NVIDIA's RTX 40 series notebook graphics have improved significantly thanks to its energy-efficient Ada architecture. The flagship RTX 4090 delivers almost 50% higher gaming performance than its predecessor, the top-of-the-line RTX 3080 Ti, while consuming almost 50% less power for the same performance.
Soon, the RTX 4070/4060/4050 notebook will be available, but it doesn't seem to be quite the same.
According to one manufacturer,The RTX 4070 3DMark Time Spy and Fire Strike are only 11-15% better than the RTX 3070, which is obviously not enough in the face of more than 50% increase in RTX 4090.
The RTX 4060 is slightly better, with a 25-30% improvement over the RTX 3060, but it's not too exciting.
This is just a theoretical run, but in actual games, it will be scaled down by experience.
In terms of specs, the RTX 4070 integrates 4,608 CUDA cores with 128-bit video memory, which is 10% less than the RTX 3070 and halved from 448GB/s to 256GB/s.
The RTX 4060 3072 CUDA cores are 128-bit wide, 20% less than the RTX 3060 and one-third less than the RTX 3060.
Although the frequency of the two new products has been improved a lot, and the architecture has made great progress, the loss of core number and bit width is really not small.