Apple M5 Ultra: a new performance king among SoCs, challenging NVIDIA’s discrete graphics cards

Recent leaks and benchmark results have put Apple’s new M5 Ultra processor in the spotlight of high-performance computing. According to data from the Blender test database, the chip’s 80-core GPU has demonstrated impressive power capable of competing with NVIDIA’s most advanced solutions. Most notably, the M5 Ultra came very close to the flagship NVIDIA GeForce RTX 5090, while significantly outperforming the RTX 5080.

A revolutionary leap in performance: M5 Ultra vs. the market leaders

Information leaked from testing labs points to a real breakthrough in Apple Silicon development. The new M5 Ultra, equipped with an 80-core GPU, scored around 15,350 points in the Blender Open Data synthetic test. To understand the scale of this achievement, it is worth comparing it with previous generations: the M5 Max scored around 8,112 points, while its predecessor, the M3 Ultra, scored approximately 7,234 points. This result demonstrates not just an evolutionary step, but a revolutionary leap in graphics performance.

The comparison with discrete desktop graphics cards looks even more impressive. The NVIDIA GeForce RTX 5080, considered one of the most powerful graphics cards on the market, scores an average of about 9,100 points in similar tests. The M5 Ultra turned out to be almost 40% faster, which is a significant achievement for an integrated graphics solution that is part of a single chip.

The gap to the RTX 5090, the undisputed market leader, is especially interesting. NVIDIA’s flagship usually scores around 16,000 points in Blender. According to the new data, the M5 Ultra trails it by less than 5%. This suggests that Apple has reached near parity with the very top tier of discrete graphics cards, opening new horizons for professional Mac users.

The road to leadership: architectural advantages and software optimization

Such significant progress by the Apple M5 Ultra can be explained by several factors. First, there is the continued evolution of the Apple Silicon architecture, which shows high efficiency in both CPU and GPU workloads. An increase in GPU core count, cache optimization, and improved intercomponent communication make it possible to achieve such impressive results.

Second, although synthetic benchmark results do not always fully reflect real-world performance in practical scenarios, they still provide a clear picture of the potential. It is important to understand that Blender, like any software, uses specific programming interfaces. In the case of the M5 Ultra, the test likely ran using the Metal API, while NVIDIA graphics cards use CUDA and OptiX. This can lead to certain differences in performance in real, complex workloads. However, the fact remains: Apple’s chip is demonstrating power that was previously available only in the most expensive discrete graphics cards.

An important aspect that sets Apple’s solution apart is the huge amount of unified memory available to both the CPU and GPU. This can be a major advantage when working with extremely large models, complex 3D scenes, and large datasets in artificial intelligence, where a lack of RAM can become a critical limitation for traditional systems.

Market prospects and challenges

The arrival of the Apple M5 Ultra with such high performance scores signals the company’s serious ambitions in the professional workstation and high-end laptop segments. This could significantly affect a market traditionally dominated by discrete graphics card manufacturers. Users looking for a powerful yet energy-efficient solution may find an attractive alternative in the new Macs.

However, it is important to remember that NVIDIA is likely to retain its advantage in certain specific scenarios. For example, in tasks that require maximum computing power for specialized scientific calculations or in deep learning, where CUDA-compatible solutions benefit from a deeply developed ecosystem of software and optimizations. In addition, NVIDIA continues to actively develop its own technologies, such as DLSS, which significantly improve performance in games and professional applications.

Despite this, Apple M5 Ultra’s achievement is a major step forward. It shows that integrated system-on-chip (SoC) designs can compete effectively with discrete components, offering users a more unified and optimized solution. The continued development of Apple Silicon, along with competition from other manufacturers, promises many more exciting innovations in the future, which ultimately benefits end users by giving them access to ever more powerful and efficient technologies.