RTX Spark’s 20-Core CPU Disappoints In Cinebench 2026’s Multi-Core Leak Despite Set To High Performance Mode, Single-Core Results Show Promise But Only Against M3 Max

Omar Sohail
RTX Spark disappoints in Cinebench 2026 multi-core results
The Blackwell GPU and unified memory are the only two impressive things about the RTX Spark

The decision to employ older Cortex-X925 and Cortex-A725 cores on the RTX Spark to form a 20-core CPU configuration will probably haunt NVIDIA because a new Cinebench 2026 leak reveals that the chipset produces subpar multi-core results, despite Microsoft’s Surface Laptop Ultra housing the SoC set to the ‘High Performance’ profile. Fortunately, the RTX Spark manages to retain some dignity in the single-core benchmark.

High Performance mode on Surface Laptop Ultra isn’t enabled by default, but hidden configuration allows RTX Spark’s Cortex cores to pull nearly 50W in new test

We’ve previously reported that the altered Cortex-X925 cores running RTX Spark were specifically designed to tackle PC-related workloads, as these tailor-made cores help maintain higher frequencies. Even then, this configuration is slower than Apple’s previous-generation M3 Max in Geekbench 6, a silicon that’s now three years old. Now, thanks to TechPowerUp Forums user “Fouquin,” we learn how the RTX Spark performs in Cinebench 2026.

Related Story SEGA’s “Legendary” Games Are Coming To NVIDIA RTX Spark 30 Years After NV1 Powered The First Virtua Fighter on PC

The single-core and multi-core scores were 540 and 5,771, respectively, indicating that the RTX Spark performs on par with the likes of the M3 Max in single-threaded workloads, but SoCs like the M4 Max outpace it considerably in the multi-threaded category, being 35.6 percent faster than the RTX Spark. Additional details reveal that while the Surface Laptop Ultra is designed to run the chipset between 80W and 95W, there are some limitations.

For instance, as mentioned above, the Surface Laptop Ultra doesn’t have the High Performance mode enabled by default, meaning that some performance will be left off the table when the RTX Spark is pushed through its paces in benchmarks. Fouquin states that he managed to enable the hidden configuration, allowing the RTX Spark to boost to 50W, with the Cortex cores operating at a clock speed of 2.80GHz.

However, keep in mind that Cortex-X925 cores can boost to 4.00GHz under the right cooling conditions, so either it’s Microsoft that has purposely gimped the RTX Spark’s performance through software limitations or the heatsink just isn’t capable enough to dissipate 50W of power. In any case, we’ll learn more about these results when the first commercial laptops featuring the RTX Spark begin shipping.

RTX Spark’s hidden gem continues to be its Blackwell GPU and unified memory

It’s heartbreaking to see that NVIDIA’s first laptop-based SoC hasn’t performed according to expectations, but that’s only on the CPU side. The Blackwell GPU and unified memory offer a pairing that’s capable enough for running AAA games like PRAGMATA and Alan Wake 2 with path tracing enabled while running LLMs with 120B parameters.

We’ve even discussed previously that the RTX Spark has a massive advantage over Apple and Qualcomm on the GPU side, so it’s imperative that both companies race with a purpose to beat their latest rival in this category. After all, those CPU limitations will likely be overcome with a successor next year, severely reducing that performance gap.

News Source: TechPowerUp

Omar Sohail Photo

About the author: Omar Sohail is a reporter and analyst for Wccftech's mobile section, specializing in the technology and business of the mobile industry. His expertise lies in the intricate hardware supply chain, covering developments in semiconductor manufacturing, chip lithography, and camera sensor technology.

Follow Wccftech on Google to get more of our news coverage in your feeds.

Button