Nvidia RTX Spark N1X beats AMD and Intel in Geekbench

Nvidia’s RTX Spark N1X has posted CPU results that put its 20-core version ahead of AMD’s Ryzen AI Max+ 395 and Intel’s Core Ultra X9 388H in multi-core Geekbench 7, although it trails both the Intel chip and Apple’s M5 Max in single-core performance.

The platform appeared in the Geekbench 7 database in two configurations, both running Windows 11 on Arm with 64GB of memory.

RTX Spark N1X benchmark results

The higher-end N1X uses a 20-core CPU arranged as two 10-core clusters, with a base frequency of 4.0GHz. Its integrated Blackwell GPU includes 6,144 CUDA cores. The system scored 2,570 points in single-core and 23,126 points in multi-core Geekbench 7.

Recommended reading

DJI Osmo 360 II launch set for August 13

The second configuration has 18 cores in a 10+8 layout and runs at 3.9GHz. It uses a smaller Blackwell GPU with 5,120 CUDA cores, while retaining 64GB of memory. Its scores were 2,541 single-core and 21,776 multi-core.

Those results add CPU evidence to Nvidia’s broader RTX Spark platform push, which includes a custom Arm-based PC platform alongside Blackwell graphics.

How it compares with AMD, Intel and Apple

AMD’s Ryzen AI Max+ 395 averages about 2,429 single-core and 20,609 multi-core points in Geekbench 7. The 20-core N1X is therefore approximately 6% faster in single-core and 12% faster in multi-core. The AMD processor has 16 Zen 5 cores, an integrated Radeon 8060S GPU with 40 RDNA 3.5 compute units, and support for up to 128GB of LPDDR5X memory.

Intel’s Core Ultra X9 388H, a 16-core Panther Lake mobile processor with Arc B390 graphics, averages about 2,767 single-core and 19,304 multi-core points. Nvidia’s 20-core system is roughly 7% slower in single-core, but approximately 20% faster with all cores active.

Apple remains well ahead in these CPU tests. Based on 19 Geekbench 7 results, the 18-core M5 Max averages around 3,766 single-core and 35,527 multi-core points. That leaves the 20-core N1X about 32% behind in single-core and 35% behind in multi-core performance.

The comparison is incomplete in one important respect: Geekbench primarily measures CPU performance, so it says little about the N1X’s Blackwell GPU. Its 6,144 CUDA cores could provide a substantial advantage in AI workloads, but these results do not measure that capability.

Nvidia also has not disclosed pricing or a release date in the reported results. On the evidence available, the N1X looks competitive against AMD and Intel for sustained multi-core workloads, but it is not a general performance leader—and its most distinctive GPU advantage remains untested here.