NVIDIA N1X SoC Boasts 6144 Blackwell Cores, Matching GeForce RTX 5070 GPU Cores and Outperforming Modern iGPUs

Recent benchmarks from Geekbench have unveiled early performance metrics for NVIDIA’s upcoming N1X System-on-Chip (SoC), which boasts a GPU architecture featuring an impressive core count comparable to the RTX 5070.

NVIDIA N1X SoC Features “Blackwell”GPU with RTX 5070 Core Count, Yet Performance Challenges Remain

The latest entry for NVIDIA’s N1X has appeared in Geekbench’s OpenCL database, providing the first glimpse into its GPU capabilities, previously overshadowed by its CPU specifications. The details have generated significant interest among tech enthusiasts.

According to the Geekbench listing, the N1X is equipped with a Blackwell GPU that matches the RTX 5070 in core count, featuring 6144 cores distributed across 48 Streaming Multiprocessors (SMs).The RTX 5070 utilizes the GB205 chip; however, it’s uncertain whether the N1X employs the same silicon or a distinct modification. The RTX 5070 also encompasses 192 Texture Mapping Units (TMUs), 80 Render Output Units (ROPs), 48 Ray Tracing (RT) Cores, and 192 Tensor Cores, along with 12 GB of GDDR7 memory that offers a bandwidth of up to 672 GB/s, achieving clock speeds of up to 2512 MHz within a 250W thermal design power (TDP).

Geekbench OpenCL performance for NVIDIA N1X — Image Source: Geekbench

The performance metrics displayed in Geekbench reveal that the N1X’s GPU operates at a modest frequency of just over 1.05 GHz, significantly lower than its RTX 5070 counterpart. Furthermore, this chip does not utilize dedicated memory; instead, it integrates LPDDR5X standard memory shared across the SoC. Anticipated TDP for the N1X is estimated between 100W and 120W, indicating that even under maximum load scenarios, it may achieve around 100W for graphics-intensive or AI tasks, which is still substantially less than the RTX 5070 by approximately 150W.

The preliminary benchmarking results indicate that the N1X has scored 46, 361 points, placing it ahead of other contemporary integrated GPUs but significantly trailing the RTX 5070. This is largely due to the power limitations and the fact that this is still an early engineering sample. NVIDIA is expected to enhance the performance metrics of the N1X significantly before the official launch.

Speculative reports suggest that the N1X SoC could be launched in 2026, with predictions varying between the first and second halves of the year. This launch is poised to intensify competition within the SoC market, positioning NVIDIA to pose a strong challenge against established players in the x86 and ARM ecosystems, including Intel, AMD, Qualcomm, and Apple.

NVIDIA N1X SoC Specifications (Preliminary):

Name	NVIDIA N1X
Process Technology	TSMC N3
Architecture	Grace + Blackwell
CPU Cores	20 (10 Cortex-X925 + 10 Cortex-A725)
GPU Cores	6144 Cores (48 SMs)
Memory	LPDDR5X (Up to 128 GB)
Memory Speed	8533 MT/s
TDP	100-120W (Peak)
Expected Launch	2026