Over the past two generations, Nvidia has taught us that mobile and desktop graphics cards of the same name are quite similar. They often had the same number of CUDA cores, and the differences were primarily in frequencies and power limits.
The generation of Ampere desktop cards has become much more powerful than the Turing cards, but a significant increase in power consumption has played a significant role. For example, the RTX 3070 is about 50-60% faster than the RTX 2070 in 4K, much faster. But energy consumption increased by 20%. Yes, the new cards’ energy efficiency has improved significantly, but absolute terms’ power consumption has increased significantly.
Of course, this became a stumbling block in the development of mobile RTX 30 adapters. It is impossible to push the same desktop GPUs into mobile power limits without a catastrophic reduction in frequencies, so Nvidia cut many mobile card parameters. At least this is true for existing top models, including the RTX 3070.
Just compare, the desktop RTX 3070 has 5888 CUDA cores, while the mobile has 5120. That is, the first in this matter is 15% better than the second. Also, the frequencies are noticeably reduced, but the bus and memory size remains the same.
So how much faster is a desktop card than a mobile card? By the same 15%? In fact, if you understand a little on this issue, you know that modern Nvidia mobile video cards within the same model can be very different due to power limits. For the RTX 3070, the TDP range is from 80 to 125 W, and there are cards with a higher limit. Of course, performance will also greatly depend on this.
The TechSpot specialists made a very cool comparison, which we will now analyze. There is a desktop RTX 3070, a couple of mobile RTX 3070s with different TDPs, and many other Nvidia adapters for a more interesting comparison.
On the source site, you can see a huge number of diagrams. Here I will give the most informative and interesting ones, and then we will understand.
So what do we get? Well, first of all, the mobile RTX 3070 lags very far behind the desktop anyway. The source tested the 80-watt and 115-watt RTX 3070. In the first case, the desktop adapter is faster than its mobile counterpart by an average of 50-60%, and in the second – by 30-35%. This is a performance gap, and when you buy a laptop with an RTX 3070, you are not getting the same thing as a desktop PC with the same graphics card, although the names are the same.
Between the two versions of the mobile RTX 3070, the difference is about 15-20%, and in some games, it can reach 30%. This is also a lot, but still much less than 50-60%.
And now, briefly, several important conclusions.
- 1. The average difference between 80-watt mobile RTX 3070 and RTX 3060 is only 10%
- 2.80-watt RTX 3070 is approximately equal to 115-watt RTX 3060. That is, in laptops, sometimes it makes no sense to overpay for a card at a higher level
- 3.80-watt RTX 3070 is 10-15% faster than RTX 2070 Super Max-Q
- 4.15W RTX 3070 30% faster than 115W RTX 2070
- 5. The desktop RTX 2070 Super is only 10% faster than the 115W RTX 3070.
All in all, the mobile RTX 3070 is definitely a very cool card. It is much faster than its predecessor, it is very productive in absolute terms, and at least in Full HD, it will provide you with 60-80 fps even in the coolest games at maximum speed, even in the 80-watt incarnation. Cyberpunk 2077, monstrous in its requirements, produces 55-65 fps at Ultra, which is cool.
Simultaneously, the card lags behind the desktop sister by a whopping 30-60%. These are simply different gaming classes, and you need to understand this when choosing and buying a laptop. It is advisable not to take models with 4K screens since the mobile RTX 3070 is clearly not for this resolution.
Let me remind you that Nvidia recently put pressure on laptop manufacturers. They are now obliged to indicate Ampere mobile adapters’ characteristics on their websites, which means that you can easily choose the version of RTX 3070 that you need: more powerful or more economical.
A source: TechSpot