TECH

RTX Spark vs M5 Pro

A direct comparison has been published between NVIDIA's RTX Spark and Apple's M5 Pro, evaluating which chip delivers the better laptop experience for users who need both productivity performance and GPU-accelerated workloads. The comparison comes at a pivotal moment, as RTX Spark represents NVIDIA's first entry into the PC CPU market with an Arm-based system-on-chip that integrates a desktop-class Blackwell GPU directly alongside ARM CPU cores, while Apple's M5 Pro is the latest iteration of the company's custom silicon that has defined premium laptop performance since the M1 debut in 2020.

On the CPU front, Trusted Reviews notes that Apple's M5 Pro holds advantages in single-threaded performance thanks to years of hardware-software co-design and TSMC's latest manufacturing process. However, RTX Spark's 20-core Arm CPU closes the gap significantly in multi-threaded workloads, particularly in scenarios that benefit from the chip's unified memory architecture and NVLink-C2C interconnect providing 600 GB/s of bandwidth between CPU and GPU. The comparison highlights that while Apple's lead in raw CPU benchmarks persists, the performance gap has narrowed to levels where real-world differences in productivity applications are minimal.

Where the comparison tilts decisively in RTX Spark's favor is GPU-accelerated workloads. The integrated Blackwell GPU with 6,144 CUDA cores — matching the desktop RTX 5070 — supports DLSS 4.5 with Multi-Frame Generation, full DirectX 12 Ultimate ray tracing, and can run AI models with up to 120 billion parameters locally. Trusted Reviews notes that for gaming, 3D rendering, and on-device AI inference, RTX Spark laptops offer capabilities that Apple's M5 Pro cannot match, making them a compelling choice for users who need workstation-class GPU performance in a portable form factor. The article concludes that the choice between the two platforms will largely depend on whether users prioritize Apple's ecosystem integration and proven battery life, or RTX Spark's superior GPU performance and AI capabilities when devices ship later this year.

Nvidia’s RTX Spark “Superchip” grabbed all the attention when it was first announced in June. At the time, it was revealed that RTX Spark-powered laptops will be made by Asus, Dell, HP, Lenovo, MSI, and Microsoft. Earlier this month, Microsoft unveiled the Surface Laptop Ultra with a 20-core Nvidia Grace CPU and up to 128 GB of LPDDR5X unified memory. Now, this machine has been tested against the much cheaper M5 Pro-powered MacBook Pro with some interesting results.

Starting with CPU benchmarks, the Surface Laptop Ultra (20-core CPU, 128 GB RAM) was consistently slower than the MacBook Pro (M5 Pro, 64 GB RAM). The testing included Cinebench 2026, Cinebench 2024, and Blender. In some cases, it was even slower than the Legion Pro 5i (Intel Core Ultra 9 275HX, 130W) and the HP ZBook Ultra (Ryzen AI Max+ Pro 395, 70W). We saw similar results with the Asus ProArt P16, which is even more expensive (as tested) than the Surface Laptop Ultra.

Creative workflows...It may come as no surprise to some, but the M5 Pro-powered MacBook Pro (14-inch) consistently beat the Surface Laptop Ultra with its 20-core N1X chip. Hardware Canucks posted their results showing comparisons in several creativity tools such as Adobe Photoshop, Lightroom, DaVinci Resolve Studio, as well as some AI workflows. Keep in mind that Nvidia has designed and marketed the RTX Spark platform to be an AI workhorse.

In every test, the Surface Laptop Ultra lagged behind the MacBook Pro. With the Adobe suite of applications, both Photoshop and Lightroom had the Surface Laptop Ultra (80W) trailing far behind. Photoshop, in particular, had it at the bottom, performing worse than the Intel Core Ultra 9 275HX (130W) and the Ryzen AI Max+ Pro 395 (70W).

On the GPU side, Premiere Pro and DaVinci Resolve performance was compared, and the Surface Laptop Ultra’s 6,144 CUDA-core GPU was a bit slower than the M5 Pro’s 20-core GPU (15W). Notably, Premiere Pro isn’t yet optimized for the RTX Spark platform, and DaVinci Resolve is heavily optimized for Apple’s M-series chips.

AI workflows...Coming to running LLMs, specifically the GGUF (GPT-Generated Unified Format) ones that support both Mac and Windows, only one test had the N1X (128 GB) beat the M5 Pro (64 GB), and that was Qwen 3.8 Q4. On the heavier Q8 model, the M5 Pro pulls ahead once again, and the gap remains as the loads increase.

The comparison with the M5 Pro MacBook Pro is quite interesting because of the price points. The Surface Laptop Ultra (as tested) is priced at $5,900, while the MacBook Pro (as tested) comes in at $3,700. That’s $2,200 more for worse performance. It should be noted that apps will become more optimized for RTX Spark, and the performance gap will reduce over time. Furthermore, gaming-focused scenarios still favor the RTX Spark, though our testing shows it performs close to the RTX 5060 or RTX 5070 laptop GPUs.

The NVIDIA RTX Spark (featuring the N1X chip) and the Apple M5 Pro represent a major showdown between ARM-based architectures and unified memory systems, both designed for content creators and local artificial intelligence (AI) workloads on laptops.

Key Specifications:

• NVIDIA RTX Spark (N1X):

• CPU: Up to 20 ARM-based cores (available in 18- or 20-core variants, balancing performance and efficiency). 

• GPU: Blackwell architecture with up to 6,144 CUDA cores. 

• Unified Memory / Bandwidth: Utilizes LPDDR5X with a bandwidth of approximately 273 GB/s to 300 GB/s (depending on the model/configuration). 

• Operating System: Runs Windows 11 (utilizing x86 emulation via Prism).

• Apple M5 Pro:

• CPU: Apple’s proprietary ARM-based architecture (available in 14-, 15-, or 18-core configurations). 

• GPU: Apple’s high-efficiency integrated GPU. 

• Unified Memory / Bandwidth: Approximately 307 GB/s of memory bandwidth. 

• Operating System: macOS (with native optimization via MLX and creative software suites).

Performance and Comparison:

• Artificial Intelligence (AI) and Local Models:

• The RTX Spark excels in generative AI workflows (such as image generation, Stable Diffusion, or Flux) thanks to NVIDIA’s CUDA software ecosystem and dedicated accelerators. 

• Conversely, for the inference of smaller language models using optimized frameworks (like MLX on Mac), the M5 Pro holds an advantage in efficiency and raw speed in certain scenarios—though the M5 Max outperforms both, benefiting from significantly higher memory bandwidth. • Content Creation and Video Editing:

• The MacBook Pro with the M5 Pro chip is consistently faster and more stable in applications such as Adobe Photoshop, Lightroom, Premiere Pro, and DaVinci Resolve Studio (all of which feature mature optimization for Apple Silicon chips). 

• The RTX Spark (found in models like the Asus ProArt or Surface Laptop Ultra) still suffers from a lack of optimization in certain editing software, leading to higher latency or dropped frames when handling complex timelines, especially when compared directly to Apple’s hardware.

• Bandwidth:

• The M5 Pro offers approximately 307 GB/s of bandwidth, surpassing the 273 GB/s – 300 GB/s range of the RTX Spark in terms of token decoding speed and memory bottleneck mitigation.

The NVIDIA RTX Spark (an SoC featuring a 20-core ARM CPU and an integrated Blackwell-based GPU equivalent to an RTX 5070) and the Apple M5 Pro (an Apple Silicon architecture with up to 18 cores) represent distinct visions for high-performance portable computing.

Key differences:

• Graphics Performance and Gaming: The RTX Spark holds a significant advantage in dedicated graphics, full ray tracing (DirectX 12 Ultimate), support for DLSS 5 with neural rendering, and compatibility with hundreds of games at launch (running either natively or via emulation within the Windows/Microsoft ecosystem). While the M5 Pro excels in efficiency, it is limited when it comes to demanding AAA games and the DirectX graphics ecosystem.

• Local Artificial Intelligence (AI): The RTX Spark integrates cores with direct support for the CUDA ecosystem and is capable of running large-scale local AI models, whereas the M5 Pro relies on the closed-source optimization of macOS and Apple’s unified architecture.

• CPU and Efficiency: The M5 Pro shines in single-thread performance and seamless hardware-software integration (macOS), consuming less power during everyday productivity tasks. The RTX Spark (utilizing licensed ARM cores, such as the Cortex-X925) focuses on raw GPU power and leverages partnerships with Microsoft to power laptops like the ASUS ProArt and the Microsoft Surface Laptop Ultra.

• Price and Ecosystem: Laptops equipped with the RTX Spark are positioned as high-end, premium devices—often comparable in price to MacBook Pro models—yet they require emulation to run traditional x86 Windows software. In contrast, the M5 Pro operates entirely natively within the macOS environment.

 

mundophone