One of the first performance results for NVIDIA’s Arm-based RTX Spark platform has emerged. According to the results shared in the Clang test, RTX Spark passed the Apple M5 by a significant margin and received a result close to the 15-core M5 Pro. NVIDIA RTX Spark passed the Apple M5 by 54 percent in the first test. In the Clang test shared by @lafaiel on X, RTX Spark received 43,149 points. In the same test, the 10-core Apple M5 remained at 27,996 points.
According to these results, RTX Spark achieved a 54.13 percent higher score against M5. A similar picture emerged on the compilation speed side. RTX Spark reached 212.5 Klines per second. Apple M5 saw 137.9 Klines per second in the same test. The difference was recorded as 54.10 percent on the compilation speed side. RTX Spark comes with a combination of 20-core Grace CPU and Blackwell GPU. The structure used by NVIDIA on this platform gave a strong result, especially in multi-threaded developer workloads, thanks to its high core count.
The Clang test is among the scenarios that measure code compilation performance in languages such as C, C and Objective-C. Apple’s M5 Pro chips were also included in the comparison. The 15-core M5 Pro scored 46,374 points and achieved a compilation speed of 228.4 Klines per second. RTX Spark scored 6.95 percent lower against this version. The compilation speed difference was listed as 6.96 percent. The 18-core M5 Pro was at the top of the list with 55,165 points.
In this model, the compilation speed was measured as 271.7 Klines per second. RTX Spark scored 21.78 percent lower against the 18-core M5 Pro. The compilation speed difference remained at 21.79 percent. In the test results, Intel Core Ultra 9 285HX was also slightly ahead of RTX Spark. In the comparison reported by Wccftech, it was emphasized that Intel’s mobile processor has a 24-core structure and higher power consumption.
AMD Ryzen AI Max 395, on the other hand, fell behind RTX Spark in the same test. The RTX Spark platform is positioned as one of NVIDIA’s new steps in the Arm-based PC market. The platform includes a 20-core Arm processor, Blackwell architecture GPU, 128 GB LPDDR5X combined memory and 6,144 CUDA cores. NVIDIA’s DGX Spark documents also contain AI performance information for the system, up to 1 PFLOP at FP4 precision. On the Apple side, M5 Pro comes with 15-core and 18-core CPU options.
The 15-core version has 5 super cores and 10 performance cores. The 18-core version includes 6 super cores and 12 performance cores. Apple’s technical documentation lists 307 GB/s memory bandwidth for the M5 Pro. The shared results are based on a single Clang test. Different tests are needed for RTX Spark’s performance in gaming, artificial intelligence, content production, emulation and Windows on Arm application compatibility.
NVIDIA’s new platform will be used on new Windows devices prepared especially for developer workloads and native artificial intelligence applications.
Comments
You can write your views about this story. Comments may be moderated according to site settings.