Graphics Performance and Thermal Throttling
Gaming performance is obviously one of the biggest marketing points of the ROG Zephyrus, so we decided to put it through our normal workload of GPU benchmarks.
In these tests, we are comparing the Max-Q GTX 1080 found in the Zephyrus with its GTX 1080 and 1070 desktop counterparts to validate the claim of "GTX 1080-level performance" in these smaller form factor machines.
We are using our Frame Rating testing methodology for these tests, just as we do for all of our GPU reviews. For more information on the methodology, you can consult this article explaining the entire process.
We are performing all of our tests at 1080p, the native resolution of the Zephyrus's display.
Dirt Rally
Our Dirt Rally results at Ultra settings puts the Zephyrus's performnace between the desktop GTX 1080 and the GTX 1070, with the 50th percentile FPS trending more towards the GTX 1070. Additionally, we find higher frame variance with the Max-Q GPU though they are under 2ms at the 95th percentile indicating a smooth gaming experience.
Grand Theft Auto V
In GTA V, we see an interesting scenario play out. At 1080P with Very High settings, the GTX 1080 and 1070 perform almost identically on our GPU testbed. However, when we look at the Max-Q result compared to these two, it is substantially lower. This is likely due to the differences in CPU between our desktop PC and the Zephyrus. The mobile i7-7700HQ struggles with GTA V's demands.
Hitman (2016)
Our first DX12 game in these results, Hitman at Ultra settings shows the GTX 1080 in the Zephyrus to be closer to the desktop GTX 1070 than the 1080. Frame time variance also sees an almost 2ms increase when moving to the Max-Q GPU in the Zephryus though it is still in a reasonable range.
Rise of the Tomb Raider
When playing Rise of the Tomb Raider in DX12 mode and ultra settings, we see the Max-Q 1080 performing much closer to the desktop GTX 1080 in average FPS. However, the 95th percentile frame variance is about 3-4x higher than the desktop GPUs. Hittin the 5ms+ mark indicates more judder than we would like to see.
The Witcher 3
Once again, The Witcher 3 at Ultra settings show the Zephyrus to perform much closer to a desktop GTX 1070 than 1080. However, frame variance is more in line with the desktop GPUs than in most of the other games.
Overall, the GTX 1080 with Max-Q design seems to perform closer to a desktop GTX 1070 than a GTX 1080. This makes the GTX 1080 branding at best meaningless and at worst disingenuous to the actual performance of the GPU. But this has been the case for mobile variants of graphics hardware for many generations. NVIDIA claimed to eradicate that mentality with the GTX 10-series mobile product launch, but the thermal and power limits of these form factors are clearly having an impact.
Also, as we would expect with a thermally limited design, the frame variance is higher in the Zephyrus than we would expect from this same silicon in its larger desktop counterparts.
It's worth noting that from an experiential standpoint, I feel that all of these games were in a playable state. While you might have some occasional frame stutter, overall these games look beautiful and smooth on the 120Hz G-Sync display in the Zephyrus.
Thermal Limits
To further test the thermal capabilities of the Zephyrus, we decided to run a more long-term gaming test. While most of our GPU benchmarks take place for just 60 seconds, we decided to keep a game running for a longer period and compare the benchmark results of the beginning and end runs.
For this expanded testing, we turned to Dirt Rally's benchmark mode. This mode is very close to real world performance while racing, just with an AI controlled driver instead of user input. Additionally, a loop setting can be enabled, allowing us to walk away so the notebook gets nice and toasty in a consistent manner.
We ran Dirt Rally at the same 1080p Ultra settings as above and gathered our normal 60-second benchmark run. Then, we continued to let the same loop in benchmark mode for an additional 45 minutes and then did another benchmark run.
The results are impressive. Despite gaming for a solid 45 minutes, the ASUS ROG Zephyrus performed nearly identically in the two benchmark results. This speaks well to the consistency of the thermal experience with the Zephyrus. Gamers shouldn't worry about their game slowing down as they engage in extended gameplay sessions.
Nice piece no doubt. Battery
Nice piece no doubt. Battery life is a killer though. More disabling tech needed to be incorporated when on battery (i.e. on chip iGPU switch over with disabled dGPU, lower multipliers, cores disabled, lower DDR4 speeds etc.). <2hrs on battery is unusable in the real world.
Is the RAM running in dual
Is the RAM running in dual channel? With on board chips and then two DIMMs, I thought this non-symmetric design might kick it back to single channel?
Intel chips have been smart
Intel chips have been smart enough to do a split dual/single channel partition of the RAM for a little while now. In most scenarios you won’t notice the difference
What the hell is the point of
What the hell is the point of an ultraportable that lasts less than two hours? Seriously who approves such a product?
Maybe I’m in the minority
Maybe I’m in the minority here, but I’d be more interested in a push for more durable, reliable, and serviceable laptops rather than this focus on thinness.
This is honestly a stupid way
This is honestly a stupid way to go with laptop design. If I’m buying a laptop for gaming, I don’t want thin and light to compromise battery life, thermals, noise, throttling and all sorts of other criteria. I /want/ a thicker laptop that has a larger battery, better cooling and temperatures. I’m all for Nvidia GPUs, but this idea is complete non-sense and a waste of time.
So it’s a 1080m…
So it’s a 1080m…
“You’ll find four USB 3.1
“You’ll find four USB 3.1 Type-A ports”
should be: You’ll find four USB 3.1 Gen 1 Type-A ports. But the spec sheet table lists them as “4 x USB 3.0″(old USB-IF nomenclature) and at least there is a TB3 so that’s not so bad.
Really a gaming laptop and Ultrabook/Thin And Light means only one thing: thermal throttling and crappy battery life. And the deal killer is that Thin/Apple like form over functionality design for a gaming laptop that costs $2700.
Why does it have a
Why does it have a significantly smaller battery than my Surface Book? Two hours is a joke. I had a Clevo gaming laptop that wasn’t much thicker than this thing back in 2005 that had three hours of battery life…
I think the compromises here are in all the wrong places.
Needs a 90whr battery.
In a
Needs a 90whr battery.
In a world where external GPUs are looming I don’t think the form factor makes sense.
Having a single machine to do it all is a good goal. But, Nvidia and AMD should OEM eGPUs in a small, sleek form factor. No need for an enclosure, if the GPU itself is designed to sit on the desk with an integrated 120v PSU on the PCB. I don’t think an eGPU has to be much bigger than a large conventional GPU if it’s done right.
I think this laptop is
I think this laptop is amazing.
Yes….the battery life let’s it down, but all Asus has to do is design and manufacture an external power-bank to attach to the Zephyrus, thus making it more portable.