Conflict background

NVIDIA Pascal graphics cards in 2016 practically erased the difference between regular and mobile graphics cards.

Several generations ago, NVIDIA promoted the concept of mobile graphics cards with the Max Q prefix, accustoming gamers to the fact that the mobile version of the graphics card was approximately on par with its older brother, and the overall difference in speed was 5-8%. This was especially noticeable with the Pascal and Turing series of GPUs, which were released between 2016 and 2021. Then literally every second review of the conventional GTX 1070 or RTX 2060 began with the thesis “modern player does not need to choose between compactness and performance.”

However, in 2021, with the release of the RTX 30XX line, codenamed Ampere, Mr. Jensen Huang, in his constant leather jacket, confused all the cards for players, causing storm of indignation. It turned out to be completely normal situation in which potential player studies reviews of graphics card he likes and goes to the store to buy the appropriate laptop, only to later be horrified at how slow, buzzing and throttling everything is.

Below we will take look at the main root causes of this phenomenon and figure out how to identify potential pig in poke. Considering that Radeon products are practically not represented on the mobile market, most of the stones will fall into NVIDIA's garden.

Problem one: different chips, different characteristics

GeForce Ampere graphics cards broke everything.

Starting with the GeForce Ampere line, NVIDIA engineers resorted to trick, using different versions of chips for mobile and standard GPUs. At one time, many journalists were shocked by the fact that the laptop version of the GeForce RTX 3080 is not based on the usual GA102 chip as in the desktop GeForce RTX 3080, but on junior GA104 processor from the GeForce RTX 3070 Ti graphics card.

Naturally, no one warned about this in the promotional materials, so over time, Reddit accumulated huge number of posts from dissatisfied gamers who complained that laptop with an RTX 3080 runs on average 50 - 70% slower than regular PC with the same graphics card. There is no need to talk about any kind of brotherhood; at best, these are distant relatives.

With the release of the next generation of accelerators from NVIDIA, codenamed Ada Lovelace, the situation only got worse. In fact, only the younger fourth-generation models are identical, plus or minus, to their older brothers. For example, the mobile and standard versions of the GeForce RTX 4060 have lot in common: both are based on the AD107 GPU, have 3072 CUDA cores and 128-bit memory bus and 8 GB of memory, resulting in both accelerators demonstrating almost identical performance in games.

And here is where all the confusion begins. For example, both variations of the top-end GeForce RTX 4090 are built on the same chip, but the mobile version has 8 GB less video memory and lower clock speeds, which is why its final performance is on par with the junior desktop model GeForce RTX 4080.

The desktop GeForce RTX 4080 uses the AD103 GPU with 16GB of VRAM, while the mobile version uses the less powerful AD 104 with 12GB of VRAM. In fact, it would be more correct to call the RTX 4080 Mobile 4070 Ti and sell it cheaper, since its performance is 30 percent lower.

The situation with the GeForce RTX 4070 is even worse - the standard version is built on more powerful core (AD104 versus AD106), it has more video memory (12 GB versus 8 GB) and more powerful memory bus (192 bits versus 128 bits), which results in 50 percent productivity gap. This is too big difference even with the change of generations; in fact, we are talking about GPUs of different levels, which should be called differently.

Popular Gaming Laptops

Problem two: graphics card power consumption limit

The uninteresting indicator of energy consumption now plays critical role.

Small problem: there are 2 laptops for sale - Acer Nitro 5 and Gigabyte G5 KF. The first one has GeFroce RTX 4050 graphics card, the second one has the older RTX 4060. At the same time, the Gigabyte G5 KF is even little cheaper. Which one will you take? It would seem that the answer is obvious, take more affordable and faster laptop and enjoy good deal, right? Sorry, but no. In reality, the graphics card in the Gigabyte G5 KF is even slower due to the fact that the manufacturer has limited its TDP to modest 75 W, while the RTX 4050 graphics card installed in the Nitro 5 has TDP of up to 115 W.

The technical parameter TDP, which no one paid attention to several years ago, has begun to play decisive role with the help of NVIDIA. So, TDP (Thermal Design Power) is an indicator measured in watts that means how much energy graphics card can “eat” and how much heat it generates at maximum load. Due to the constant miniaturization of laptops, manufacturers have essentially had to choose between power, portability and runtime.

And as it turns out, many laptop manufacturers can limit TDP for variety of reasons. For example, cooling system that worked perfectly in tandem with one processor and graphics card may not cope with another combination of hardware, leading to high heating, noise and throttling. Other manufacturers are confused by the fact that powerful graphics card instantly eats up the battery power and instead of the stated 10 hours of browsing, you get, at best, an hour and half of playing Doom Ethernal. Well, someone might want to make the world’s thinnest gaming laptop and stuff conventional RTX 4090 into it. Just so you understand, the regular version of this card weighs 2.5 kg, occupies more than 3 slots in the motherboard and does not fit into every case. But songs are composed to the madness of the brave, you give the abolition of the laws of physics!

Perhaps the most egregious example is the 2021 mobile RTX 3080, which can have TDP difference of up to two times from laptop to laptop. Needless to say that the full-fledged version, which according to its status is supposed to consume 150 W, will be much faster than its counterpart with the limit cut to 75 - 90 W? With the massive release of mobile graphics cards RTX 40XX Ada Lovelace, the situation has only gotten worse and now literally every review of gaming laptop is devoted to the features of the power system and TDP of the graphics card. This is especially acute for younger models like the RTX 4050/4060 with TDP range from 45 to 115 W. Just think about these numbers.

How to avoid getting scammed?

ROG Zephyrus Duo 15 SE has dynamic TDP that adjusts to the turbo boost frequency.

And it’s not fun at all because laptop manufacturers are not required to indicate such parameters in the specifications, so before purchasing you will have to look for this data on the official website, and if it is not there, rely on geek reviewers. It is obvious that marketing has defeated common sense, and people who unsuspectingly compare these graphics cards are actually being deceived.

Let's say you found an interesting laptop at discount with conditional GeForce RTX 3060 on board. You are aware that the RTX 3060 is strong and modern mid-level graphics card, ideal for Full HD gaming. But how can you understand what is hidden behind this sign in the case of the mobile version? First, you need to compare the model of the graphics chip and the technical parameters of both cards. Many manufacturers do not indicate the chip version, so you will have to Google this information.

As for the characteristics, they can be compared on GPUCheck or any similar service for comparing graphics cards. If you are satisfied with the results you see, then you should find out the TDP of the selected accelerator in the laptop you like and compare it with the original TDP of the mobile version. In most cases, this information is available in product cards on the e-catalog or the official website. If the manufacturer is silent about this information, you will have to Google reviews yourself.