen enemies. I found that on other phones, merely looking at intense light sources would show a noticeable decrease in frames per second, but the OnePlus 3T generally handled Dead Trigger 2 very well under most conditions. The framerate average sits comfortably at 55, higher than the average of 45 to 50 I find on Snapdragon 820 devices. The short length of each level also means frequent loading screens, which further drag down that average.
However, while playing the game it's quite clear the phone is unable to sustain a solid 60 frames per second. It's not a terrible thing, and a framerate cap of 30 would certainly give the illusion of perfect performance. Temperatures didn't rise past 41°C | 105.8°F, and the phone felt perceptibly hotter than when playing other games on this list. For close to thirty minutes of sustained gameplay, and considering the graphics quality the game offers, the OnePlus 3T ends up performing admirably here.
GTA: San Andreas [AVG: 30FPS]
Grand Theft Auto games have been getting ported to Android for the past few years up until San Andreas; with no likely port of GTA IV anytime soon, this remains one of the best-looking and certainly most expansive sandbox games on Android. The game is chock-full of content and activity, with tons of traffic (set to highest for the purpose of this test), angry pedestrians, and action to be had should you run over the wrong crowd. While the game doesn't look nearly as good as others down this list, the amount of things going on at any given moment and scope certainly contribute to its tax on the processor. It wasn't until the Exynos 7420 on the Note5 – an exceptional phone for the price – that I could reliably achieve the framerate cap when testing for extended periods of time (by comparison, the OnePlus 2 completely failed this test).
The OnePlus 3T does a remarkable job with a sustained average of 30 frames per second. There were some choke-points, namely when activity grew too extreme at a three-stars wanted level. Heavy car collisions that 'dominoed' into multiple vehicles would consistently knock the game out of its framerate cap, and the same went for chained explosions (yet not individual explosions, only those that interacted with the gameworld). Busy areas with long, far-away view distances and many NPCs could also strain the framerate almost as much as the action upon arriving to the area. These framerate drops were always short lived, though, as seen in the graph above. In the end, gameplay was quite smooth and as good as I've ever had it on a modern smartphone. However, some caveats to note are that it is ultimately capped at 30 frames per second, and at maximum resolution it is still "only" running at 1080p.
Nuance aside, the game didn't heat the device past 40.5°C | 104.9°F. Looking at the GPU and CPU load graphs, we find that the game does call the CPU more than other games we tested for this article, with the major spike near the 2:30 mark correlating with a large chained explosion which made itself known in the framerate as well. This is a good port overall, though not without bugs — almost every device I've played it on in 2016 and 2017 have issues with polygons spazzing out accross the screen, and I had a few early crashes before I managed a smooth-sailing run.
Warhammer 40K: Freeblade — [AVG: 59FPS]
It wasn't long ago when we started testing Warhammer 40K: Freeblade for the gaming sections for reviews, but it has quickly turned into one of my favorite games to test. It's visually stunning for a mobile game, though it does make use of a fair few tricks to achieve its graphics fidelity. While environments are well-textured and the game looks quite sharp, there is a very low polygon count on everything but character and NPC models. This is what the game puts front and center, too, and it's no coincidence that both your personal mech and the main enemies take large portions of the screen, with camerawork during fights emphasizing the modelling and texturing of these well-crafted designs.
After going through the first track of the campaign – a rather homogeneous environment located in an urban setting, perfect for testing endurance over time – I received a rather consistent framerate as shown in the graph. The dips of around 5 frames per second most frequently came during the explosive arrival of new enemy hordes and vehicles, with quick stabilization even during the firefights. The smoothest portions of gameplay were actually the one-on-one fights against other mechs, which involve low-intensity quick-time events while locking the player into an impressive fight that showcases every detail, as shown below.
There's not much else to say: the game allows for quite a few graphics settings which I maxed, including lifting the framerate cap from 30 to 60. The resulting framerate average is exceptional, although it wasn't maintained perfectly throughout the session as you can see. That said, I personally perceive the dips in gameplay only when the difference is of 20 frames per second or higher in a short time interval, with jumps from 60 to 30 or lower being particularly jarring. Nothing in my sample suggests the OnePlus 3T has issues with this game, though, and the outside temperature was quite lower than on other titles at just 40°C | 104°F flat after 10-minute sessions.
Riptide GP2 — [AVG: 51FPS]
Last but not least, we have Riptide GP2, a game you might also have seen in video reviews or listicles. The watersport cousin of Asphalt 8 does indeed feature excellent graphics with a healthy amount of granularity under "Advanced Settings", featuring tons of effects and reflections, with substantially less use of cheap tricks to mask its imperfections. The game's physics also do a convincing job at simulating waves and the waterbike's behavior, making for satisfactory gameplay. Detailed models, crisp textures and very complex environments with detail, long distances and many moving objects make this game a graphics crown jewel and a good performance benchmark.
Unfortunately, this is where the OnePlus 3T performed the worst in, although it still managed to output a respectable 51 frames per second average in the end. This is nearly a pixel-perfect average of what I perceived in game, as looping the first level had me fluctuating between 40 and 60 frames per second extremely consistently; every lap had framerate dips at the same specific segments, which didn't offer particularly different visuals than the rest of each level. However, I speculate that what really bogged down performance was the draw distance and the amount of objects on the screen, as after turns revealing large straight sections of the track, fluidity took a hit every time. This is compounded by the amount of large structures with many moving elements found in some of these segments, and by the end of my 30 minute session I had the timing and location of these heavy areas memorized to a T.
I didn't see much of a drop in performance over time, although the specific laggy areas – particularly at the beginning (or end) of a lap – did start dipping below 40 after 15 minutes of gameplay or so. That said, and as shown in the graph, the framerate is still somewhat consistent. Outside temperature of the device managed to reach 41°C | 105.8°F, making it one of the more uncomfortable games, though not by much.
Dash Charge — Play While Charging
Another small advantage worth mentioning is Dash Charge, OnePlus' new charging protocol. Not only is Dash Charge fast, being able to fully charge a device in less than an hour and a half, but it's also different to other charging standards as it does not throttle if you use the phone while charging it. This is specially useful in some real-life scenarios such as airports or even driving (getting more juice out of your GPS-guided commute), but it also applies to games as well. In an in-depth analysis of Dash Charge, we found that playing Asphalt 8 while plugged in, at maximum brightness for 20 minutes, still charged the battery from 14% to 51%. Many OnePlus 3 or 3T owners can vouch for the efficacy of Dash Charge, and while it's not the best perk in the world, it does help with long gaming sessions, or even regular gaming sessions, as the phone will perform as usual, heat up at similar rates (the charger itself does the heavy lifting, reaching up to 50°C | 122°F), and still giving you very similar charging speeds as if you weren't playing the game at all. That said, I'd still be cautious, just in case.
Conclusion
I am very excited to see what OEMs can offer in 2017, and the OnePlus 3T will be the standard to which I'll hold every new device this year. While the 1080p screen on the OnePlus 3T might give it an "unfair edge" against other devices with 1440p displays, many 3D games nowadays offer resolution scaling settings, and some OEM ROMs also allow you to change the resolution of the device. If they don't, a simple adb command does the trick. For this reason, when testing 2017 flagships, we'll analyze performance at both 1080p and, if the game allows it, 1440p as well.
Going back to the OnePlus 3T, the device is a gaming powerhouse indeed. There are some features it offers above competitors that do help mobile gaming in particular, such as the faster loading times introduced by F2FS and the unthrottled charging speeds brought by Dash Charge. It's also true, though, that OnePlus introduced more aggressive processor behavior in certain games — my hope is that they reintroduce the feature with a user interface, like Samsung does, so that users can add their own games to the list of games it targets, or disable the feature altogether should the user not want or need it.
Despite these caveats, it's ultimately a powerful performer that near-maxes out all of these games. It might not be the specific champion in all instances, but it offers a profile I can easily contrast against 2017 flagships. Overall, it's a nice device for gaming that will likely stand the test of time better than its predecessors. I do hope, however, that the prowess on display here and in upcoming devices prompts game makers to further up the ante and deliver even better graphics, as I do think that progress has been stale with few new releases truly impressing those hungry for great visuals on mobile. With smartphones getting more and more powerful with every cycle, it should only be a matter of time before some game studios decide to take full advantage of the improvements in graphics performance and the Vulkan API.
Check out XDA's OnePlus 3T Forums! >>>
from xda-developers http://ift.tt/2l1smID
via IFTTT
Aucun commentaire:
Enregistrer un commentaire