On June 3, 2024, AMD officially launched the brand-new Ryzen AI 300 series processors. The release of this series has garnered widespread attention, not only because of its new naming convention but, more importantly, because the Ryzen AI 300 series processors bring significant updates across three major architectures (Zen 5, XDNA 2, and RDNA 3.5). Moreover, these processors not only feature the most powerful NPU currently available but also fully comply with Microsoft's Copilot+PC standards, marking the beginning of a new era for AI PCs.

On July 28, new laptop models equipped with AMD Ryzen AI 300 series processors hit the market. How does this new generation of processors perform in reality? "Microcomputer" has already obtained the first test units, so let's unveil the answers together!

A Brief Analysis of the AMD Ryzen AI 300 Series Processors

In early 2023, AMD introduced the first dedicated NPU in the Ryzen 7040 series mobile processors, officially ushering in the era of AI PCs. Thus, the Ryzen 7040 series mobile processors can be considered the first generation of AMD AI PC processors, with an NPU computing power of 10 TOPS. In early 2024, the Ryzen 8040 series mobile processors were released, optimized from the previous generation, with a 60% increase in NPU computing power to 16 TOPS, and are known as the second generation of AMD AI PC processors. The Ryzen AI 300 series processors, launched on the 28th, are the third generation of AMD AI PC processors, with significant updates in CPU architecture, NPU architecture, and GPU architecture.

Advertisement

● Zen 5 CPU Architecture: Comprehensive Optimization with a 16% IPC Increase

The AMD Ryzen AI 300 series processors utilize TSMC's 4nm production process, and the CPU has been upgraded to the Zen 5 architecture. This is a new generation of CPU architecture that AMD has heavily invested in, with IPC performance once again achieving a leap forward.

In simple terms, the Zen 5 architecture has four main areas of improvement, including: enhancing the accuracy of branch prediction to reduce latency; using wider pipelines and vector units to achieve higher throughput; achieving higher parallelism through deeper window size design; and strengthening support for AI acceleration.

Compared to the Zen 4 architecture, the Zen 5 architecture sees up to a twofold increase in instruction bandwidth for front-end instructions, data bandwidth from L2 cache to L1 cache, data bandwidth from L1 cache to floating-point units, and performance related to AI and AVX-512 instruction sets. The end result of these improvements is an average IPC performance increase of up to 16% compared to the Zen 4 architecture.

According to data released by AMD, the Zen 5 architecture performs even better in AI applications. Compared to Zen 4, it can achieve up to a 32% increase in single-core performance in machine learning applications, and up to a 35% increase in AES-XTS encryption algorithm performance.

● XDNA 2 NPU Architecture: NPU Computing Power Soars to 50 TOPSThe AMD Ryzen AI 300 series processors feature an NPU with the newly upgraded XDNA 2 architecture, marking a significant breakthrough in AI performance and specifications.

Compared to the previous XDNA architecture, the AI engine module of XDNA 2 has increased from 20 to 32, with each module having twice the MACs and 1.6 times the on-chip memory. It also supports the Block Floating Point block computation format and enhanced non-linear support. Consequently, its computing power has skyrocketed from 10 TOPS to 50 TOPS, leading the Qualcomm Snapdragon X Elite (45 TOPS), Intel Lunar Lake (48 TOPS), and Apple M4 (38 TOPS), making it the NPU with the highest computing power to date.

The XDNA 2 architecture can support up to 8 concurrent spatial streams, enabling better multitasking capabilities. It also features column-based power gating, which can achieve twice the energy efficiency compared to the Ryzen 7040 series. It's worth mentioning that AMD has added support for the Block Floating Point format, which offers high efficiency close to 8-bit integer operations and precision close to 16-bit floating-point operations. It should be emphasized that, thanks to the 50 TOPS NPU computing power, the AMD Ryzen AI 300 series processors fully meet the requirements of Microsoft Copilot+PC, providing excellent support for AI features brought by Microsoft Copilot+PC, such as Recall and real-time subtitles.

● RDNA 3.5 GPU Architecture: Enhanced GPU Performance and Energy Efficiency

In addition to the powerful Zen 5 CPU architecture and XDNA 2 architecture, the integrated GPU architecture of the Ryzen AI 300 series processors has also been upgraded to the brand-new RDNA 3.5 graphics architecture, offering the Radeon 800M series integrated graphics cards.

Building on the RDNA 3 architecture, RDNA 3.5 further optimizes energy efficiency and enhances performance. The per-watt performance of the RDNA 3.5 architecture is stronger, mainly reflected in the texture sampling rate and the interpolation and comparison rate of the vector ISA, which are twice as fast as before. The RDNA 3.5 architecture also improves memory management efficiency, including improved raw batch processing to reduce memory access, better compression techniques, and optimized LPDDR5 memory access.

According to data released by AMD, the Radeon 890M graphics card integrated in the Ryzen AI 9 HX 370 processor (54W), combined with AMD Fluid Motion Frames technology, can even significantly outperform the RTX 2050 discrete graphics card (60W) in 1080p medium gaming settings.

The AMD Ryzen AI 300 series processor family currently includes three models: Ryzen AI 9 HX 375, Ryzen AI 9 HX 370, and Ryzen AI 9 365. Here is a brief introduction to AMD's new processor naming convention: "AMD Ryzen AI" represents the brand of the processor; "9 HX" and "9" represent the brand level, with only Ryzen AI 9 HX and Ryzen AI 9 levels currently available, and the previous Ryzen 9, Ryzen 7, Ryzen 5 series classifications, as well as HX, HS, H, U suffixes, may no longer apply; the following numbers "3" and "7" represent the generation and product series of the Ryzen AI processor, which is currently the third generation of Ryzen AI processors; the final "0" represents the specific product SKU.

Launch Model: ASUS ZenBook 16 Air AI Ultra-Thin Laptop

On July 28th, new laptop models featuring the AMD Ryzen AI 300 series processors were launched. The ASUS ZenBook 16 Air AI ultra-thin laptop we received at the earliest opportunity is one of the first models to hit the market. As a new high-end ultra-thin laptop product featuring the AMD Ryzen AI 300 series processor, what is the actual experience of the ASUS ZenBook 16 Air AI?Product Specifications

Operating System: Windows 11 Home (64-bit)

Main Display: 16-inch OLED ASUS Good Screen (2880x1800@120Hz, 550 nits peak brightness)

Processor: AMD Ryzen AI 9 HX 370 (12-core 24-thread, max frequency 5.1GHz)

Memory: Onboard 32GB LPDDR5x 7500 dual-channel memory

Storage: 1TB PCIe 4.0 SSD

Graphics Card: AMD Radeon 890M (16 CU units, 2900MHz)

Battery: 78Wh

Dimensions: 353.6mm x 243mm x 11.9mm

Weight: Approximately 1.49kg (without power adapter)Reference Price: 12,999 yuan

From the product's name, it is evident that the ASUS ZenBook 16 Air is an ultra-thin 16-inch laptop. This machine has been redesigned from the inside out, with an overall lightweight design. The full-metal CNC unibody construction not only provides a sturdy structure but also makes the body more compact. As a 16-inch model, this machine weighs a mere 1.49 kg, with the thinnest part of the body being only 11.9 mm. It is noticeably slim when held, and it is almost weightless to carry around.

Inheriting the classic design style of the ASUS ZenBook series, this machine's A-face is highly recognizable—with an overall Snowy White color scheme and the iconic geometric line design on the A-face, it is immediately identifiable as part of the ASUS ZenBook family. Additionally, the A-face of this machine is made of high-tech ceramic aluminum material, which is the latest achievement after four years of ASUS research and development. It has a warm ceramic-like feel, texture, and lifespan, offering not only excellent aesthetics but also features such as stain resistance, durability, and wear resistance. Compared to other metal materials used in thin and light laptops, the ASUS ZenBook 16 Air is more elegant and fashionable.

The ASUS ZenBook 16 Air features a 16-inch OLED ASUS good screen with a screen resolution of 2880x1800. This is a cinema-level high-specification screen, supporting a contrast ratio of up to 1,000,000:1, and has passed the DisplayHDR True Black 500 certification, capable of displaying and playing high-quality HDR content.

Moreover, this screen also supports a 120Hz refresh rate, 0.2ms ultra-fast response time, and has a 10-bit color depth, capable of displaying approximately 1 billion colors. The peak brightness reaches 550 nits, and each screen is precisely calibrated at the factory, supporting Dolby Vision, delivering outstanding professional color and display effects.

At the same time, equipped with a smart ambient light sensor, the screen of this machine can automatically adjust the brightness and color temperature according to the surrounding light. To meet users' professional color needs, the ASUS ZenBook 16 Air also has four color space modes built-in for users to switch between: native color space, sRGB, DCI-P3, and Display P3. In the native color space mode, the color gamut coverage of this machine reaches 99.9% sRGB, 91.2% Adobe RGB, and 99.6% DCI-P3, which is quite impressive.

The ASUS ZenBook 16 Air is equipped with a 1080p IR infrared dual camera at the top of the screen, which can capture high-definition and delicate images and supports various AI portrait processing functions, system login via facial recognition, and automatic screen dimming, which is very helpful for office workers.

The C-face design of this machine is also simple, with a comfortable white backlit keyboard that is easy to type on for extended periods without fatigue. The top of the C-face features a geometric grille ventilation design, and the heat dissipation holes processed by CNC technology help to improve heat dissipation efficiency while also preventing dust and water from entering the interior of the laptop. Additionally, this machine features a 7.1-inch large touchpad with a 240Hz touch sampling rate, providing a comfortable operation feel, and the support for various smart gestures greatly enhances user convenience.

Although the ASUS ZenBook 16 Air has a compact and lightweight body, its interface configuration is very comprehensive. This machine is equipped with an SD card slot, one USB-A 3.2 Gen2 interface, two USB4 interfaces, one HDMI 2.1 TMDS interface, and a 3.5mm audio interface. Clearly, such interface design can not only meet the needs of ordinary users but also the needs of creators.

In terms of hardware configuration, the ASUS ZenBook 16 Air is first launched with the AMD Ryzen AI 9 HX 370 processor and AMD Radeon 890M graphics card. In terms of memory, it comes with 32GB LPDDR5x 7500 dual-channel memory onboard. From tests, the memory's read speed reaches 86967MB/s, the write speed is up to 101.51GB/s, and the latency is 113.8ns. Such performance fully leverages the capabilities of high-frequency memory, providing ample assurance for the powerful performance of the entire machine.The test unit we received is equipped with a Samsung 1TB PCIe 4.0 SSD, which achieves a sequential read speed of 5062.67MB/s and a sequential write speed of 3874.9MB/s in CrystalDiskMark 8.0.4. Such performance can bring a swift experience to users when loading software and copying data.

The ASUS ZenBook 16 Air comes with a large 78Wh battery, and under the settings of "Quiet Mode," "Best Battery Life," "Airplane Mode," 50% screen brightness, 20% speaker volume, and with power-saving mode turned off, it achieved a score of 16 hours and 32 minutes in the PCMark 10 "Video Playback" test. This excellent performance means that when carrying the ZenBook 16 Air for daily tasks, there is no need to worry about the battery running out.

AMD Ryzen AI 9 HX 370 Processor Real-World Test

The ASUS ZenBook 16 Air is powered by the AMD Ryzen AI 9 HX 370 processor, codenamed Strix Point, which is manufactured using TSMC's 4nm process technology and features 4 Zen 5 cores + 8 Zen 5c cores, forming a 12-core 24-thread configuration. It should be noted that AMD's combination of Zen 5 and Zen 5c is not the traditional "big.LITTLE" architecture, as the Zen 5c is not a "power-efficient core." It simply adopts a more compact design, but the core architecture and IPC are identical to the Zen 5.

The AMD Ryzen AI 9 HX 370 processor has a base frequency of 2.0GHz and a maximum boost frequency of 5.1GHz. The default TDP of this processor is 28W, but the configurable cTDP range is 15-54W. This means that despite the lack of traditional suffixes like H, HS, U, etc., in the naming, the AMD Ryzen AI 9 HX 370 processor is highly versatile and can be used in both gaming laptops and creative laptops, such as the ASUS TUF Gaming Ryzen AI Edition, as well as in ultra-thin laptops like the ASUS ZenBook 16 Air.

Additionally, the cache capacity of the AMD Ryzen AI 9 HX 370 processor has been increased to 36MB, which is a significant improvement. Compared to the 24MB of the Ryzen 7 8845HS processor, it represents a 50% increase, and it also surpasses the Core Ultra 7 155H (24MB cache) by more than 50%.

The AMD Ryzen AI 9 HX 370 processor integrates an AMD Ryzen AI NPU based on the XDNA 2 architecture, with an NPU compute power of 50 TOPS, and the total processor compute power can reach up to 80 TOPS. Moreover, this processor also includes an AMD Radeon 890M graphics card based on the RDNA 3.5 architecture, which uses 16 CU units and has a graphics frequency increased to 2900MHz. The new graphics card also supports technologies such as AMD SmartShift MAX, AMD Smart Access Memory, AV1 hardware encoding and decoding, ray tracing, and AMD FidelityFX Super Resolution (FSR).

What is the actual performance of the AMD Ryzen AI 9 HX 370 processor like? And how is the AI application experience on laptops equipped with this processor? We will test this through the ASUS ZenBook 16 Air.

As mentioned earlier, the AMD Ryzen AI 9 HX 370 processor is highly versatile and can be used in both gaming laptops and ultra-thin laptops. Our test unit this time is the ultra-thin laptop ASUS ZenBook 16 Air. Due to the limitations of the ultra-thin body, the performance release of this machine is conservative, with the processor's TDP at 28W in "full-speed mode."

Additionally, different laptop manufacturers, models, and performance tuning modes can all lead to performance differences. Therefore, even with the same processor, there can be a "thousand faces" in performance. Thus, the test results of this time only represent the performance of the ASUS ZenBook 16 Air and do not represent the highest performance of the Ryzen AI 9 HX 370 processor. Nonetheless, our test results are also very meaningful for the ultra-thin laptop market equipped with the Ryzen AI 9 HX 370 processor.For a more intuitive comparison, in addition to the ASUS ZenBook 16 Air with a 28W TDP, we have also included two comparison models in this test: one is the Ryzen 7 8845HS platform with a 40W performance release, and the other is the Core Ultra 7 155H platform, also with a 28W TDP. All three test units are equipped with 32GB LPDDR5x 7500/7467 (Core Ultra platform) dual-channel memory, and are set to the highest performance mode during testing.

**Processor Benchmark Performance Test**

Looking at the processor benchmark performance results, the Ryzen AI 9 HX 370 processor shows an impressive performance. Facing the Ryzen 7 8845HS platform with a 40W power consumption, the Ryzen AI 9 HX 370 platform, which only has a 28W power consumption, shows no fear. It achieved a single-thread score of 1917pts and a multi-thread score of 17239pts in CINEBENCH R23, leading the Ryzen 7 8845HS platform by approximately 11% and 14% respectively. In the multi-thread test of CINEBENCH 2024, the Ryzen AI 9 HX 370 platform also has a lead of about 13%. After all, both processors have a maximum frequency of 5.1GHz, yet the Ryzen AI 9 HX 370 processor still has an approximate lead of about 11% in single-thread performance, which is the benefit brought by the new Zen 5 architecture. Moreover, the power consumption of the Ryzen AI 9 HX 370 platform is even lower, which is indeed a significant progress in energy efficiency.

Compared to the Core Ultra 7 155H, which also has a 28W TDP, the Ryzen AI 9 HX 370 processor also shows a significant advantage. It leads by about 9% in the single-thread of CINEBENCH R23, and the lead in the single-thread of 3DMark CPU Profile is even more exaggerated, reaching 12.55%. In terms of multi-thread performance, the Ryzen AI 9 HX 370 has a huge advantage in the multi-thread test of CINEBENCH 2024, with a lead of up to 28.92%; it also has an approximate lead of 18% in the multi-thread test of CINEBENCH R23.

**Real-world Application Test**

In the x264 video encoding test, the performance improvement of the Ryzen AI 9 HX 370 platform compared to the Ryzen 7 8845HS platform is not significant, but in the currently more efficient x265 video encoding test, the performance advantage is demonstrated. The Ryzen AI 9 HX 370 platform leads by about 9%. In the 7-Zip compression/decompression test, which more relies on the processor's multi-thread performance, the advantage of the Ryzen AI 9 HX 370, which already has stronger multi-thread performance, is even greater, leading by about 12% in the total test score.

Compared to the Core Ultra 7 155H, the advantage of the Ryzen AI 9 HX 370 processor in the x264 video encoding test is about 5%, but in the 7-Zip and x265 video encoding tests, the Ryzen AI 9 HX 370 completely overwhelms the competition, with the maximum gap reaching 33%. Such a significant difference suggests that the Ryzen AI 9 HX 370 would likely have no problem competing with the Core Ultra 9 185H.

Next, let's look at the performance of the three platforms in scenarios such as web browsing, 3D rendering, and photo editing. As expected, the Ryzen AI 9 HX 370 platform completely outperforms the other two comparison platforms. Compared to the Ryzen 7 8845HS platform, the Ryzen AI 9 HX 370 platform leads by about 19% in the V-Ray 5.0.20 CPU Benchmark, by about 17% in the UL Procyon photo editing test, and by about 13% in WebXPRT 4.

Compared to the Core Ultra 7 155H platform, the Ryzen AI 9 HX 370 platform, aside from not showing a significant advantage in web browsing, performs quite prominently in the V-Ray 5.0.20 CPU Benchmark and UL Procyon photo editing test, leading by about 39% and 27% respectively.

We used the Blender 4.2 software to manually render the "BMW27" model to see how long it takes for the two Ryzen platforms to complete the rendering. The results show that the Ryzen 7 8845HS platform with a 40W performance release took 168 seconds, while the Ryzen AI 9 HX 370 platform with only a 28W performance release took 140 seconds, saving 28 seconds. This demonstrates that the new Ryzen AI 9 HX 370 not only has higher efficiency but also better energy performance.In summary, the all-new AMD Ryzen AI 9 HX 370 processor boasts impressive performance and power efficiency, allowing it to be easily integrated into ultra-slim laptops like the ASUS ZenBook 16 Air, which is as thin as 1.1cm. Despite its slim profile, it delivers powerful performance, significantly outperforming its predecessors and competitors. This means that the arrival of the Ryzen AI 9 HX 370 processor opens up new possibilities for laptops that combine high performance with ultra-portability.

**AMD Radeon 890M Graphics Card Testing**

Moving on to graphics performance, the AMD Ryzen AI 9 HX 370 processor features an integrated Radeon 890M graphics card based on the RDNA 3.5 architecture, equipped with 16 CU units and a graphics frequency boosted to 2900MHz. The RDNA 3.5 architecture further optimizes power efficiency and enhances performance. So, how does the Radeon 890M graphics card perform?

In the 3DMark graphics test, the Intel Arc graphics card integrated in the Core Ultra 7 155H managed to gain an advantage, leading by about 6% in the DX12, 1440p resolution 3DMark Time Spy. However, in the ray-tracing-based 3DMark Port Royal, the Intel Arc graphics card lags far behind the Radeon 890M, trailing by about 17%. It appears that the Radeon 890M supports ray-tracing performance better when gaming.

Compared to the previous generation Radeon 780M graphics card, the Radeon 890M has a significant lead in Time Spy and Port Royal, with a 19% advantage in Time Spy.

3DMark scores are theoretical tests, and we must also judge the actual performance in real games through popular titles. It can be seen that at 1080p resolution, the Radeon 890M graphics card performs best in "Overwatch: Return," achieving an average frame rate of 112.88fps at high settings, leading the Radeon 780M by about 18% and the Intel Arc graphics card by about 54%. In addition to the performance improvements of the RDNA 3.5 graphics card itself, we believe this is also related to the 36MB large cache of the Ryzen AI 9 HX 370 processor. After all, "Overwatch: Return" is an online game, and data during the game will be stored in the cache, thereby reducing the number of memory accesses and improving the game's performance. Of course, the impact of the game's own optimization on the graphics card cannot be ruled out.

In addition to "Overwatch: Return," the Radeon 890M's actual frame rates in several other games are also completely ahead of the Radeon 780M and Intel Arc graphics cards. Although the lead in some games is not very large, this also indicates that the RDNA 3.5 architecture has indeed brought about an improvement in graphics performance. Barring any surprises, the Radeon 890M is poised to earn the title of "the strongest integrated graphics card."

The D5 renderer is a well-known universal rendering software that utilizes DirectX ray tracing technology to achieve a powerful and excellent rendering experience on mobile platforms. The Radeon 890M graphics card supports ray tracing technology; will it bring a more outstanding performance in the D5 renderer? We rendered an 8K image in the D5 renderer, and the Radeon 890M graphics card completed the task in just 348 seconds, while the Radeon 780M graphics card took 370 seconds, demonstrating the higher efficiency of the Radeon 890M.

**Ryzen AI 9 HX 370 Platform AI Application Full Experience**

In the era of AI PCs, relying solely on the powerful performance of the CPU and GPU is far from enough to deliver an outstanding AI experience. The protagonist of the AI PC era— the NPU— must also be included. For the Ryzen AI 300 series processors, it is the synergy of the Zen 5 architecture CPU, RDNA 3.5 architecture GPU, and XDNA 2 architecture NPU that brings an excellent AI experience and overall user experience.As one of the earliest manufacturers of x86 processors with built-in NPU, and also the first to drive Windows Studio Effects through the NPU, AMD has partnered with the vast majority of mainstream AI ecosystem partners worldwide to provide AI-enhanced performance experiences for over 100 applications. For the ASUS ZenBook 16 Air equipped with the Ryzen AI 9 HX 370 processor, AMD has integrated a new generation NPU based on the XDNA 2 architecture, with a computing power of 50 TOPS, combining the computing power of the CPU and GPU, the overall platform computing power reaches 80 TOPS, fully preparing for the AI PC era. Taking the ASUS ZenBook 16 Air as an example, let's briefly experience the AI capabilities based on the new generation of AMD AI PC processors.

Firstly, the Ryzen AI 300 series processors fully meet the requirements of Microsoft Copilot+PC, so the Copilot+ feature suite launched by Microsoft can be enabled at any time. However, this part of Microsoft's functionality has not yet been officially enabled in China. If these features are officially enabled in the future, then laptops equipped with the Ryzen AI 300 series processors will be fully prepared. The above images are for illustrative purposes only.

Another significant change is that the NPU of the Ryzen AI 300 series processors can finally be displayed in the Windows Task Manager, allowing users to more conveniently view the real-time usage of the NPU.

Additionally, through AMD's extensive cooperation and in-depth collaborative design with ISVs, AMD has brought users a variety of useful AI software in partnership with its collaborators. Among the most significant is a software called "Amuse." Amuse 2.0.0 is a software that integrates Stable Diffusion AI drawing, text-to-image, and image-to-image capabilities. It is a single installation program, making it very simple to install without the need for any additional dependent plugins. After a "plug-and-play" installation, you can quickly launch AI drawing applications such as Stable Diffusion on your computer.

We know that AI drawing software like Stable Diffusion is very "memory and VRAM hungry." On some low-specification models, we often can only generate images with a resolution of 512x512, and cannot generate larger resolution images. How does AMD solve this problem for users? AMD has introduced the AMD XDNA Super Resolution feature based on the NPU. If you enable this feature in Amuse 2.0.0, during the process of generating a 512x512 image, AMD XDNA Super Resolution will call upon the NPU to automatically upscale the image to a resolution of 1024x1024. As we experienced in the image above, the entire process is essentially seamless.

Thanks to the powerful integrated computing power of the Ryzen AI 9 HX 370 platform, especially the enhanced GPU performance, completing text-to-image creative tasks with the ASUS ZenBook 16 Air is also very easy. In Amuse 2.0.0, we can generate a 512x512 image through the SDXL model in just 2.2 seconds, with a generation efficiency of 2.9 it/s, making the entire process quite fast.

Another issue is that most integrated graphics thin and light laptops share VRAM, and when VRAM is insufficient, generating large images like 1024x1024 with Stable Diffusion often results in errors. The ASUS ZenBook 16 Air does not have this problem because its VRAM size supports manual allocation, with a maximum of 8GB that can be allocated to VRAM.

After manually allocating 8GB of VRAM, we can generate 1024x1024 resolution images with the SDXL model in Amuse 2.0.0, and the system will not report errors even after generating dozens of times.

Furthermore, with the help of the Amuse Design feature, if we want to create images through our own creations, we only need to draw part of the material in Amuse Design, and the local large model will generate the image.

Additionally, there is the Camo Studio software, which supports the NPU based on AMD XDNA 2 and can help users enhance video image quality during video calls or live streaming. For example, in "Portrait" mode, we can use our hands to create emoticons, such as likes, waves, etc., and it offers many more features than Windows Studio Effects.The last AI software to introduce is the Gigapixel AI photo enlargement and repair software from Topaz Labs. In its latest release, Gigapixel AI has introduced a Beta mode called "Recovery," which primarily uses Stable Diffusion on the backend to restore lost details in images. We used Gigapixel AI to enlarge a 512x512 image by two times, and the entire process took 13 minutes and 23 seconds on the ASUS ZenBook 16 Air. Although the time was a bit long, the enlargement effect was quite good, and it is recommended for users who need it.

Summary: Fast-forwarding the development of AI PCs

The AMD Ryzen AI 300 series processors with a new naming scheme represent a super heavy upgrade for both AMD and the PC market. Looking back at the development history of AMD Ryzen mobile processors, each time there is such a leapfrog upgrade iteration, it always brings surprises, and this time is no exception.

Through the actual test of the new AMD Ryzen AI 9 HX 370 processor platform, we indeed felt the powerful potential of the combination of Zen 5 architecture, RDNA 3.5 architecture, and XDNA 2 architecture. The AMD Ryzen AI 9 HX 370 processor's performance in various tests not only surpassed the previous generation with higher power consumption but also strongly outperformed competitors.

More importantly, the AI application experience on the new platform is remarkable. The ultra-high computing power brought by the "CPU+NPU+GPU" of the AMD Ryzen AI 300 series processors has laid a solid foundation for the improvement of AI application experiences and can be said to have pressed the fast-forward button for the development of the AI PC ecosystem! It is believed that in the near future, we will see more and more AI software and application scenarios that fully leverage the potential of AI PCs, making users' AI application experiences readily accessible.

The ASUS ZenBook 16 Air with the AMD Ryzen AI 9HX 370 processor is just an AI notebook positioned as ultra-thin, and it can already exert such powerful strength. Therefore, in performance notebooks with normal thickness or gaming notebooks with more internal space, the performance potential we can imagine is even greater. Where is the performance limit of the AMD Ryzen AI 9 HX 370 processor or the Ryzen AI 300 series processors?

In any case, AMD has taken the lead today, and now the pressure is on the competitors. Want to try a super thin and light notebook with outstanding comprehensive strength? This summer, products featuring the AMD Ryzen AI 300 series processors are truly worth paying attention to!