Nvidia RTX Spark Geekbench Scores Leaked: 23,424 Multi-Core Rivals Intel i9-14900KS

2 Sources

Share

Nvidia's ARM-based RTX Spark processor achieves 23,424 multi-core in Geekbench tests, matching Intel i9-14900KS performance. The 20-core chip surpasses AMD Ryzen AI Max+ 395 and Intel Core Ultra 9 275HX but trails Snapdragon X2 Elite in single-core tasks.

News article

RTX Spark ARM-Based Processors Surface in Benchmark Listings

Nvidia RTX Spark Geekbench scores have emerged online, revealing performance metrics for the company's upcoming ARM-based gaming platform ahead of its Fall 2026 launch in Windows laptops

1

2

. Two separate listings showcase 20-core and 18-core configurations, marking a significant shift from conventional x86 architecture. Jensen Huang first unveiled the platform at Computex in May, stating the company's intention to "reinvent the PC" after 40 years of largely unchanged architecture

1

. The RTX Spark platform targets gaming and agentic AI workloads while maintaining low power consumption, leveraging Nvidia's experience building ARM gaming chips for Nintendo Switch 2

1

.

Multi-Core CPU Benchmarks Challenge Intel and AMD

The flagship 20-core RTX Spark processor delivered impressive multi-core performance with a Geekbench score of 23,424, nearly matching the Intel i9-14900KS which scored 23,785 in identical testing

1

. This multi-core performance surpasses AMD Ryzen AI Max+ 395, also known as Strix Halo, and outpaces Intel Core Ultra 9 275HX, demonstrating clear advantages over x86 contemporaries

2

. The 18-core variant performs within 6% of the flagship configuration, suggesting consistent performance scaling across the lineup

2

. These results indicate RTX Spark will excel at multitasking scenarios like running multiple applications simultaneously or handling rendering workloads.

Single-Core Performance Trails Competition

While multi-core performance impresses, single-core performance presents a different picture. The 20-core RTX Spark achieved 2,593 in single-core testing, falling short of competitors including Snapdragon X2 Elite and Apple M5

1

. Single-core scores measure how efficiently processors handle individual tasks like photo editing, suggesting potential sluggishness in these scenarios

1

. The chip matches Apple M3 Max performance but cannot reach M4 Pro levels

2

. However, driver optimizations and software refinements before the Fall 2026 release could improve these numbers

2

. Qualcomm's Snapdragon X2 Extreme has recorded scores exceeding 25,000 in multi-core tests, setting a high bar for Windows on Arm laptops

1

.

GPU Configurations Differ Across RTX Spark Variants

Benchmark listings reveal distinct GPU specifications between configurations. The 20-core processor integrates a 6,144 CUDA core RTX Blackwell GPU, while the 18-core version features 5,120 CUDA cores

2

. This difference will directly impact PC gaming performance and GPU-intensive tasks beyond what CPU benchmarks reveal

2

. The platform will support DLSS 4.5, enhancing gaming capabilities

1

. Nvidia is ensuring compatibility with common anti-cheat and DRM software, addressing a critical concern for Windows gaming

1

. Most Windows games should run on the platform, though complete gaming performance data remains unavailable.

Launch Partners and Market Positioning

Nvidia has confirmed partnerships with Asus, MSI, Microsoft, and additional manufacturers for RTX Spark-powered devices launching Fall 2026

1

. These Windows on Arm laptops and mini PCs will ship with Windows 11 pre-installed, targeting premium gaming and productivity segments

1

2

. The platform's competitive multi-core benchmarks position it favorably against Apple M5 standard configurations and Intel Panther Lake processors, though Apple M5 Pro and Max variants maintain performance advantages

1

. Watch for additional GPU benchmarks and real-world gaming tests as the launch approaches, which will provide clearer performance comparisons across different workloads and usage scenarios.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved