Against the backdrop of growing demand for personal devices with powerful local artificial intelligence capabilities, Nvidia continues to expand its presence in the processor market. In August 2026, test results for the new Nvidia RTX Spark N1X platform were discovered in the Geekbench 7 benchmark database. The appearance of the chip in two configurations — 18-core and 20-core — confirms the manufacturer's ambitious plans to capture the niche of high-performance laptops and workstations.
Technical specifications and chip configurations
According to data obtained from the benchmark, the RTX Spark N1X platform is based on the latest Blackwell architecture. The flagship version of the chip features 20 cores organized into two clusters of 10 cores each, with a base frequency of 4.0 GHz. The graphics processor integrated into the die boasts an impressive array of 6144 CUDA cores. The second configuration, likely aimed at more balanced devices, received 18 cores (10+8 scheme) and a frequency of 3.9 GHz. In this version, the GPU is reduced to 5120 CUDA cores. Both versions were tested in a system with 64 GB of RAM running Windows 11 on ARM.
Performance results in Geekbench 7
In Geekbench 7 synthetic tests, the 20-core version of the N1X demonstrated impressive results: 2570 points in single-threaded mode and 23,126 points in multi-threaded mode. The 18-core modification showed slightly more modest figures — 2541 points in single-core and 21,776 points in multi-core tests. These figures indicate that Nvidia has managed to create a competitive processor capable of handling heavy computational tasks traditionally performed by desktop solutions.
Comparison with competitors: AMD and Intel
Particularly interesting is the analysis of N1X performance compared to current market leaders. In a showdown with the AMD Ryzen AI Max+ 395 (16 Zen 5 cores), which scores around 2429 and 20,609 points respectively, the Nvidia chip demonstrates a 6% advantage in single-threaded mode and a full 12% in multi-threaded mode. The situation changes when comparing with the Intel Core Ultra X9 388H (Panther Lake architecture). Here, the 20-core N1X lags by approximately 7% in single-threaded performance (2767 points for Intel), but confidently wins by about 20% under full load on all cores, showing that the bet on a large number of cores in the Blackwell architecture is justified.
Prospects and limitations
Despite its successes, Nvidia is not yet ready to challenge the leader of the mobile chip market — Apple. The 18-core Apple M5 Max, averaging 3766 and 35,527 points, remains unrivaled, outpacing the N1X by 32% and 35% respectively. However, it is important to consider the specifics of the tests: Geekbench focuses on the CPU, whereas the main advantage of the N1X is the powerful Blackwell GPU with thousands of CUDA cores. It is precisely in generative artificial intelligence tasks and compute acceleration that the new Nvidia chip may gain a decisive advantage that standard benchmarks cannot reflect.
