NUMA Regression 5.18 Benchmarks by Michael Larabel reproducing regression from https://www.phoronix.com/scan.php?page=news_item&px=Linux-518-Stress-NUMA-Goes-Boom on the system in 2022-05-28 state across v5.17 to v5.18, still present on Linux Gity, and testing the Linux Git with patch from https://lore.kernel.org/lkml/[email protected] /T/#m172e8b7b57fdbed3aceca6fc667cdbf85e2af49b. Patch seems to resolve the stress-ng NUMA performance.
HTML result view exported from: https://openbenchmarking.org/result/2205284-PTS-NUMAREGR17&rdt .
NUMA Regression 5.18 Processor Motherboard Chipset Memory Disk Graphics Audio Monitor Network OS Kernel Desktop Display Server OpenGL Vulkan Compiler File-System Screen Resolution v5.17 v5.18 20220528 Git 20220528 Git + sched-rcu-exped patch AMD Ryzen 9 5950X 16-Core @ 3.40GHz (16 Cores / 32 Threads) ASUS ROG CROSSHAIR VIII HERO (WI-FI) (4006 BIOS) AMD Starship/Matisse 32GB 1000GB Sabrent Rocket 4.0 Plus AMD Radeon RX 6800 16GB (2475/1000MHz) AMD Navi 21 HDMI Audio ASUS MG28U Realtek RTL8125 2.5GbE + Intel I211 + Intel Wi-Fi 6 AX200 Ubuntu 22.04 5.17.0-051700-generic (x86_64) GNOME Shell 42.0 X Server + Wayland 4.6 Mesa 22.0.1 (LLVM 13.0.1 DRM 3.44) 1.2.204 GCC 11.2.0 ext4 3840x2160 5.18.0-051800-generic (x86_64) 4.6 Mesa 22.0.1 (LLVM 13.0.1 DRM 3.46) 5.18.0-051800daily20220528-generic (x86_64) 5.18.0-sched-rcu-exped (x86_64) OpenBenchmarking.org Kernel Details - Transparent Huge Pages: madvise Compiler Details - --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none=/build/gcc-11-gBFGDP/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-gBFGDP/gcc-11-11.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v Processor Details - Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0xa201016 Security Details - itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines IBPB: conditional IBRS_FW STIBP: always-on RSB filling + srbds: Not affected + tsx_async_abort: Not affected
NUMA Regression 5.18 stress-ng: NUMA Difference Standard Error Standard Deviation v5.17 v5.18 20220528 Git 20220528 Git + sched-rcu-exped patch 412.88 8.37x 2.70 1.13% 49.33 1.00x 0.35 2.45% 49.66 1.01x 0.44 3.42% 468.81 9.50x 1.26 0.47% OpenBenchmarking.org
Stress-NG Test: NUMA OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: NUMA v5.17 v5.18 20220528 Git 20220528 Git + sched-rcu-exped patch 100 200 300 400 500 SE +/- 2.70, N = 3 SE +/- 0.35, N = 12 SE +/- 0.44, N = 15 SE +/- 1.26, N = 3 412.88 49.33 49.66 468.81 1. (CC) gcc options: -O2 -std=gnu99 -lm -laio -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lz -pthread
Phoronix Test Suite v10.8.5