Revisiting The SNC3 vs. HEX Mode Performance With Intel Xeon 6 Granite Rapids

Written by Michael Larabel in Memory on 21 October 2025 at 11:00 AM EDT. Page 2 of 4. Add A Comment.

👁 High Performance Conjugate Gradient benchmark with settings of X Y Z: 144 144 144, RT: 60. Default was the fastest.

👁 QuantLib benchmark with settings of Size: S. Default was the fastest.

👁 libxsmm benchmark with settings of M N K: 256. Default was the fastest.

👁 Xcompact3d Incompact3d benchmark with settings of Input: ABL-Atmospheric-Boundary-Layer. Default was the fastest.

👁 OpenRadioss benchmark with settings of Model: Cell Phone Drop Test. Default was the fastest.

👁 OpenRadioss benchmark with settings of Model: Bird Strike on Windshield. Default was the fastest.

👁 RELION benchmark with settings of Test: Basic, Device: CPU. Default was the fastest.

👁 LAMMPS Molecular Dynamics Simulator benchmark with settings of Model: 20k Atoms. Default was the fastest.

👁 DaCapo Benchmark benchmark with settings of Java Test: Tradesoap. Default was the fastest.

👁 DaCapo Benchmark benchmark with settings of Java Test: Tradebeans. Default was the fastest.

The SNC3 mode is the sensible default and for most workloads does deliver the best performance. For those curious about the difference of SNC3 vs. HEX...

👁 Timed Godot Game Engine Compilation benchmark with settings of Time To Compile. Default was the fastest.

👁 Timed Linux Kernel Compilation benchmark with settings of Build: defconfig. Default was the fastest.

👁 Timed Linux Kernel Compilation benchmark with settings of Build: allmodconfig. Default was the fastest.

👁 Timed LLVM Compilation benchmark with settings of Build System: Ninja. Default was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: gpt-oss-20b-Q8_0, Test: Prompt Processing 2048. Default was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: Mistral-7B-Instruct-v0.3-Q8_0, Test: Prompt Processing 2048. Default was the fastest.

👁 BRL-CAD benchmark with settings of VGR Performance Metric. Default was the fastest.

For most workloads sticking to the SNC3 default makes sense but for some other workloads it can be beneficial to change to HEX mode for the one NUMA domain per processor. Let's take a look at some of the workloads where HEX mode ends up better for two Intel Xeon 6980P flagship processors.

Add A Comment - Next Page

Page: 1 2 3 4 Next Page

URL: https://www.phoronix.com/review/intel-xeon-snc3-hex-benchmarks/2

⇱ Revisiting The SNC3 vs. HEX Mode Performance With Intel Xeon 6 Granite Rapids - Phoronix

Revisiting The SNC3 vs. HEX Mode Performance With Intel Xeon 6 Granite Rapids