If you have been thinking about when gaming laptops will begin to push to 1440p panels more seriously, then this week’s CES Demystifying Nvidia is aimed directly at you. Behold: The company’s laptop GPU has spanned a generation, this time with the Ampere architecture and the RTX 3000 series brand.
A total of three GPU models were announced, named after GeForce RTX 3080, 3070, and 3060. They are scheduled to be launched in the “70+” laptop model starting on January 26. Nvidia has listed the “top OEM” Acer, Alienware, Asus, Gigabyte, HP, Lenovo, MSI and Razer, as well as the upcoming RTX 3000 series of laptops, as well as “local OEM and system manufacturers.”
Double check naming convention
Nvidia’s sales strategy positions the RTX 3060 laptop model as “faster than laptops equipped with RTX 2080 Super”, although this model may be more specifically suitable for 1080p systems. These two high-end models are often referred to as part of the 1440p system, and this resolution has been left in the cold for gaming laptops for a long time (it can be said that it will benefit from Nvidia’s proprietary DLSS upgrade solution). Although Nvidia’s latest promotional material mentions that it is an excellent value-for-money upgrade compared to the previous generation of notebook GPUs, we are still waiting for OEMs to release specific prices and specifications for their models in late January. (Also, we want to know if these laptops will sell out so fast that ordinary people cannot get them.)
Nvidia continues to use its smaller laptop models to borrow the naming convention of desktop models, and has tested it in some ways. The transition of laptops to the RTX 3000 series is achieved by including the second-generation ray tracing kernel and the third-generation tensor kernel, which have been proven on the RTX 3000 desktop model and can provide performance upgrades beyond pure rasterization , While reducing the space required on the physical hardware.
However, if you put the specifications of desktops and laptops together, the degradation of each product will be much greater than you would expect when you use the name alone. Especially for RTX 3080, the CUDA kernel has dropped by nearly 30%, while RTX 3070 has dropped by 13%. (The strange thing is that the RTX 3060 laptop version will have 6.6% More The CUDA core is higher than its desktop version, but 21.1% less than the desktop 3060 Ti. )
To be honest, the laptop version of RTX 3080 should probably be changed to another name, as it is also downgraded to a 256-bit memory bus and GDDR6 VRAM (instead of the pipe hot GDDR6X variant in the desktop model, which is attached to a 320-bit bus) . It’s not that you don’t need to downgrade to work within the heat dissipation and power limitations of the laptop case. We are just talking. Speaking of this, the clock speed of the RTX 3000 notebook computer series will also change across the board, and the range announced by Nvidia will definitely change further when it is put in the wild and installed in various OEM cases.
Surprisingly, despite the degraded power supply of VRAM, the RTX 3080 still has an optional capacity upgrade Reach 16GB GDDR6. This is not the rumored RTX 3080 Ti (it is reported that it may be equipped with up to 20GB GDDR6X VRAM), but if you are looking for serious VRAM in a studio-class laptop in 2021, you may want to crack more piggy banks1 Before the 26th.
Define “whisper”, Nvidia
In the rest of this week’s announcement, some efficiency gains are mentioned. These gains may be difficult to test because their improvements are mainly around “balanced power…per frame” and “simultaneous access to all GPU VRAM”. statement. (The latter is called “sizable BAR” and cannot be ready at boot time and needs to update the driver.) There is also the “WhisperMode 2.0” issue, which obviously allows users to select the fan noise level of their choice and view the system . Performance has also been adjusted accordingly. It remains to be seen how the performance changes from “listening to your own thoughts” to “standing in the airport hangar”.
We look forward to experiencing these GPUs in person to understand exactly how these promises are fulfilled. From the looks of it, Nvidia will once again beat AMD to enter the market in a given generation, while Red Team only provides the vague “first RDNA2 laptop.” The GPU series is promised to be realized in the first half of 2020.
Nvidia’s listing image