[Hardware] NVIDIA GeForce RTX 3070 Mobility GPU Specifications & Benchmark Leak Out, GA104 With 5120 Cores, 8 GB GDDR6 Memory & 17% Slower Than Desktop RTX 3070

[_Happy boy]

The specifications of NVIDIA's GeForce RTX 3070 mobility graphics card have been confirmed within a listing over at Geekbench. The listing was spotted by TUM_APISAK and shows the next-gen RTX 30 series mobility GPU running in the OpenCL test suite.

NVIDIA GeForce RTX 3070 Mobility GPU To Feature GA104 GPU With 5120 Cores, 8 GB Memory & 1300 MHz Clock in Max-Q Variant, On Par With An RTX 2080 SUPER
Just a few days ago, we got to see the first listings for the NVIDIA GeForce RTX 30 series mobility lineup along with some preliminary specifications. It seems like those specifications might end up being true since they match the ones from today's leak. Also, based on the clock speeds alone, we can tell that this is the Max-Q variant of the GeForce RTX 3070 instead of the Max-P variant which would offer much higher clock speeds.

So coming to the specifications, the NVIDIA GeForce RTX 3070 Mobility will feature 40 SMs on the GA104 GPU SKU (presumably, the GA104-770). This chip would feature a total of 5120 cores and a maximum clock speed of up to 1300 MHz (~1290 MHz). The graphics card also features 8 GB VRAM which runs across a 256-bit wide bus interface but it remains to be seen if NVIDIA is utilizing 12 or 14 Gbps DRAM modules for its mobility lineup. In the case of 14 Gbps, the GPU will be able to offer a maximum bandwidth of 448 GB/s.

Preliminary specifications coming in from NotebookCheck also seem to indicate that the Max-Q variant will feature a TGP between 80-90W while the faster Max-P variant will feature a TGP of 115-150W. The NVIDIA GeForce RTX 3070 Mobility was tested on an HP Omen 15 laptop which also had the Intel Core i7-10870H and 16 GB of DDR4 memory installed.

 

 

In terms of performance, the GPU scored 110839 points in the OpenCL test. The score puts the GeForce RTX 3070 mobility on par with the GeForce RTX 2080 SUPER but at the same time, it is slower than the desktop GeForce RTX 3070 by 17% which should be expected since the variant does not only offer lower cores but also comes with much lower clocks being a Max-Q design. It looks like the Max-P variant should close the gap and bring the performance difference below 10 percent but that remains to be seen.

Laptop makers will also be offering high-end and enthusiast configurations of NVIDIA GeForce RTX 30 Ampere GPUs with AMD's Ryzen 5000H 'Cezanne-H' CPUs and Intel's Tiger Lake-H 11th Gen CPU lineup at CES 2021.