laitimes

No graphics card! NVIDIA has a second spring in China: self-driving chips are being bought by automakers

In the hot new energy vehicle market, NVIDIA has successfully card slots, although it does not build cars, but its automatic driving computing chip is being bought by car manufacturers, especially China's new power brands.

According to incomplete statistics, the Xiaopeng G9, WEILAI ET7/ET5, and Ideal X01 (not yet released) that are scheduled to be delivered this year will officially announce that they will use NV's ADAS chip Orin.

In addition, Jidu Automobile's 2023 mass-produced "automotive robots" and Polar Star Automobiles have also confirmed that they will use Orin.

No graphics card! NVIDIA has a second spring in China: self-driving chips are being bought by automakers

Orin can be classified into the Tegra family, which is NV's system-level SoCs based on ARM CPUs and self-developed GPUs. Very early on, Tegra also tried to use for mobile phones, tablets, boxes, etc., although the graphics performance is powerful and the power consumption is touching, the software adaptation is also poor, I did not expect that since the transformation of Tegra X2, NV has found a new direction for it to find an automotive platform, and now shines.

In terms of parameters, Orin is manufactured using 7nm process, the CPU is 12-core Cortex-A78, GPU integrated Ampere, with 2024 CUDA units, integer computing power up to 254TOPs, support 4-way 10 Gigabit network, 8K 30P decoding and 4K 60P encoding.

Among the models currently announced, the NIO ET5/ET7 is the most violent, and the bicycle is equipped with 4 Orins, which is the terrifying scale of 48-core CPU/8096 CUDA/1016T hashrate. For comparison, Tesla's self-developed autopilot chip FSD, two bicycles are only 100T computing power.

Daniel Ives, president of Wedbush Securities, pointed out to the media that the close cooperation between Chinese automakers and NVIDIA ADAS chips will further narrow the gap between them and Tesla in the field of assisted/autonomous driving, and this will also bring more sustainable revenue to NVIDIA.

Read on