The arrival of Nvidia’s GeForce RTX 30 Series GPUs on laptops, following their smashing success on desktops, promises big performance gains and attainable ray tracing on mobile machines. Based on the same “Ampere” architecture that delivers incredible performance at a fair price in full-size desktop graphics cards, the RTX 30 Series laptops GPUs have come zooming in this week to high expectations.
Having just completed our first round of real-world testing of RTX 30 Series laptops, though, it turns out that things are not quite that straightforward. Through the process of reviewing three different laptops, each with a new GeForce RTX 3070 GPU, we have seen some impressive performance numbers—but also some variations in frame rate and performance that we could not ignore.
Now, one thing is clear from the get-go: The mobile versions of these GPUs are not going to match the power of their full-size desktop counterparts. Chalk that up to their sizes, the greater wattage available to desktop-PC power supplies, and, well, the sheer laws of physics. Some variation in performance from laptop to laptop is also normal, even when they have the same core parts. The chassis and the cooling design affect results. Differences in CPUs and memory configuration and amount, too, come into play.
But in our testing, the gaps between theoretically equal (in name) RTX 3070 laptop GPUs were, in some cases, significant. A few results looked more like they belonged to GPUs of different tiers entirely, as much as 34 frames per second (fps) in one instance, despite the consumer-facing information labeling them as GeForce RTX 3070.
Three RTX 3070 Laptops Walk Into a Bar…
The table coming up below shows the results for these three GeForce RTX 3070-bearing laptops we received in advance of the mobile Ampere launch. They are the following…
We ran each across a range of initial tests. These comprise two of 3DMark’s subtests: Sky Diver (a lower-end test) and Fire Strike (a more gaming-PC focused one). Also in the mix is Unigine’s Superposition, a difficult DirectX 11 rendering trial. We also tested each laptop with the AAA games Rise of the Tomb Raider and Far Cry 5 at 1080p and two settings presets. (See more about how we test laptops.)
You can quickly see the issue: For three GPUs with the same name, there’s quite a range of results, especially where the Asus TUF Dash laptop is concerned. It’s a multi-faceted issue, but the larger part of it ultimately lies with the wattage deployed to each GPU, Nvidia’s Max-Q design branding, and how laptop makers disclose this information.
How Max-Q Has ‘Matured’
To help understand the issue, you first need to understand Max-Q. We have a full explainer of what exactly Max-Q has meant during the past few years, but here’s the quick version.
Nvidia worked with manufacturers to develop Max-Q implementations of top-end GPUs (for example, a Max-Q RTX 2070) to fit inside thinner gaming laptops. These Max-Q editions have had down-tuned power in order to restrict their thermal output, necessitating less cooling hardware and space, thus allowing them to fit in slimmer machines. The compromise is that the limited power, naturally, led to lower frame rates versus the “full” version of any given GPU. (Incidentally, in laptop-maker lingo, the full-strength version of these GPUs deployed in larger machines were often dubbed “Max-P.”)
This was an acceptable solution for years. So long as you knew you were buying a Max-Q version of the GPU, you were presumably willing to see a decrease in power in order to gain a thinner laptop. The different GPUs still roughly fell where they should in the hierarchy, with smaller percentage differences in performance than shown above. The full-power (Max-P) implementations of mobile GPUs over the last few years were generally only seen in thicker (sometimes 17-inch) laptops, and thinner machines saw modest performance reductions, comparatively speaking.
The launch of the laptop GeForce RTX 30-Series, though, has muddied up this status quo. Nvidia isn’t mandating that vendors publicly list whether or not the GPU is tuned down for Max-Q, and the meaning of the Max-Q branding itself is also shifting. Nvidia issued this statement on the topic:
Max-Q branding is not going away.
Nvidia’s original statement on the matter, given to the site Hot Hardware, had a different last line:
Ultimately, like all laptop features and specs, it is up to the OEM to market what their particular laptop configuration supports.
We suspect Nvidia and its OEMs may have caught some heat for that statement, potentially leaving consumers to fend for themselves. (Indeed, we’d be the first to fire things up.)
The Range of Things to Come
Why? In essence, the Max-Q suffix is no longer a quick-glance designation of its capabilities, and you may well not see it on a given laptop’s spec sheet. Instead, Max-Q is more of a technology-platform designation for the new GPUs, encompassing a host of power-saving aspects, and you’ll instead need to look at the detailed specs of each laptop for a gauge of how the GPU is implemented, for an idea of how well it will perform.
That’s a difficult ask for laptop shoppers, who need to dive in to the specs to get a theoretical idea of a laptop’s power. If you’re an enthusiast, you may get a clue of sorts from reading specs, but the less tech-savvy will have difficulty parsing the nuances. Even GPU names alone can be confusing for many shoppers.
On top of that, with the RTX 30 Series, each GPU can be implemented at a wide range of TDP values and varying peak boost clocks, really blurring the lines between GPUs of the same name. See Nvidia’s GeForce RTX 30 Series summary chart below for what I mean…
As you can see, each of these Ampere GPUs has a range of peak boost clock speeds and wattages they can be configured for. If a GeForce RTX 3080 GPU is configured at the lower end of its range, it’s possible (though we have not confirmed this in practice) that an RTX 3070 with higher wattage (and the right combination of supporting parts) could deliver better performance. The same base GPU configured at a different wattage would also result in differing performance levels.
How Power Limits Come Into Play
That’s how we come full circle to the results of our three RTX 3070 GPUs. The chip in the Gigabyte Aero 15 itself is listed as a “Max-Q” GPU in Gigabyte’s system specs, but it clearly outperforms the Asus TUF Dash F15’s RTX 3070, which is not labeled as such. (We queried our marketing contact at Asus, who noted, “Max-Q or Max-P is no longer referenced anymore, as it’s just an RTX 3070 8GB GDDR6.”)
We hope this Wild West of laptop GPU designations doesn’t settle into the new normal. Given that a “Max-Q” flagging (or lack thereof) is no longer a clear indicator of a GPU’s power implementation, consumers could end up at the mercy of OEMs to say exactly what they are selling, and to interpret what is said.
The Alienware m15 R4, meanwhile, had the best results of the bunch. It is not listed as Max-Q, and the company confirmed to me that it’s a “full power” version with additional cooling to support that implementation. It helps that it sports a sizable power brick charger to go with it, able to push more wattage to the system.
You might safely assume the Alienware machine had been set up for a higher power ceiling than the Asus, and we, indeed, assumed that from the outset. But we ran some tests to get some concrete backing for that.
One thing to note is the impact of Nvidia’s Dynamic Boost 2 feature, which comes in with these Ampere GPUs. When boosted, the GPU can effectively “borrow” unused wattage from the rest of the system (say, when the processor isn’t being fully utilized). This can push the total wattage past the TDP ceiling of the GPU on its own. (Note that the TGP stands for Total Graphics Power, and measures the wattage of the chip power plus the graphics memory.)
On the Alienware and Asus systems, we looped 3DMark’s Fire Strike test, and used the utility GPU-Z to log the GPU wattage as the test ran. The Asus TUF Dash F15 peaked at 62 watts TDP and 86 watts TGP, far lower than the Alienware m15 R4’s 107 watts TDP and 141 watts TGP.
If you saw those TDP and frame-rate gaps without knowing which GPUs they were, you could be excused for assuming they’re entirely different GPUs. That illustrates the problem with the lack of clarity around this topic moving forward: Paying for an RTX 3070 set up in a lighter power-consumption state could mean you get something implemented in a way that brings it closer to a RTX 3060. In other words, some RTX 3070s might be equal in name only, and it could be on you to figure out which is which.
Now, let’s not ascribe malice here. Manufacturers may very well account for the lower wattage in the prices of their laptops—they’re not necessarily trying to fool you. They may well—and should—cite clear GPU power level and peak boost numbers. But if you see a laptop with the GeForce RTX 3070 GPU listed as a component and are wondering how its maker is offering that much power at that price, the GPU nitty-gritty may be one big place to look for the nuances.
And…That’s Why Independent Testing Matters
You’ll need to look carefully at the specs when shopping, and ultimately, this will place more value on independent testing results for individual laptops, like those from PC Labs. Previously, testing gaming laptops was less fraught; it was less about testing each GPU implementation to the nth degree (as we knew roughly what to expect from full-power and Max-Q versions). But it looks like laptop GPU evaluation won’t be quite so straightforward from here on out.
That’s our job, though. Hopefully, the situation becomes clearer for consumers after the Ampere launch has had some time to percolate and settle. As I noted above, in reaching out to Nvidia, the company reiterated its statement cited to other outlets, but with that tweaked line that it is “reviewing additional ways to give consumers more insight prior to purchase” and will share those details soon. We’ll keep you apprised of any changes in RTX 30 Series GPUs’ naming and spec conventions, if any come to pass. We hope they do—maximum transparency matters, especially when you’re paying four figures for the laptop of your dreams.
