MMIO Stale Data Mitigation Benchmarks Tests for a future article. Intel Core i9-11900K testing with a ASUS ROG MAXIMUS XIII HERO (1402 BIOS) and ASUS Intel RKL GT1 31GB on Ubuntu 21.10 via the Phoronix Test Suite.
HTML result view exported from: https://openbenchmarking.org/result/2206160-PTS-MMIOSTAL02&rdt&grt&export=txt .
MMIO Stale Data Mitigation Benchmarks Processor Motherboard Chipset Memory Disk Graphics Audio Monitor Network OS Kernel Desktop Display Server OpenGL Vulkan Compiler File-System Screen Resolution Clear buffers, SMT vulnerable mmio_stale_data=off Intel Core i9-11900K @ 5.10GHz (8 Cores / 16 Threads) ASUS ROG MAXIMUS XIII HERO (1402 BIOS) Intel Tiger Lake-H 32GB 2000GB Corsair Force MP600 ASUS Intel RocketLake-S GT1 [UHD 750] (1300MHz) Intel Tiger Lake-H HD Audio MX279 2 x Intel I225-V + Intel Wi-Fi 6 AX210/AX211/AX411 Ubuntu 21.10 5.19.0-rc2-mmio-stale-data-patched (x86_64) GNOME Shell 40.5 X Server 1.20.13 4.6 Mesa 22.2.0-devel (git-fd27d51 2022-06-15 impish-oibaf-ppa) 1.3.217 GCC 11.2.0 ext4 1920x1080 ASUS Intel RKL GT1 31GB (1300MHz) 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-ZPT0kp/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-ZPT0kp/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 Disk Details - NONE / errors=remount-ro,relatime,rw / Block Size: 4096 Processor Details - Scaling Governor: intel_pstate powersave (EPP: balance_performance) - CPU Microcode: 0x53 - Thermald 2.4.6 Java Details - OpenJDK Runtime Environment (build 11.0.15+10-Ubuntu-0ubuntu0.21.10.1) Python Details - Python 3.9.7 Security Details - Clear buffers, SMT vulnerable: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Mitigation of Clear buffers; SMT vulnerable + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced IBRS IBPB: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected - mmio_stale_data=off: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Vulnerable + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced IBRS IBPB: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected
MMIO Stale Data Mitigation Benchmarks apache: 1 apache: 20 apache: 100 apache: 200 blender: BMW27 - CPU-Only blender: Fishy Cat - CPU-Only blender: Pabellon Barcelona - CPU-Only compilebench: Compile compilebench: Initial Create compilebench: Read Compiled Tree dacapobench: H2 dacapobench: Jython dacapobench: Tradesoap dacapobench: Tradebeans darktable: Boat - CPU-only darktable: Masskrug - CPU-only darktable: Server Rack - CPU-only darktable: Server Room - CPU-only rocksdb: Rand Read rocksdb: Update Rand rocksdb: Read While Writing rocksdb: Read Rand Write Rand gimp: resize gimp: rotate gimp: auto-levels gimp: unsharp-mask octave-benchmark: ior: 4MB - Default Test Directory memtier-benchmark: Redis npb: BT.C npb: EP.C npb: IS.D npb: SP.C nettle: aes256 nettle: chacha nettle: sha512 nettle: poly1305-aes nginx: 1 nginx: 20 nginx: 100 nginx: 200 onnx: GPT-2 - CPU - Standard onnx: yolov4 - CPU - Standard onnx: bertsquad-12 - CPU - Standard onnx: fcn-resnet101-11 - CPU - Standard onnx: ArcFace ResNet-100 - CPU - Standard onnx: super-resolution-10 - CPU - Standard pgbench: 100 - 50 - Read Only pgbench: 100 - 50 - Read Only - Average Latency pgbench: 100 - 50 - Read Write pgbench: 100 - 50 - Read Write - Average Latency postmark: Disk Transaction Performance rawtherapee: Total Benchmark Time redis: GET redis: SET renaissance: Finagle HTTP Requests renaissance: In-Memory Database Shootout sockperf: Throughput sockperf: Latency Ping Pong sockperf: Latency Under Load sqlite-speedtest: Timed Time - Size 1,000 stress-ng: MMAP stress-ng: NUMA stress-ng: Futex stress-ng: MEMFD stress-ng: Atomic stress-ng: Crypto stress-ng: Malloc stress-ng: Forking stress-ng: IO_uring stress-ng: SENDFILE stress-ng: CPU Cache stress-ng: CPU Stress stress-ng: Semaphores stress-ng: Matrix Math stress-ng: Vector Math stress-ng: x86_64 RdRand stress-ng: Memory Copying stress-ng: Socket Activity stress-ng: Context Switching stress-ng: Glibc C String Functions stress-ng: Glibc Qsort Data Sorting stress-ng: System V Message Passing svt-av1: Preset 10 - Bosphorus 4K svt-av1: Preset 12 - Bosphorus 4K svt-av1: Preset 10 - Bosphorus 1080p svt-av1: Preset 12 - Bosphorus 1080p svt-hevc: 7 - Bosphorus 4K svt-hevc: 10 - Bosphorus 4K svt-hevc: 7 - Bosphorus 1080p svt-hevc: 10 - Bosphorus 1080p svt-vp9: VMAF Optimized - Bosphorus 4K svt-vp9: VMAF Optimized - Bosphorus 1080p svt-vp9: PSNR/SSIM Optimized - Bosphorus 4K svt-vp9: PSNR/SSIM Optimized - Bosphorus 1080p tensorflow-lite: SqueezeNet tensorflow-lite: Inception V4 tensorflow-lite: NASNet Mobile tensorflow-lite: Mobilenet Float tensorflow-lite: Mobilenet Quant tensorflow-lite: Inception ResNet V2 build-linux-kernel: defconfig build-llvm: Unix Makefiles Clear buffers, SMT vulnerable mmio_stale_data=off 20094.59 92526.25 106471.07 89131.24 130.68 172.36 451.46 2573.57 771.55 4673.46 2588 2994 2991 2035 4.069 4.444 0.217 3.343 53591223 669208 2242464 2094059 5.761 8.387 8.790 10.613 5.181 1774.60 3068855.25 23989.46 1663.96 999.60 6620.49 12018.99 1466.73 721.60 4466.63 111694.60 501759.2 501994.94 497111.33 6963 532 995 103 2071 7126 372478 0.135 31450 1.590 8524 52.139 4502788.7 3212061.33 2734.8 2324.2 839493 2.350 14.365 39.616 217.77 24.61 2542009.31 741.66 339665.98 13464.65 7727010.12 70423.80 25726.27 243060.73 209.47 23175.10 1718628.66 57560.79 58932.06 77956.30 1613.78 9216.66 12430837.52 1716938.57 185.31 11271215.82 80.701 109.186 231.466 425.403 45.98 85.44 140.77 282.53 57.18 193.67 62.55 201.31 1813.30 26947.2 7077.56 1432.14 2756.90 26868.1 78.918 615.362 19834.64 95187.42 107861.66 89743.41 130.65 172.20 451.67 2635.02 762.22 4702.80 2571 2995 3002 2068 4.095 4.444 0.217 3.349 53660896 670521 2233150 2078180 5.792 8.417 8.726 10.650 5.209 1604.93 3017015.03 24007.45 1614.26 1001.31 6602.49 12158.15 1501.73 739.88 4465.80 111539.67 506778.75 510403.10 506101.81 6961 509 1020 103 1963 6997 372390 0.134 31491 1.588 8723 52.576 4042373.31 3183488.08 2699.3 2308.0 826045 2.380 19.160 39.971 217.49 24.64 2532900.28 742.02 352590.37 13419.93 7783046.99 70631.43 25255.71 242573.02 214.32 22936.55 1716750.63 57656.07 58931.54 77931.96 1614.44 9193.96 13841930.87 1717173.49 185.14 11307266.30 81.030 108.802 231.753 429.784 45.89 85.61 140.47 282.71 56.93 192.98 62.21 199.83 1816.11 26719.8 7082.13 1437.31 2716.21 27460.1 79.057 617.889 OpenBenchmarking.org
Apache HTTP Server Concurrent Requests: 1 OpenBenchmarking.org Requests Per Second, More Is Better Apache HTTP Server 2.4.48 Concurrent Requests: 1 Clear buffers, SMT vulnerable mmio_stale_data=off 4K 8K 12K 16K 20K SE +/- 20.97, N = 3 SE +/- 94.85, N = 3 20094.59 19834.64 1. (CC) gcc options: -shared -fPIC -O2
Apache HTTP Server Concurrent Requests: 20 OpenBenchmarking.org Requests Per Second, More Is Better Apache HTTP Server 2.4.48 Concurrent Requests: 20 Clear buffers, SMT vulnerable mmio_stale_data=off 20K 40K 60K 80K 100K SE +/- 339.98, N = 3 SE +/- 274.39, N = 3 92526.25 95187.42 1. (CC) gcc options: -shared -fPIC -O2
Apache HTTP Server Concurrent Requests: 100 OpenBenchmarking.org Requests Per Second, More Is Better Apache HTTP Server 2.4.48 Concurrent Requests: 100 Clear buffers, SMT vulnerable mmio_stale_data=off 20K 40K 60K 80K 100K SE +/- 105.45, N = 3 SE +/- 294.96, N = 3 106471.07 107861.66 1. (CC) gcc options: -shared -fPIC -O2
Apache HTTP Server Concurrent Requests: 200 OpenBenchmarking.org Requests Per Second, More Is Better Apache HTTP Server 2.4.48 Concurrent Requests: 200 Clear buffers, SMT vulnerable mmio_stale_data=off 20K 40K 60K 80K 100K SE +/- 325.91, N = 3 SE +/- 196.56, N = 3 89131.24 89743.41 1. (CC) gcc options: -shared -fPIC -O2
Blender Blend File: BMW27 - Compute: CPU-Only OpenBenchmarking.org Seconds, Fewer Is Better Blender 3.2 Blend File: BMW27 - Compute: CPU-Only Clear buffers, SMT vulnerable mmio_stale_data=off 30 60 90 120 150 SE +/- 0.09, N = 3 SE +/- 0.03, N = 3 130.68 130.65
Blender Blend File: Fishy Cat - Compute: CPU-Only OpenBenchmarking.org Seconds, Fewer Is Better Blender 3.2 Blend File: Fishy Cat - Compute: CPU-Only Clear buffers, SMT vulnerable mmio_stale_data=off 40 80 120 160 200 SE +/- 0.06, N = 3 SE +/- 0.06, N = 3 172.36 172.20
Blender Blend File: Pabellon Barcelona - Compute: CPU-Only OpenBenchmarking.org Seconds, Fewer Is Better Blender 3.2 Blend File: Pabellon Barcelona - Compute: CPU-Only Clear buffers, SMT vulnerable mmio_stale_data=off 100 200 300 400 500 SE +/- 0.43, N = 3 SE +/- 0.49, N = 3 451.46 451.67
Compile Bench Test: Compile OpenBenchmarking.org MB/s, More Is Better Compile Bench 0.6 Test: Compile Clear buffers, SMT vulnerable mmio_stale_data=off 600 1200 1800 2400 3000 SE +/- 31.35, N = 15 SE +/- 25.45, N = 6 2573.57 2635.02
Compile Bench Test: Initial Create OpenBenchmarking.org MB/s, More Is Better Compile Bench 0.6 Test: Initial Create Clear buffers, SMT vulnerable mmio_stale_data=off 170 340 510 680 850 SE +/- 4.90, N = 3 SE +/- 11.46, N = 3 771.55 762.22
Compile Bench Test: Read Compiled Tree OpenBenchmarking.org MB/s, More Is Better Compile Bench 0.6 Test: Read Compiled Tree Clear buffers, SMT vulnerable mmio_stale_data=off 1000 2000 3000 4000 5000 SE +/- 0.00, N = 3 SE +/- 29.34, N = 3 4673.46 4702.80
DaCapo Benchmark Java Test: H2 OpenBenchmarking.org msec, Fewer Is Better DaCapo Benchmark 9.12-MR1 Java Test: H2 Clear buffers, SMT vulnerable mmio_stale_data=off 600 1200 1800 2400 3000 SE +/- 46.91, N = 20 SE +/- 65.73, N = 20 2588 2571
DaCapo Benchmark Java Test: Jython OpenBenchmarking.org msec, Fewer Is Better DaCapo Benchmark 9.12-MR1 Java Test: Jython Clear buffers, SMT vulnerable mmio_stale_data=off 600 1200 1800 2400 3000 SE +/- 9.16, N = 4 SE +/- 20.30, N = 4 2994 2995
DaCapo Benchmark Java Test: Tradesoap OpenBenchmarking.org msec, Fewer Is Better DaCapo Benchmark 9.12-MR1 Java Test: Tradesoap Clear buffers, SMT vulnerable mmio_stale_data=off 600 1200 1800 2400 3000 SE +/- 17.52, N = 3 SE +/- 22.55, N = 20 2991 3002
DaCapo Benchmark Java Test: Tradebeans OpenBenchmarking.org msec, Fewer Is Better DaCapo Benchmark 9.12-MR1 Java Test: Tradebeans Clear buffers, SMT vulnerable mmio_stale_data=off 400 800 1200 1600 2000 SE +/- 22.05, N = 20 SE +/- 21.71, N = 20 2035 2068
Darktable Test: Boat - Acceleration: CPU-only OpenBenchmarking.org Seconds, Fewer Is Better Darktable 3.6.0 Test: Boat - Acceleration: CPU-only Clear buffers, SMT vulnerable mmio_stale_data=off 0.9214 1.8428 2.7642 3.6856 4.607 SE +/- 0.003, N = 3 SE +/- 0.005, N = 3 4.069 4.095
Darktable Test: Masskrug - Acceleration: CPU-only OpenBenchmarking.org Seconds, Fewer Is Better Darktable 3.6.0 Test: Masskrug - Acceleration: CPU-only Clear buffers, SMT vulnerable mmio_stale_data=off 0.9999 1.9998 2.9997 3.9996 4.9995 SE +/- 0.009, N = 3 SE +/- 0.006, N = 3 4.444 4.444
Darktable Test: Server Rack - Acceleration: CPU-only OpenBenchmarking.org Seconds, Fewer Is Better Darktable 3.6.0 Test: Server Rack - Acceleration: CPU-only Clear buffers, SMT vulnerable mmio_stale_data=off 0.0488 0.0976 0.1464 0.1952 0.244 SE +/- 0.001, N = 3 SE +/- 0.000, N = 3 0.217 0.217
Darktable Test: Server Room - Acceleration: CPU-only OpenBenchmarking.org Seconds, Fewer Is Better Darktable 3.6.0 Test: Server Room - Acceleration: CPU-only Clear buffers, SMT vulnerable mmio_stale_data=off 0.7535 1.507 2.2605 3.014 3.7675 SE +/- 0.004, N = 3 SE +/- 0.003, N = 3 3.343 3.349
Facebook RocksDB Test: Random Read OpenBenchmarking.org Op/s, More Is Better Facebook RocksDB 7.0.1 Test: Random Read Clear buffers, SMT vulnerable mmio_stale_data=off 11M 22M 33M 44M 55M SE +/- 381547.95, N = 3 SE +/- 449997.89, N = 3 53591223 53660896 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
Facebook RocksDB Test: Update Random OpenBenchmarking.org Op/s, More Is Better Facebook RocksDB 7.0.1 Test: Update Random Clear buffers, SMT vulnerable mmio_stale_data=off 140K 280K 420K 560K 700K SE +/- 384.85, N = 3 SE +/- 888.10, N = 3 669208 670521 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
Facebook RocksDB Test: Read While Writing OpenBenchmarking.org Op/s, More Is Better Facebook RocksDB 7.0.1 Test: Read While Writing Clear buffers, SMT vulnerable mmio_stale_data=off 500K 1000K 1500K 2000K 2500K SE +/- 9766.85, N = 3 SE +/- 17624.42, N = 3 2242464 2233150 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
Facebook RocksDB Test: Read Random Write Random OpenBenchmarking.org Op/s, More Is Better Facebook RocksDB 7.0.1 Test: Read Random Write Random Clear buffers, SMT vulnerable mmio_stale_data=off 400K 800K 1200K 1600K 2000K SE +/- 4401.84, N = 3 SE +/- 13179.82, N = 3 2094059 2078180 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
GIMP Test: resize OpenBenchmarking.org Seconds, Fewer Is Better GIMP 2.10.24 Test: resize Clear buffers, SMT vulnerable mmio_stale_data=off 1.3032 2.6064 3.9096 5.2128 6.516 SE +/- 0.036, N = 13 SE +/- 0.060, N = 5 5.761 5.792
GIMP Test: rotate OpenBenchmarking.org Seconds, Fewer Is Better GIMP 2.10.24 Test: rotate Clear buffers, SMT vulnerable mmio_stale_data=off 2 4 6 8 10 SE +/- 0.010, N = 3 SE +/- 0.019, N = 3 8.387 8.417
GIMP Test: auto-levels OpenBenchmarking.org Seconds, Fewer Is Better GIMP 2.10.24 Test: auto-levels Clear buffers, SMT vulnerable mmio_stale_data=off 2 4 6 8 10 SE +/- 0.039, N = 3 SE +/- 0.017, N = 3 8.790 8.726
GIMP Test: unsharp-mask OpenBenchmarking.org Seconds, Fewer Is Better GIMP 2.10.24 Test: unsharp-mask Clear buffers, SMT vulnerable mmio_stale_data=off 3 6 9 12 15 SE +/- 0.02, N = 3 SE +/- 0.01, N = 3 10.61 10.65
GNU Octave Benchmark OpenBenchmarking.org Seconds, Fewer Is Better GNU Octave Benchmark 6.2.0 Clear buffers, SMT vulnerable mmio_stale_data=off 1.172 2.344 3.516 4.688 5.86 SE +/- 0.042, N = 9 SE +/- 0.024, N = 5 5.181 5.209
IOR Block Size: 4MB - Disk Target: Default Test Directory OpenBenchmarking.org MB/s, More Is Better IOR 3.3.0 Block Size: 4MB - Disk Target: Default Test Directory Clear buffers, SMT vulnerable mmio_stale_data=off 400 800 1200 1600 2000 SE +/- 12.73, N = 15 SE +/- 24.03, N = 15 1774.60 1604.93 MIN: 392.17 / MAX: 2478.06 MIN: 820.91 / MAX: 2498.45 1. (CC) gcc options: -O2 -lm -lmpi
Memtier_benchmark Protocol: Redis OpenBenchmarking.org Ops/sec, More Is Better Memtier_benchmark 1.3 Protocol: Redis Clear buffers, SMT vulnerable mmio_stale_data=off 700K 1400K 2100K 2800K 3500K SE +/- 27004.28, N = 8 SE +/- 18522.57, N = 3 3068855.25 3017015.03 1. (CXX) g++ options: -O2 -levent_openssl -levent -lcrypto -lssl -lpthread -lz -lpcre
NAS Parallel Benchmarks Test / Class: BT.C OpenBenchmarking.org Total Mop/s, More Is Better NAS Parallel Benchmarks 3.4 Test / Class: BT.C Clear buffers, SMT vulnerable mmio_stale_data=off 5K 10K 15K 20K 25K SE +/- 38.49, N = 3 SE +/- 35.59, N = 3 23989.46 24007.45 1. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.0
NAS Parallel Benchmarks Test / Class: EP.C OpenBenchmarking.org Total Mop/s, More Is Better NAS Parallel Benchmarks 3.4 Test / Class: EP.C Clear buffers, SMT vulnerable mmio_stale_data=off 400 800 1200 1600 2000 SE +/- 23.94, N = 3 SE +/- 27.41, N = 15 1663.96 1614.26 1. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.0
NAS Parallel Benchmarks Test / Class: IS.D OpenBenchmarking.org Total Mop/s, More Is Better NAS Parallel Benchmarks 3.4 Test / Class: IS.D Clear buffers, SMT vulnerable mmio_stale_data=off 200 400 600 800 1000 SE +/- 0.82, N = 3 SE +/- 1.53, N = 3 999.60 1001.31 1. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.0
NAS Parallel Benchmarks Test / Class: SP.C OpenBenchmarking.org Total Mop/s, More Is Better NAS Parallel Benchmarks 3.4 Test / Class: SP.C Clear buffers, SMT vulnerable mmio_stale_data=off 1400 2800 4200 5600 7000 SE +/- 13.70, N = 3 SE +/- 2.93, N = 3 6620.49 6602.49 1. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.0
Nettle Test: aes256 OpenBenchmarking.org Mbyte/s, More Is Better Nettle 3.8 Test: aes256 Clear buffers, SMT vulnerable mmio_stale_data=off 3K 6K 9K 12K 15K SE +/- 128.94, N = 3 SE +/- 13.25, N = 3 12018.99 12158.15 MIN: 7702.46 / MAX: 20580.16 MIN: 7916.5 / MAX: 20586.6 1. (CC) gcc options: -O2 -ggdb3 -lnettle -lm -lcrypto
Nettle Test: chacha OpenBenchmarking.org Mbyte/s, More Is Better Nettle 3.8 Test: chacha Clear buffers, SMT vulnerable mmio_stale_data=off 300 600 900 1200 1500 SE +/- 5.10, N = 3 SE +/- 5.54, N = 3 1466.73 1501.73 MIN: 654.37 / MAX: 4535.7 MIN: 654.5 / MAX: 4611.78 1. (CC) gcc options: -O2 -ggdb3 -lnettle -lm -lcrypto
Nettle Test: sha512 OpenBenchmarking.org Mbyte/s, More Is Better Nettle 3.8 Test: sha512 Clear buffers, SMT vulnerable mmio_stale_data=off 160 320 480 640 800 SE +/- 3.21, N = 3 SE +/- 7.38, N = 3 721.60 739.88 1. (CC) gcc options: -O2 -ggdb3 -lnettle -lm -lcrypto
Nettle Test: poly1305-aes OpenBenchmarking.org Mbyte/s, More Is Better Nettle 3.8 Test: poly1305-aes Clear buffers, SMT vulnerable mmio_stale_data=off 1000 2000 3000 4000 5000 SE +/- 1.30, N = 3 SE +/- 1.15, N = 3 4466.63 4465.80 1. (CC) gcc options: -O2 -ggdb3 -lnettle -lm -lcrypto
nginx Concurrent Requests: 1 OpenBenchmarking.org Requests Per Second, More Is Better nginx 1.21.1 Concurrent Requests: 1 Clear buffers, SMT vulnerable mmio_stale_data=off 20K 40K 60K 80K 100K SE +/- 160.88, N = 3 SE +/- 361.48, N = 3 111694.60 111539.67 1. (CC) gcc options: -lcrypt -lz -O3 -march=native
nginx Concurrent Requests: 20 OpenBenchmarking.org Requests Per Second, More Is Better nginx 1.21.1 Concurrent Requests: 20 Clear buffers, SMT vulnerable mmio_stale_data=off 110K 220K 330K 440K 550K SE +/- 579.06, N = 3 SE +/- 662.40, N = 3 501759.20 506778.75 1. (CC) gcc options: -lcrypt -lz -O3 -march=native
nginx Concurrent Requests: 100 OpenBenchmarking.org Requests Per Second, More Is Better nginx 1.21.1 Concurrent Requests: 100 Clear buffers, SMT vulnerable mmio_stale_data=off 110K 220K 330K 440K 550K SE +/- 1049.35, N = 3 SE +/- 224.71, N = 3 501994.94 510403.10 1. (CC) gcc options: -lcrypt -lz -O3 -march=native
nginx Concurrent Requests: 200 OpenBenchmarking.org Requests Per Second, More Is Better nginx 1.21.1 Concurrent Requests: 200 Clear buffers, SMT vulnerable mmio_stale_data=off 110K 220K 330K 440K 550K SE +/- 446.08, N = 3 SE +/- 251.45, N = 3 497111.33 506101.81 1. (CC) gcc options: -lcrypt -lz -O3 -march=native
ONNX Runtime Model: GPT-2 - Device: CPU - Executor: Standard OpenBenchmarking.org Inferences Per Minute, More Is Better ONNX Runtime 1.11 Model: GPT-2 - Device: CPU - Executor: Standard Clear buffers, SMT vulnerable mmio_stale_data=off 1500 3000 4500 6000 7500 SE +/- 2.20, N = 3 SE +/- 0.29, N = 3 6963 6961 1. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
ONNX Runtime Model: yolov4 - Device: CPU - Executor: Standard OpenBenchmarking.org Inferences Per Minute, More Is Better ONNX Runtime 1.11 Model: yolov4 - Device: CPU - Executor: Standard Clear buffers, SMT vulnerable mmio_stale_data=off 120 240 360 480 600 SE +/- 1.30, N = 3 SE +/- 10.67, N = 12 532 509 1. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
ONNX Runtime Model: bertsquad-12 - Device: CPU - Executor: Standard OpenBenchmarking.org Inferences Per Minute, More Is Better ONNX Runtime 1.11 Model: bertsquad-12 - Device: CPU - Executor: Standard Clear buffers, SMT vulnerable mmio_stale_data=off 200 400 600 800 1000 SE +/- 12.86, N = 12 SE +/- 4.51, N = 3 995 1020 1. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
ONNX Runtime Model: fcn-resnet101-11 - Device: CPU - Executor: Standard OpenBenchmarking.org Inferences Per Minute, More Is Better ONNX Runtime 1.11 Model: fcn-resnet101-11 - Device: CPU - Executor: Standard Clear buffers, SMT vulnerable mmio_stale_data=off 20 40 60 80 100 SE +/- 0.17, N = 3 SE +/- 0.17, N = 3 103 103 1. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
ONNX Runtime Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard OpenBenchmarking.org Inferences Per Minute, More Is Better ONNX Runtime 1.11 Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard Clear buffers, SMT vulnerable mmio_stale_data=off 400 800 1200 1600 2000 SE +/- 3.03, N = 3 SE +/- 44.46, N = 12 2071 1963 1. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
ONNX Runtime Model: super-resolution-10 - Device: CPU - Executor: Standard OpenBenchmarking.org Inferences Per Minute, More Is Better ONNX Runtime 1.11 Model: super-resolution-10 - Device: CPU - Executor: Standard Clear buffers, SMT vulnerable mmio_stale_data=off 1500 3000 4500 6000 7500 SE +/- 77.89, N = 4 SE +/- 75.32, N = 12 7126 6997 1. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
PostgreSQL pgbench Scaling Factor: 100 - Clients: 50 - Mode: Read Only OpenBenchmarking.org TPS, More Is Better PostgreSQL pgbench 14.0 Scaling Factor: 100 - Clients: 50 - Mode: Read Only Clear buffers, SMT vulnerable mmio_stale_data=off 80K 160K 240K 320K 400K SE +/- 1809.62, N = 3 SE +/- 3420.46, N = 3 372478 372390 1. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
PostgreSQL pgbench Scaling Factor: 100 - Clients: 50 - Mode: Read Only - Average Latency OpenBenchmarking.org ms, Fewer Is Better PostgreSQL pgbench 14.0 Scaling Factor: 100 - Clients: 50 - Mode: Read Only - Average Latency Clear buffers, SMT vulnerable mmio_stale_data=off 0.0304 0.0608 0.0912 0.1216 0.152 SE +/- 0.001, N = 3 SE +/- 0.001, N = 3 0.135 0.134 1. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
PostgreSQL pgbench Scaling Factor: 100 - Clients: 50 - Mode: Read Write OpenBenchmarking.org TPS, More Is Better PostgreSQL pgbench 14.0 Scaling Factor: 100 - Clients: 50 - Mode: Read Write Clear buffers, SMT vulnerable mmio_stale_data=off 7K 14K 21K 28K 35K SE +/- 67.52, N = 3 SE +/- 100.77, N = 3 31450 31491 1. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
PostgreSQL pgbench Scaling Factor: 100 - Clients: 50 - Mode: Read Write - Average Latency OpenBenchmarking.org ms, Fewer Is Better PostgreSQL pgbench 14.0 Scaling Factor: 100 - Clients: 50 - Mode: Read Write - Average Latency Clear buffers, SMT vulnerable mmio_stale_data=off 0.3578 0.7156 1.0734 1.4312 1.789 SE +/- 0.003, N = 3 SE +/- 0.005, N = 3 1.590 1.588 1. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
PostMark Disk Transaction Performance OpenBenchmarking.org TPS, More Is Better PostMark 1.51 Disk Transaction Performance Clear buffers, SMT vulnerable mmio_stale_data=off 2K 4K 6K 8K 10K SE +/- 95.67, N = 3 SE +/- 102.67, N = 3 8524 8723 1. (CC) gcc options: -O3
RawTherapee Total Benchmark Time OpenBenchmarking.org Seconds, Fewer Is Better RawTherapee Total Benchmark Time Clear buffers, SMT vulnerable mmio_stale_data=off 12 24 36 48 60 SE +/- 0.02, N = 3 SE +/- 0.03, N = 3 52.14 52.58 1. RawTherapee, version 5.8, command line.
Redis Test: GET OpenBenchmarking.org Requests Per Second, More Is Better Redis 6.0.9 Test: GET Clear buffers, SMT vulnerable mmio_stale_data=off 1000K 2000K 3000K 4000K 5000K SE +/- 24337.32, N = 3 SE +/- 44687.51, N = 4 4502788.70 4042373.31 1. (CXX) g++ options: -MM -MT -g3 -fvisibility=hidden -O3
Redis Test: SET OpenBenchmarking.org Requests Per Second, More Is Better Redis 6.0.9 Test: SET Clear buffers, SMT vulnerable mmio_stale_data=off 700K 1400K 2100K 2800K 3500K SE +/- 7906.99, N = 3 SE +/- 10443.72, N = 3 3212061.33 3183488.08 1. (CXX) g++ options: -MM -MT -g3 -fvisibility=hidden -O3
Renaissance Test: Finagle HTTP Requests OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.14 Test: Finagle HTTP Requests Clear buffers, SMT vulnerable mmio_stale_data=off 600 1200 1800 2400 3000 SE +/- 8.40, N = 3 SE +/- 28.48, N = 3 2734.8 2699.3 MIN: 2523.04 / MAX: 2994.67 MIN: 2469.47 / MAX: 3073.31
Renaissance Test: In-Memory Database Shootout OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.14 Test: In-Memory Database Shootout Clear buffers, SMT vulnerable mmio_stale_data=off 500 1000 1500 2000 2500 SE +/- 12.64, N = 3 SE +/- 7.82, N = 3 2324.2 2308.0 MIN: 2160.37 / MAX: 2574.03 MIN: 2135.4 / MAX: 2564.1
Sockperf Test: Throughput OpenBenchmarking.org Messages Per Second, More Is Better Sockperf 3.7 Test: Throughput Clear buffers, SMT vulnerable mmio_stale_data=off 200K 400K 600K 800K 1000K SE +/- 7370.73, N = 5 SE +/- 6450.15, N = 25 839493 826045 1. (CXX) g++ options: --param -O3 -rdynamic
Sockperf Test: Latency Ping Pong OpenBenchmarking.org usec, Fewer Is Better Sockperf 3.7 Test: Latency Ping Pong Clear buffers, SMT vulnerable mmio_stale_data=off 0.5355 1.071 1.6065 2.142 2.6775 SE +/- 0.015, N = 25 SE +/- 0.025, N = 5 2.350 2.380 1. (CXX) g++ options: --param -O3 -rdynamic
Sockperf Test: Latency Under Load OpenBenchmarking.org usec, Fewer Is Better Sockperf 3.7 Test: Latency Under Load Clear buffers, SMT vulnerable mmio_stale_data=off 5 10 15 20 25 SE +/- 1.49, N = 25 SE +/- 0.10, N = 5 14.37 19.16 1. (CXX) g++ options: --param -O3 -rdynamic
SQLite Speedtest Timed Time - Size 1,000 OpenBenchmarking.org Seconds, Fewer Is Better SQLite Speedtest 3.30 Timed Time - Size 1,000 Clear buffers, SMT vulnerable mmio_stale_data=off 9 18 27 36 45 SE +/- 0.03, N = 3 SE +/- 0.14, N = 3 39.62 39.97 1. (CC) gcc options: -O2 -lz
Stress-NG Test: MMAP OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: MMAP Clear buffers, SMT vulnerable mmio_stale_data=off 50 100 150 200 250 SE +/- 0.48, N = 3 SE +/- 0.85, N = 3 217.77 217.49 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: NUMA OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: NUMA Clear buffers, SMT vulnerable mmio_stale_data=off 6 12 18 24 30 SE +/- 0.14, N = 3 SE +/- 0.17, N = 3 24.61 24.64 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Futex OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Futex Clear buffers, SMT vulnerable mmio_stale_data=off 500K 1000K 1500K 2000K 2500K SE +/- 26117.34, N = 5 SE +/- 49539.09, N = 15 2542009.31 2532900.28 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: MEMFD OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: MEMFD Clear buffers, SMT vulnerable mmio_stale_data=off 160 320 480 640 800 SE +/- 0.76, N = 3 SE +/- 0.59, N = 3 741.66 742.02 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Atomic OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Atomic Clear buffers, SMT vulnerable mmio_stale_data=off 80K 160K 240K 320K 400K SE +/- 4514.52, N = 3 SE +/- 3752.59, N = 3 339665.98 352590.37 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Crypto OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Crypto Clear buffers, SMT vulnerable mmio_stale_data=off 3K 6K 9K 12K 15K SE +/- 47.88, N = 3 SE +/- 29.69, N = 3 13464.65 13419.93 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Malloc OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Malloc Clear buffers, SMT vulnerable mmio_stale_data=off 2M 4M 6M 8M 10M SE +/- 53293.72, N = 3 SE +/- 27882.04, N = 3 7727010.12 7783046.99 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Forking OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Forking Clear buffers, SMT vulnerable mmio_stale_data=off 15K 30K 45K 60K 75K SE +/- 348.18, N = 3 SE +/- 172.90, N = 3 70423.80 70631.43 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: IO_uring OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: IO_uring Clear buffers, SMT vulnerable mmio_stale_data=off 6K 12K 18K 24K 30K SE +/- 64.12, N = 3 SE +/- 255.70, N = 6 25726.27 25255.71 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: SENDFILE OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: SENDFILE Clear buffers, SMT vulnerable mmio_stale_data=off 50K 100K 150K 200K 250K SE +/- 1107.84, N = 3 SE +/- 888.45, N = 3 243060.73 242573.02 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: CPU Cache OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: CPU Cache Clear buffers, SMT vulnerable mmio_stale_data=off 50 100 150 200 250 SE +/- 2.51, N = 15 SE +/- 3.07, N = 3 209.47 214.32 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: CPU Stress OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: CPU Stress Clear buffers, SMT vulnerable mmio_stale_data=off 5K 10K 15K 20K 25K SE +/- 102.57, N = 3 SE +/- 36.93, N = 3 23175.10 22936.55 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Semaphores OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Semaphores Clear buffers, SMT vulnerable mmio_stale_data=off 400K 800K 1200K 1600K 2000K SE +/- 5573.05, N = 3 SE +/- 10137.25, N = 3 1718628.66 1716750.63 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Matrix Math OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Matrix Math Clear buffers, SMT vulnerable mmio_stale_data=off 12K 24K 36K 48K 60K SE +/- 53.26, N = 3 SE +/- 16.25, N = 3 57560.79 57656.07 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Vector Math OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Vector Math Clear buffers, SMT vulnerable mmio_stale_data=off 13K 26K 39K 52K 65K SE +/- 11.95, N = 3 SE +/- 17.22, N = 3 58932.06 58931.54 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: x86_64 RdRand OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: x86_64 RdRand Clear buffers, SMT vulnerable mmio_stale_data=off 20K 40K 60K 80K 100K SE +/- 24.81, N = 3 SE +/- 10.17, N = 3 77956.30 77931.96 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Memory Copying OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Memory Copying Clear buffers, SMT vulnerable mmio_stale_data=off 300 600 900 1200 1500 SE +/- 0.83, N = 3 SE +/- 0.75, N = 3 1613.78 1614.44 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Socket Activity OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Socket Activity Clear buffers, SMT vulnerable mmio_stale_data=off 2K 4K 6K 8K 10K SE +/- 5.15, N = 3 SE +/- 5.40, N = 3 9216.66 9193.96 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Context Switching OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Context Switching Clear buffers, SMT vulnerable mmio_stale_data=off 3M 6M 9M 12M 15M SE +/- 160660.40, N = 3 SE +/- 204019.40, N = 15 12430837.52 13841930.87 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Glibc C String Functions OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Glibc C String Functions Clear buffers, SMT vulnerable mmio_stale_data=off 400K 800K 1200K 1600K 2000K SE +/- 10949.44, N = 3 SE +/- 9914.06, N = 3 1716938.57 1717173.49 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: Glibc Qsort Data Sorting OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: Glibc Qsort Data Sorting Clear buffers, SMT vulnerable mmio_stale_data=off 40 80 120 160 200 SE +/- 0.42, N = 3 SE +/- 0.24, N = 3 185.31 185.14 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
Stress-NG Test: System V Message Passing OpenBenchmarking.org Bogo Ops/s, More Is Better Stress-NG 0.14 Test: System V Message Passing Clear buffers, SMT vulnerable mmio_stale_data=off 2M 4M 6M 8M 10M SE +/- 41159.70, N = 3 SE +/- 3240.59, N = 3 11271215.82 11307266.30 1. (CC) gcc options: -O2 -std=gnu99 -lm -lapparmor -latomic -lc -lcrypt -ldl -ljpeg -lrt -lsctp -lz -pthread
SVT-AV1 Encoder Mode: Preset 10 - Input: Bosphorus 4K OpenBenchmarking.org Frames Per Second, More Is Better SVT-AV1 1.0 Encoder Mode: Preset 10 - Input: Bosphorus 4K Clear buffers, SMT vulnerable mmio_stale_data=off 20 40 60 80 100 SE +/- 0.74, N = 12 SE +/- 0.60, N = 12 80.70 81.03 1. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq -pie
SVT-AV1 Encoder Mode: Preset 12 - Input: Bosphorus 4K OpenBenchmarking.org Frames Per Second, More Is Better SVT-AV1 1.0 Encoder Mode: Preset 12 - Input: Bosphorus 4K Clear buffers, SMT vulnerable mmio_stale_data=off 20 40 60 80 100 SE +/- 0.08, N = 3 SE +/- 0.40, N = 3 109.19 108.80 1. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq -pie
SVT-AV1 Encoder Mode: Preset 10 - Input: Bosphorus 1080p OpenBenchmarking.org Frames Per Second, More Is Better SVT-AV1 1.0 Encoder Mode: Preset 10 - Input: Bosphorus 1080p Clear buffers, SMT vulnerable mmio_stale_data=off 50 100 150 200 250 SE +/- 3.26, N = 3 SE +/- 2.47, N = 5 231.47 231.75 1. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq -pie
SVT-AV1 Encoder Mode: Preset 12 - Input: Bosphorus 1080p OpenBenchmarking.org Frames Per Second, More Is Better SVT-AV1 1.0 Encoder Mode: Preset 12 - Input: Bosphorus 1080p Clear buffers, SMT vulnerable mmio_stale_data=off 90 180 270 360 450 SE +/- 0.28, N = 3 SE +/- 0.85, N = 3 425.40 429.78 1. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq -pie
SVT-HEVC Tuning: 7 - Input: Bosphorus 4K OpenBenchmarking.org Frames Per Second, More Is Better SVT-HEVC 1.5.0 Tuning: 7 - Input: Bosphorus 4K Clear buffers, SMT vulnerable mmio_stale_data=off 10 20 30 40 50 SE +/- 0.34, N = 3 SE +/- 0.31, N = 3 45.98 45.89 1. (CC) gcc options: -fPIE -fPIC -O3 -O2 -pie -rdynamic -lpthread -lrt
SVT-HEVC Tuning: 10 - Input: Bosphorus 4K OpenBenchmarking.org Frames Per Second, More Is Better SVT-HEVC 1.5.0 Tuning: 10 - Input: Bosphorus 4K Clear buffers, SMT vulnerable mmio_stale_data=off 20 40 60 80 100 SE +/- 0.05, N = 3 SE +/- 0.13, N = 3 85.44 85.61 1. (CC) gcc options: -fPIE -fPIC -O3 -O2 -pie -rdynamic -lpthread -lrt
SVT-HEVC Tuning: 7 - Input: Bosphorus 1080p OpenBenchmarking.org Frames Per Second, More Is Better SVT-HEVC 1.5.0 Tuning: 7 - Input: Bosphorus 1080p Clear buffers, SMT vulnerable mmio_stale_data=off 30 60 90 120 150 SE +/- 1.26, N = 3 SE +/- 1.49, N = 3 140.77 140.47 1. (CC) gcc options: -fPIE -fPIC -O3 -O2 -pie -rdynamic -lpthread -lrt
SVT-HEVC Tuning: 10 - Input: Bosphorus 1080p OpenBenchmarking.org Frames Per Second, More Is Better SVT-HEVC 1.5.0 Tuning: 10 - Input: Bosphorus 1080p Clear buffers, SMT vulnerable mmio_stale_data=off 60 120 180 240 300 SE +/- 0.42, N = 3 SE +/- 0.47, N = 3 282.53 282.71 1. (CC) gcc options: -fPIE -fPIC -O3 -O2 -pie -rdynamic -lpthread -lrt
SVT-VP9 Tuning: VMAF Optimized - Input: Bosphorus 4K OpenBenchmarking.org Frames Per Second, More Is Better SVT-VP9 0.3 Tuning: VMAF Optimized - Input: Bosphorus 4K Clear buffers, SMT vulnerable mmio_stale_data=off 13 26 39 52 65 SE +/- 0.42, N = 11 SE +/- 0.65, N = 3 57.18 56.93 1. (CC) gcc options: -O3 -fcommon -fPIE -fPIC -fvisibility=hidden -pie -rdynamic -lpthread -lrt -lm
SVT-VP9 Tuning: VMAF Optimized - Input: Bosphorus 1080p OpenBenchmarking.org Frames Per Second, More Is Better SVT-VP9 0.3 Tuning: VMAF Optimized - Input: Bosphorus 1080p Clear buffers, SMT vulnerable mmio_stale_data=off 40 80 120 160 200 SE +/- 2.14, N = 5 SE +/- 2.00, N = 5 193.67 192.98 1. (CC) gcc options: -O3 -fcommon -fPIE -fPIC -fvisibility=hidden -pie -rdynamic -lpthread -lrt -lm
SVT-VP9 Tuning: PSNR/SSIM Optimized - Input: Bosphorus 4K OpenBenchmarking.org Frames Per Second, More Is Better SVT-VP9 0.3 Tuning: PSNR/SSIM Optimized - Input: Bosphorus 4K Clear buffers, SMT vulnerable mmio_stale_data=off 14 28 42 56 70 SE +/- 0.02, N = 3 SE +/- 0.07, N = 3 62.55 62.21 1. (CC) gcc options: -O3 -fcommon -fPIE -fPIC -fvisibility=hidden -pie -rdynamic -lpthread -lrt -lm
SVT-VP9 Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080p OpenBenchmarking.org Frames Per Second, More Is Better SVT-VP9 0.3 Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080p Clear buffers, SMT vulnerable mmio_stale_data=off 40 80 120 160 200 SE +/- 0.16, N = 3 SE +/- 0.19, N = 3 201.31 199.83 1. (CC) gcc options: -O3 -fcommon -fPIE -fPIC -fvisibility=hidden -pie -rdynamic -lpthread -lrt -lm
TensorFlow Lite Model: SqueezeNet OpenBenchmarking.org Microseconds, Fewer Is Better TensorFlow Lite 2022-05-18 Model: SqueezeNet Clear buffers, SMT vulnerable mmio_stale_data=off 400 800 1200 1600 2000 SE +/- 7.65, N = 3 SE +/- 7.43, N = 3 1813.30 1816.11
TensorFlow Lite Model: Inception V4 OpenBenchmarking.org Microseconds, Fewer Is Better TensorFlow Lite 2022-05-18 Model: Inception V4 Clear buffers, SMT vulnerable mmio_stale_data=off 6K 12K 18K 24K 30K SE +/- 148.28, N = 3 SE +/- 43.81, N = 3 26947.2 26719.8
TensorFlow Lite Model: NASNet Mobile OpenBenchmarking.org Microseconds, Fewer Is Better TensorFlow Lite 2022-05-18 Model: NASNet Mobile Clear buffers, SMT vulnerable mmio_stale_data=off 1500 3000 4500 6000 7500 SE +/- 21.37, N = 3 SE +/- 14.99, N = 3 7077.56 7082.13
TensorFlow Lite Model: Mobilenet Float OpenBenchmarking.org Microseconds, Fewer Is Better TensorFlow Lite 2022-05-18 Model: Mobilenet Float Clear buffers, SMT vulnerable mmio_stale_data=off 300 600 900 1200 1500 SE +/- 3.29, N = 3 SE +/- 3.71, N = 3 1432.14 1437.31
TensorFlow Lite Model: Mobilenet Quant OpenBenchmarking.org Microseconds, Fewer Is Better TensorFlow Lite 2022-05-18 Model: Mobilenet Quant Clear buffers, SMT vulnerable mmio_stale_data=off 600 1200 1800 2400 3000 SE +/- 56.11, N = 12 SE +/- 58.68, N = 15 2756.90 2716.21
TensorFlow Lite Model: Inception ResNet V2 OpenBenchmarking.org Microseconds, Fewer Is Better TensorFlow Lite 2022-05-18 Model: Inception ResNet V2 Clear buffers, SMT vulnerable mmio_stale_data=off 6K 12K 18K 24K 30K SE +/- 14.80, N = 3 SE +/- 86.07, N = 3 26868.1 27460.1
Timed Linux Kernel Compilation Build: defconfig OpenBenchmarking.org Seconds, Fewer Is Better Timed Linux Kernel Compilation 5.16 Build: defconfig Clear buffers, SMT vulnerable mmio_stale_data=off 20 40 60 80 100 SE +/- 0.12, N = 3 SE +/- 0.08, N = 3 78.92 79.06
Timed LLVM Compilation Build System: Unix Makefiles OpenBenchmarking.org Seconds, Fewer Is Better Timed LLVM Compilation 13.0 Build System: Unix Makefiles Clear buffers, SMT vulnerable mmio_stale_data=off 130 260 390 520 650 SE +/- 1.36, N = 3 SE +/- 0.84, N = 3 615.36 617.89
Phoronix Test Suite v10.8.5