RTX Spark Closes the Gap with Apple M4 Max in Performance Tests
NVIDIA’s new ARM-based RTX Spark processor significantly closed the performance gap with the Apple M4 Max in Geekbench 7 tests. Here are the details.
NVIDIA’s new ARM-based chip, RTX Spark, has shown significant improvement in recent performance tests conducted on the Geekbench 7 platform. Previously failing to deliver the expected results against Apple’s M3 Max processor, this new SoC (system-on-a-chip), with its 20-core configuration, has clearly narrowed the multi-core performance gap with Apple’s new 14-core M4 Max processor. Developed in collaboration between NVIDIA and MediaTek, this hardware aims to secure a strong position, particularly in the portable computer market.
- The RTX Spark 20-core processor scored only 11 percent lower than the Apple M4 Max in multi-core tests.
- According to the Geekbench 7 database, the 20-core model is approximately 6.2 percent faster in multi-core performance than the 18-core version.
- High-end RTX Spark models house 6,144 CUDA cores based on the Blackwell architecture.
RTX Spark Models Differ in Performance Tests
The results from the Geekbench 7 database confirm the existence of two different models of RTX Spark with different core configurations. These systems, supported by 64GB of combined memory, were tested in a “Balanced” power profile. In the measurements taken, the 20-core processor scored 2,570 single-core and 23,126 multi-core points, while the 18-core version scored 2,541 and 21,776 points respectively.
Although the new RTX Spark processors struggle against the single-core performance of the Apple Silicon family, they are intensifying the competition with their multi-core capacity.
Apple’s M4 Max processor continues to maintain its lead with 3,404 single-core and 25,957 multi-core scores in the tests. The most powerful version of the RTX Spark shows a graph that is 11 percent slower in multi-core performance and approximately 24.5 percent slower in single-core performance than its competitor. It is predicted that this difference may close further over time with software optimizations and driver updates.
NVIDIA Blackwell Architecture Defines Graphics Performance
Beyond processor performance, the greatest strength of the RTX Spark comes from its integrated GPU unit using the Blackwell architecture. Information obtained from Windows 11 ARM64 drivers reveals that the 18-core model has 5,120 CUDA cores, while the 20-core model has 6,144. While these figures are similar to desktop-class RTX 5070 graphics cards, it remains unclear how successful the final game and process performance will be in real-world tests.
Prototype devices, still in the development stage, may produce fluctuating results due to drivers not running at full capacity. This move by NVIDIA seems poised to take the ARM-based processor competition in the PC market to a new level. Users are eagerly awaiting to see if the RTX Spark can dethrone Apple’s M-series processors in real-world use cases.
Do you think NVIDIA’s ARM-based RTX Spark processors can change the dynamics of the laptop market by offering desktop-class performance against Apple’s M-series chips? Share your comments with us.
