Qualcomm Snapdragon X Elite is about to make landfall in the ultraportable notebook segment, powering a new wave of Windows 11 devices powered by Arm, capable of running even legacy Windows applications. The Snapdragon X Elite SoC in particular has been designed to rival the Apple M3 chip powering the 2024 MacBook Air, and some of the "entry-level" variants of the 2023 MacBook Pros. These chips threaten the 15 W U-segment and even 28 W P-segment of x86-64 processors from Intel, such as the Core Ultra "Meteor Lake," and Ryzen 8040 "Hawk Point." Erdi Özüağ, prominent tech journalist from Türkiye, has access to a Qualcomm-reference notebook powered by the Snapdragon X Elite X1E80100 28 W SoC. He compared its performance to an off-the-shelf notebook powered by a 28 W Intel Core Ultra 7 155H "Meteor Lake" processor.

There are three tests that highlight the performance of the key components of the SoCs—CPU, iGPU, and NPU. A Microsoft Visual Studio code compile test sees the Snapdragon X Elite with its 12-core Oryon CPU finish the test in 37 seconds; compared to 54 seconds by the Core Ultra 7 155H with its 6P+8E+2LP CPU. In the 3DMark test, the Adreno 750 iGPU posts identical performance numbers to the Arc Graphics Xe-LPG of the 155H. Where the Snapdragon X Elite dominates the Intel chip is AI inferencing. The UL Procyon test sees the 45 TOPS NPU of the Snapdragon X Elite score 1720 points compared to 476 points by the 10 TOPS AI Boost NPU of the Core Ultra. The Intel machine is using OpenVINO, while the Snapdragon is using Qualcomm SNPE SDK for the test. Don't forget to check out the video review by Erdi Özüağ in the source link below.

Qualcomm Snapdragon 8cx Gen 3 is a high performance Arm SoC designed to compete with Apple M3, with Windows 11 thin and light notebooks and Chromebooks being its main target devices. Microsoft pins a lot of hope in chips such as the Snapdragon 8cx series as they offer comparable performance and battery life to the current crop of M3 MacBooks. A lot of water has flown under the bridge since Windows RT, and the latest crop of Windows 11 for Arm has a much wider PC application support base thanks to official translation layers by Microsoft. CPUID has an Arm64 version of the popular CPU-Z utility, which correctly detects all the specs of the Snapdragon 8cx, but more importantly, has a Bench tab that can test the single- and multithreaded performance of the CPU.

A Chinese tech enthusiast wasted no time in putting the Snapdragon 8cx through this CPU-Z internal benchmark, and found surprisingly good performance numbers. The single-threaded bench, which loads one of chip's four Arm Cortex-X1C P-cores, registers a score of 543.7 points. This is roughly comparable to that of the AMD "Zen 2" or Intel "Comet Lake" x86-64 core. The multithreaded test, which saturates all four P-cores, and all four Cortex-A78C E-cores, springs up 3479.7 points, which again compares to entry/mainstream x86-64 processors from AMD or Intel. Not impressed? How about the fact that the Snapdragon 8cx Gen 3 is a 7 W chip that idles under 2 W for the most part, and can make do with passive cooling, posting scores comparable to 35 W x86 chips that need active cooling?

AMD's upcoming family of Ryzen 9000 series of processors on the AM5 platform will carry a new silicon SKU under the hood—Zen 5. The latest revision of AMD's x86-64 microarchitecture will feature a few interesting improvements over its current Zen 4 that it is replacing, targeting the rumored 10-15% IPC improvement. Thanks to the latest set of patches for GNU Compiler Collection (GCC), we have the patch set that proposes changes taking place with "znver5" enablement. One of the most interesting additions to the Zen 5 over the previous Zen 4 is the expansion of the AVX instruction set, mainly new AVX and AVX-512 instructions: AVX-VNNI, MOVDIRI, MOVDIR64B, AVX512VP2INTERSECT, and PREFETCHI.

AVX-VNNI is a 256-bit vector version of the AVX-512 VNNI instruction set that accelerates neural network inferencing workloads. AVX-VNNI delivers the same VNNI instruction set for CPUs that support 256-bit vectors but lack full 512-bit AVX-512 capabilities. AVX-VNNI effectively extends useful VNNI instructions for AI acceleration down to 256-bit vectors, making the technology more efficient. While narrow in scope (no opmasking and extra vector register access compared to AVX-512 VNNI), AVX-VNNI is crucial in spreading VNNI inferencing speedups to real-world CPUs and applications. The new AVX-512 VP2INTERSECT instruction is also making it in Zen 5, as noted above, which has been present only in Intel Tiger Lake processor generation, and is now considered deprecated for Intel SKUs. We don't know the rationale behind this inclusion, but AMD sure had a use case for it.