CES2015: 264-core Tegra X1 scores more points than Apple A8X

NVIDIA released the Tegra X1 processor at noon today, the day before the CES2015 exhibition in Las Vegas, the United States. The GPU of Tegra X1 adopts the Maxwell architecture, with a total of 256 stream processors, which is comparable to entry-level independent graphics cards. Only a few hours later, its running score was exposed on the website. Judging from the score, it is definitely the best ARM processor at present, without a doubt.

In terms of architecture, Tegra X1 integrates four Cortex-A57 cores and four Cortex-A53 cores, the same as Snapdragon 810 and Samsung Exynos 7 series. The GPU part adopts the Maxwell architecture, with a total of 256 stream processors, comparable to entry-level discrete graphics cards.

The picture comes from the Internet

In addition, Tegra X1 is manufactured using a 20nm process, which can effectively reduce its power consumption. The computing power under FP16 single precision is as high as 1TFlops, which is comparable to the supercomputer "ASCI Red" 15 years ago.

So, what is the performance of Tegra X1? Foreign media Anandtech has conducted a preliminary test of its GPU performance, and the results are amazing.

In the BaseMark X test, Tegra X1 scored 56.9FPS. In comparison, Apple A8X scored 40.2FPS, and Tegra K1 scored 36.3FPS.

The picture comes from the Internet

In 3DMark 1.2 Unlimited mode, Tegra X1 scored 58448. In comparison, Apple A8X scored 31781 and Tegra K1 scored 36688.

The picture comes from the Internet

In the GFXBench 3.0 Manhattan 1080p (Offscreen) test, the Tegra X1 scored 63.6FPS. In comparison, the Apple A8X scored 32.6FPS and the Tegra K1 scored 31.7FPS.

The picture comes from the Internet

I won’t go into detail here about how much we’re ahead, but it’s a crushing victory anyway.

The picture comes from the Internet

What? You said that the power consumption of Tegra X1 must be frighteningly high. Actually, you are wrong. Anandtech's preliminary power consumption test shows that the GPU power consumption of Tegra X1 is only about 1.51W, while the GPU power consumption of Apple A8X reaches 2.67W.

About Tegra X1

Before CES2015 officially opened, NVIDIA had already launched its big move. NVIDIA co-founder, president and CEO Jensen Huang announced the official release of the new generation of mobile super chip Tegra X1, which has twice the performance of the previous generation chip and can be used in mobile products such as mobile phones and tablets.

Mobile super chip Tegra X1

Last year, NVIDIA released the powerful Tegra K1 mobile chip, which was already outstanding in terms of performance, and the new generation of mobile super chip Tegra X1 is even more powerful. Tegra K1 is equipped with a 192-core Kepler GPU and supports the Unreal 4 game engine based on DirectX 11.1, providing PC-level game effects for mobile devices. The dual-core Denver CPU core based on the 64-bit ARM v8 architecture has a maximum frequency of up to 2.5GHz, which once surpassed Apple and Qualcomm Snapdragon at the time.

Tegra X1 is based on the NVIDIA Maxwell GPU architecture, which was released a few months ago and is used in the world's highest performance gaming graphics card GeForce GTX 980. It is the industry's first mobile processor with teraflop processing performance. It uses the latest Maxwell architecture and has 256 GPU cores. The hardware and performance are twice that of the previous generation product Tegra K1!

Tegra X1 has twice the performance of Tegra K1

Tegra X1 has a processing performance of more than 1 teraflop, which is faster than ASCI Red, the fastest supercomputer 15 years ago. ASCI Red is the only system in the world that can achieve teraflop processing performance. However, ASCI Red occupies an area of ​​1,600 square feet (about 149 square meters) and consumes 500,000 watts of power. In addition, it consumes another 500,000 watts of power to dissipate the space it occupies. Tegra X1 is only the size of a thumbnail and consumes less than 15 watts of power. It is not easy to save power while ensuring performance.

Performance comparison between Tegra X1, Tegra K1 and A8X

The Tegra X1 processor will be used in the newly released NVIDIA DRIVE automotive computers. DRIVE PX is an automotive autonomous driving computing platform that processes video from 12 on-board cameras to enable features such as surround vision and auto-valet. Surround vision provides a seamless 360-degree view around the car, while the latter enables true autonomous parking. DRIVE CX is a complete cockpit platform designed to achieve advanced graphics capabilities to meet the needs of an increasing number of screens. These screens are used for digital instrument clusters, infotainment, head-up displays, virtual rearview mirrors, rear-seat entertainment, etc.

[[125735]]

DRIVE CX

Smart cars equipped with NVIDIA Drive CX will be truly "cars of the future". Powered by Tegra X1, NVIDIA Drive CX will provide extremely fine image quality, and the icons on the dashboard will be cooler and more avant-garde!

DRIVE CX

“We see a future of self-driving cars, robots and drones that can see and learn with seemingly unimaginable intelligence,” said Jen-Hsun Huang, CEO and co-founder of NVIDIA. “They will enable safer driving, safer cities and a host of other benefits for all of us. To achieve this dream, we need huge advances in vision and parallel computing. The Tegra X1 mobile superchip is a major leap forward in this transformation, with one teraflop of processing performance.”

Tegra X1 supports major graphics standards, including Unreal Engine 4, DirectX12, OpenGL 4.5, CUDA, OpenGL ES 3.1 and Android Extension Pack, making it easier for developers to bring PC games to mobile platforms.