Keyboard

The Lenovo Legion Y920, like its predecessor, also features a mechanical keyboard—something which has long been hailed the holy grail of gaming peripherals. The difference is immediately noticeable, as every key features an identical level of feedback and exudes a rather loud “click” when depressed—a characteristic common to all mechanical keyboards. Unfortunately, regardless of this undeniably premium aspiration, the result is less than perfect: thanks to an uncommonly light actuation force and keys which are rather wobbly (they literally pivot under the finger until pressed), typing takes some serious adjustment on the Y920 keyboard. Overall, we feel that the vanilla Lenovo AccuType keyboard is fundamentally superior.

The keyboard is RGB backlit and (mostly) customizable. The Nerve Center software package allows for assigning different colors to different keys individually as well as even to the lighting surrounding the touchpad and top speaker grille, but it doesn’t go so far as to allow notification- and event-based lighting changes like (for instance) the Alienware software does. However, it is easy to activate one of a few different preconfigured animation styles with whatever color scheme you choose.

Touchpad

The Synaptics touchpad is fairly large and provides accurate pointer translation with a pleasantly smooth surface finish. The integrated buttons aren’t nearly as pleasant, however; the feedback is a bit mushier than ideal and it’s difficult to tell whether you’re right- or left-clicking. Drag-and-drop is, as usual, a challenge with the clickpad as well. Gestures are accurately and quickly interpreted.

The Y920’s 17.3-inch FHD 1920x1080 display features what Lenovo refers to as an “AntiGlare” finish—though really, it’s just glossy with a little less reflectivity than the average high-gloss edge-to-edge glass display. Regardless, the 127 PPI derived from the panel size and resolution should be enough for satisfactory gaming presentation, and—of equivalent importance—NVIDIA G-Sync technology is on board, which synchronizes display refresh rates with those of the GPU for a smoother gaming experience overall.

Subjectively speaking, the display appears bright with reasonably good color reproduction. It doesn’t have too warm or cool a tint to it, and the contrast levels also seem acceptable.

We measured an average brightness of 378 cd/m² with a significantly higher brightness in the center quadrant of 422.3 cd/m². The rest of the screen is fairly consistent in its luminosity, however, so brightness distribution still isn’t all that bad at 81%. Contrast, as we previously predicted, was measured at 880:1, which is pretty good but not stellar.  Compared with its closest modern competitor, the Asus ROG G752VS, the Legion Y920 handles the basics better (21% brighter, roughly the same contrast, better color accuracy).

The display is not particularly unique in its color coverage, producing (per our measurements) 84% of sRGB and 55% of AdobeRGB. That’s roughly in line with the rest of our comparison field, with only the Legion Y720 coming in behind with a much more disappointing 57% of sRGB.

The Y920 also does a good job with color accuracy out of the box. Our CalMAN 5 measurements suggested DeltaE values of 4.4 and 2.8 for Colorchecker and Greyscale respectively (ideal: 0), and the CCT Average was 6403K—very close to the 6500K ideal value. The Total Gamma reading of 2.22 also isn’t too far off the ideal mark of 2.4.

Outdoors, the Y920 is just fine in shaded areas. Introduce any sunlight, however, and the so-called “AntiGlare” finish that Lenovo refers to quickly succumbs to obstructive reflections. It isn’t nearly as bad as it is in many other notebooks with glossy panels, but it’s certainly not as resistant as true anti-glare panels. Viewing angles, on the other hand, are excellent thanks to the IPS panel.

The Intel Core i7-7820HK’s four cores (plus Hyper-Threading) are clocked at 2.9 – 3.9 GHz (2 cores: 3.7 GHz; 4 cores: 3.5 GHz), so it’s more than capable of handling any gaming endeavor. Plus, it’s unlocked, and Lenovo’s default “Turbo” settings overclock to 4.1 GHz—though (and this is the critical point) that’s provided thermal headroom exists to accommodate it. We’ll examine that second point in a moment, but the bottom line is that the CPU is not likely to age as quickly as the GPU in this machine regardless of choice. Yes, the Core i7-7820HK is technically a bit quicker, and its potential for overclocking renders it more future proof, but even if performance pans out ideally, it’s unlikely that in three or four years the CPU will be the primary bottleneck during gaming sessions on the Legion Y920. By then, it’ll be time to consider planning for a replacement if serious gaming is still on the table. And besides, overclocking the CPU is far less likely to positively affect gaming performance versus overclocking the GPU, and the additional thermal constraints posed by the higher clock rates (as we witness later in our testing) could lead to less consistent performance overall.

We’ll come back to this point in a moment, but first, let’s take a look at CPU performance as compared to competing notebooks. The only other machine in our comparison field today which shares the same processor is the Alienware 17 R4, and it manages higher scores overall than the Lenovo Legion Y920—even when the Y920 is running in its so-called Turbo Mode. Apart from the Cinebench R15 Multi-CPU benchmark (where two sets of data are shown for the Y920, one with Turbo Mode on and the other off), the rest of our benchmarks below are recorded with Turbo Mode off. However, our testing revealed that a performance bump of around 2 to 5 percent on average is possible with Turbo Mode on, so keep that in mind while digesting the data.

So then, why does the Alienware still manage higher scores even when up against the Y920 with Turbo Mode enabled? Because of its more manageable internal temperatures (as we’ll see in just a bit). The Alienware 17 R4 is able to maintain a full 4.0 GHz (1.1 Ghz above base clock rate) for the duration of a CPU stress test, whereas we most commonly witnessed frequencies of the Legion Y920’s CPU hovering in the 3.5 to 3.6 GHz range. The difference is particularly evident in multi-core benchmarks thanks to the additional thermal challenges imposed by such intensive tests.

In spite of its somewhat disappointing CPU capabilities, we’re happy to report that our tests of the Y920’s general system performance using PCMark left us with few concerns in that department. Here, the Y920 dominated the charts with a 5206 in PCMark 10 and equally impressive scores across all PCMark 8 tests which either topped or closely matched all other competitors in our comparison field. As before, the Alienware 17 R4 was once again the primary competitor in most of these tests, but both machines post such impressive results that it’d be splitting hairs to try and meaningfully compare the two.

One major reason for these stellar results is evident as we turn to our storage benchmarks. The Y920’s Samsung SM961 NVMe SSD is one of the current best storage drives around, and its sequential and 4K-64 read/write results of 2768/1589 and 747/1220 MB/s are evidence of that. Our machine also came equipped with a secondary mechanical storage drive—a 1 TB Western Digital 5400 RPM model—but there’s actually still a third bay available as well (another M.2 / NVMe bay). As you might have guessed, the two M.2 slots can be used in conjunction for RAID operation.

CPU performance is adequate, storage performance is excellent… but what about the GPU? Clearly the single most important item in a gaming machine, let’s hope it isn’t subject to the same thermal constraints that plagued the processor. Although faster GTX 1070 notebooks exist, the Legion Y920 turns in a decent performance for its hardware: we recorded a 3DMark Fire Strike Graphics score of 16629 (slightly higher than the Gigabyte P56XT’s 16410, though below the Asus G752VS score of 18346) and a 3DMark 11 Performance GPU score of 21879 (once again besting the Gigabyte’s score of 20576 but succumbing to the Asus G752VS’ 24349). The latter score is roughly 2% below the median for all 40 benchmarks (currently) in our database for the GTX 1070 (score: 22354), so it’s not ideal, but it’s not far off the mark, either. It’s overall around 20 to 30 percent below the performance of the Alienware 17 R4 and its GTX 1080.

Keep in mind that these benchmarks were performed using the laptop’s default settings, meaning that Turbo Mode was off; with it on, users should be able to eke out another percentage point or two of performance according to our tests—but the differences are minor to say the least.  For instance, with Turbo Mode enabled, 3DMark 11 scores jump slightly to 16869 (1.3% higher) and 21933 (0.3% higher) for overall/Performance GPU.  That’s still below the Asus ROG G752VS’ scores of 17222 and 24349, but for the most part (as we see in our actual gaming tests below) performance will likely only be around 3 to 4 percent different between the two models during actual gaming.

In terms of sustained performance, we turn to our Witcher 3 benchmark as always for a real-world estimation of how well the device holds up. Like our earlier Cinebench R15 sustained CPU test, we ran our Witcher benchmark twice—with Turbo Mode both on and off—and recorded and analyzed all the data for both states. To quickly summarize, we found that performance without Turbo Mode enabled was mostly consistent, but we did witness frame drops into the middle and low 50s (frames per second) for probably around a third of our test duration (which is 60 minutes in total). Meanwhile, enabling Turbo Mode basically eradicates the problem, with only some extremely minor and infrequent fluctuation down just a couple of frames per second. The bottom line is that there’s really no good reason not to have Turbo Mode enabled during gaming sessions unless the noise level simply bothers you too much. We didn’t even notice any significant difference in case surface temperatures between the two scenarios; CPU temperatures reached 93°C on average, while GPU temperatures climbed to a consistent 73 °C.

We stress test machines using a combination of Prime95 and FurMark to discern their theoretical maximum potential under the heaviest synthetic loads. For most machines, this may be merely an exercise in superfluity, but for gaming devices and workstations, it can provide valuable insight as to the top-level capabilities of a machine that could eventually encounter similar conditions—however rarely.

As we alluded to earlier, our stress tests of the Legion Y920 were particularly illuminating. Before we get started, it’s important to explain how we approached these tests. We performed each exercise with Turbo Mode off, allowed the machine to cool, and then repeated it a second time with Turbo Mode on. We recorded the results in each scenario separately to allow for comparison and perhaps a better idea of how well the Turbo Mode feature actually works.

First, under CPU stress using Prime95, we witnessed an initial clock rate of 3.5 GHz until around 30 seconds into stress, at which point it drops to around 3.1 – 3.2 GHz (after reaching a max of 82 degrees). We then switched Turbo Mode on and repeated the test. The results were similar, but the clock rates were (predictably) just a bit higher: initially 3.7 GHz and eventually down to 3.5 – 3.6 GHz (after reaching 95 degrees). This result is significant because it explains why we never see performance matching that of the Alienware 17 R4 even with Turbo Mode enabled: the Alienware is able to maintain a permanent turbo clock rate of 4.0 GHz, a full 1.1 GHz above the base frequency of the CPU and roughly 400 – 500 MHz faster than the Legion Y920 on average (roughly 10 – 12 percent faster, which is exactly what we found in our Cinebench R15 benchmarks).

GPU performance thankfully isn’t a major issue on the Y920, but stress testing still provided some interesting data points. Under FurMark synthetic GPU stress, initially, the GPU frequency hovers around 1230 MHz on average with a temperature of 68 °C. With Turbo Mode enabled, the frequency increases to roughly 1280 MHz and the average temperature is slightly higher at 69 °C. As our earlier benchmarks showed, GPU performance is adequate overall, so we’re really more concerned about consistency and GPU temperatures—both areas of proficiency for the Y920.

Combined CPU and GPU stress, meanwhile, saw GPU frequencies fluctuating between 1290 MHz and 1650 MHz and CPU frequencies confined to the 3.0 – 3.1 GHz range. The maximum GPU temperature we recorded was 70 °C (average: 69 °C) and the maximum CPU temperature was a very hot 95 °C (average: 94 °C). Switching Turbo Mode on slightly increases nearly all of these values, with GPU frequencies reaching 1300 – 1750 MHz (and temperatures remaining the same) and CPU frequencies notching upward to 3.1 – 3.2 GHz. CPU temperatures in any situation are far more bothersome to the Y920 than GPU temperatures, the latter of which never see anything above the 70s C.

In summation, enabling Turbo Mode increases both possible top-end clock rates and temperatures/fan speeds. Thermal tolerances/limits appear to remain the same though temperatures are obviously affected. The performance difference is marginal at best, but as we explained earlier during our sustained performance testing of both the CPU and GPU, there really is no good reason not to enable it unless noise levels are a prime concern.

The Y920’s speakers are a pleasant surprise—though when coming from a nearly 10 lb. notebook, perhaps it’s not as much of a shock. You’ll find two 2 W speakers accompanied by a 3 W subwoofer, and that results in better bass reproduction and—happily—relatively balanced mids and highs according to our measurements. Volume levels are not particularly loud (69.67 dB on average), though some adjustment/tweaking is possible via the Dolby Atmos post-processing software.

Speaking of which, this post-processing can be automatically invoked (in the background) via the Lenovo Nerve Center application when games are detected in the foreground. This feature is enabled by default, but in case the post-processing effects prove irritating to the user, disabling it is as easy as launching Nerve Center and switching the Sound Enhancement feature off.

It shouldn’t come as much of a surprise that the Legion Y920 isn’t meant for long runtimes while operating unplugged, but then again, that really isn’t what most gamers care about. Under load, the laptop lasted for just over two hours, while web surfing at around 150 cd/m² lasted for 4 hours and 43 minutes total. That’s not terrible for a high-end gaming PC, but again, the likelihood of this thing making frequent trips out of the house is not very high given its unwieldy size and weight and its need for AC power for true gaming performance. Compared with its peers, this is right around what we’d expect to see from a 90 Wh battery paired with these components.

Please also note the odd change in slope in the Classic Test graph. Because of this, we actually ran this battery test twice, but the result was exactly the same both times, so it ought to be accurate regardless of the cause.