Since the start of this year, the emergence of generative AI has been notable, leading to advancements in productivity for numerous individuals through applications like Wensheng words and Wensheng diagrams. Nevertheless, as generative AI becomes more widespread, it necessitates enhanced specifications for terminal hardware.
To ensure an exceptional generative AI experience and optimize work efficiency, an acquaintance of mine decided to upgrade their home computer’s graphics card. They replaced the previous “standard” RTX 3080 with the now elusive RTX 4090, which is currently unavailable in the market.
Generative AI Approaches Close Proximity with Snapdragon X Elite
While desktop PCs can harness this technology through hardware upgrades like graphics cards, laptops require a hardware platform upgrade. Unlike traditional X86 architecture PCs, ARM architecture notebooks have gained prominence in recent years due to their notable performance, energy efficiency, and mobile connectivity. These notebooks, including the recently unveiled Snapdragon X Elite computing platform at the Snapdragon Summit, are specifically designed to support high-performance, energy-efficient, and generative AI applications.
The Snapdragon X Elite outperforms the i9-13980HX, a processor tailored for high-performance gaming terminals, in terms of single-threaded performance of the Oryon CPU. The Oryon CPU delivers the same level of performance while consuming up to 70% less energy, making it a highly efficient option. However, the Snapdragon X Elite’s true advantage lies in its potential for generative AI, which makes it a promising choice for the future.
the Snapdragon X Elite has 75 TOPS of AI computing power
Currently, the primary approach employed by the mainstream X86 architecture SoC to address the computational demands of AI tasks is through the utilization of CPU+GPU. This combination ensures a balance between performance and energy efficiency. However, Qualcomm has taken a step further with the Snapdragon X Elite by incorporating an independent Hexagon NPU to cater to the requirements of AI and neural network computing. This NPU supports INT4 precision and possesses an independent computing power of up to 45 TOPS. Additionally, it can execute the Llama 70 model, which consists of 30 billion parameters, at a speed of 2 tokens/s on the terminal side, thereby delivering prompt and accurate responses.
By integrating CPU, GPU, and NPU, the Snapdragon X Elite boasts an impressive AI computing power of 75 TOPS. This enhanced capability enables the device to handle generative AI models with over 130 billion parameters. Based on the data provided by Qualcomm, the Snapdragon X Elite achieved a remarkable score of 1750 in the UL Procyon AI test. This outstanding performance empowers the device to effortlessly handle a wide range of generative AI tasks with ease.
The current flagship processors, namely 13800H and 7940HS, have the highest score of 232 when compared. However, to enhance the AI experience, one must rely on high-performance independent graphics cards, resulting in increased power consumption and hindering the achievement of a lightweight and slim notebook design.
Snapdragon X Elite, on the other hand, aims to popularize generative AI in notebook products. It offers users two significant benefits: enhanced AI computing power and expanded application scenarios.
At the onset of this year, Qualcomm spearheaded the implementation of the Stable Diffusion model with 10 billion parameters on the Snapdragon platform. The time taken to generate an image from text was reduced to less than 0.6 seconds on the Snapdragon X Elite, as opposed to the previous 15 seconds. The higher computing power of the Snapdragon X Elite device notebook also enables it to handle more intricate and demanding AI tasks, resulting in increased productivity value.
Apart from the fundamental Wensheng diagram application, we also explored other generative AI application scenarios of the Snapdragon X Elite during the Snapdragon Summit. For instance, by using a Snapdragon phone to capture a series of photos and video clips of users, and utilizing Qualcomm Snapdragon Seamless technology to share and synchronize them in real-time to a laptop equipped with the Snapdragon X Elite platform, users can effortlessly crop out the characters in the photos or videos and add them to a new background to generate a new image through the customized version of the Blackmagic Design da Vinci software. The entire process of cropping is exceptional, allowing users to quickly and easily produce new photos or videos.
In earlier times, achieving such smoothness required the use of a PC equipped with a high-performance discrete graphics card, which was not at all conducive to mobile work. However, with the Snapdragon X Elite, this can now be easily accomplished on a laptop. This is merely a small example of how generative AI is being applied to notebook products, and Qualcomm is collaborating with Microsoft, Blackmagic Design, and Adobe to explore more comprehensive use cases and solutions.
Summary:
Currently, generative A is in its early stages, and it is unavoidable to address increasingly intricate application scenarios and heightened productivity requirements in the coming years. The advancement of AI computing power is crucial in attaining these objectives. The Snapdragon X Elite has been specifically engineered to optimize power utilization and enhance productivity value, considering the compact form factor and limited battery capacity of notebook devices.
Qualcomm collaborated with OEMs including Lenovo, HP, and Honor, alongside partners like Microsoft, Blackmagic Design, and Adobe, to foster the development of the Snapdragon X Elite ecosystem. Their collective efforts aim to transform the AI capabilities of Snapdragon X Elite into tangible productivity benefits. Anticipate the release of laptop terminals integrated with Snapdragon X Elite in the upcoming year, presenting an enticing option for users.
Reference for this article：When buying a laptop in the future, you can’t afford to buy a laptop without considering the generative AI experience
