Intel Xeon Ice Lake Mitigation Comparison

Intel CPU Xeon Scalable Ice Lake mitigation benchmark impact by Michael Larabel for a future article.

HTML result view exported from: https://openbenchmarking.org/result/2105232-IB-XEONICELA80&grr&sor.

Intel Xeon Ice Lake Mitigation ComparisonProcessorMotherboardChipsetMemoryDiskGraphicsMonitorNetworkOSKernelDesktopDisplay ServerCompilerFile-SystemScreen ResolutionDefaultmitigations=off2 x Intel Xeon Platinum 8380 @ 3.40GHz (80 Cores / 160 Threads)Intel M50CYP2SB2U (SE5C6200.86B.0022.D08.2103221623 BIOS)Intel Device 0998504GB800GB INTEL SSDPF21Q800GBASPEEDVE2282 x Intel X710 for 10GBASE-T + 2 x Intel E810-C for QSFPUbuntu 21.045.11.0-17-generic (x86_64)GNOME Shell 3.38.4X ServerGCC 10.3.0ext41920x1080OpenBenchmarking.orgKernel Details- Transparent Huge Pages: madviseCompiler Details- --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --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-mutex --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none=/build/gcc-10-gDeRY6/gcc-10-10.3.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-10-gDeRY6/gcc-10-10.3.0/debian/tmp-gcn/usr,hsa --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: 4096Processor Details- Scaling Governor: intel_pstate powersave - CPU Microcode: 0xd000270Java Details- OpenJDK Runtime Environment (build 11.0.11+9-Ubuntu-0ubuntu2)Python Details- Python 3.9.4Security Details- Default: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + 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 - mitigations=off: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Vulnerable + spectre_v1: Vulnerable: __user pointer sanitization and usercopy barriers only; no swapgs barriers + spectre_v2: Vulnerable IBPB: disabled STIBP: disabled + srbds: Not affected + tsx_async_abort: Not affected

Intel Xeon Ice Lake Mitigation Comparisonrenaissance: Akka Unbalanced Cobwebbed Treerenaissance: In-Memory Database Shootoutrenaissance: Scala Dottyrenaissance: Genetic Algorithm Using Jenetics + Futureshbase: Increment - 1hbase: Increment - 1hbase: Seq Read - 1hbase: Seq Read - 1renaissance: Rand Foresthbase: Rand Read - 1hbase: Rand Read - 1keydb: renaissance: Apache Spark ALSpgbench: 100 - 250 - Read Only - Average Latencypgbench: 100 - 250 - Read Onlyrenaissance: Twitter HTTP Requestsinfluxdb: 4 - 10000 - 2,5000,1 - 10000build-linux-kernel: Time To Compilebuild-nodejs: Time To Compileinfluxdb: 64 - 10000 - 2,5000,1 - 10000build-godot: Time To Compilepjsip: INVITEpjsip: OPTIONS, Statefulmnn: inception-v3mnn: SqueezeNetV1.0tensorflow-lite: Inception V4tensorflow-lite: SqueezeNetsqlite-speedtest: Timed Time - Size 1,000hbase: Rand Write - 1hbase: Rand Write - 1pjsip: OPTIONS, Statelessdacapobench: Tradebeansredis: SEThbase: Seq Write - 1hbase: Seq Write - 1postmark: Disk Transaction Performancesockperf: Throughputchia-vdf: Square Plain C++pgbench: 100 - 250 - Read Write - Average Latencypgbench: 100 - 250 - Read Writevosk: dacapobench: H2build-mesa: Time To Compilesockperf: Latency Ping Pongctx-clock: Context Switch TimeDefaultmitigations=off30002.43218901.3161791.6696677.145176256610819222991.8471097907516052.163448.8560.2639579666118.777721644.524.780101.3121188389.474.8772588382125.9044.86869166847669.759.564372567040482170451720526.95313056462143155571399678.6822890428.6181075418.9618.34722830102.52219055.4731737.2456637.155176756510969092925.4371132879508188.293422.7390.2629583456172.604727924.424.762102.1071196123.774.9402623381325.4654.74368010347385.960.183382580441291167961762944.93313060261993209391402008.6952886528.4021071019.0148.362228OpenBenchmarking.org

Renaissance

Test: Akka Unbalanced Cobwebbed Tree

OpenBenchmarking.orgms, Fewer Is BetterRenaissance 0.10.0Test: Akka Unbalanced Cobwebbed TreeDefaultmitigations=off6K12K18K24K30KSE +/- 213.94, N = 15SE +/- 325.81, N = 1530002.4330102.52

Renaissance

Test: In-Memory Database Shootout

OpenBenchmarking.orgms, Fewer Is BetterRenaissance 0.10.0Test: In-Memory Database ShootoutDefaultmitigations=off4K8K12K16K20KSE +/- 171.28, N = 7SE +/- 197.89, N = 1518901.3219055.47

Renaissance

Test: Scala Dotty

OpenBenchmarking.orgms, Fewer Is BetterRenaissance 0.10.0Test: Scala Dottymitigations=offDefault400800120016002000SE +/- 32.02, N = 20SE +/- 48.82, N = 251737.251791.67

Renaissance

Test: Genetic Algorithm Using Jenetics + Futures

OpenBenchmarking.orgms, Fewer Is BetterRenaissance 0.10.0Test: Genetic Algorithm Using Jenetics + Futuresmitigations=offDefault14002800420056007000SE +/- 60.53, N = 5SE +/- 59.27, N = 76637.166677.15

Apache HBase

Test: Increment - Clients: 1

OpenBenchmarking.orgMicroseconds - Average Latency, Fewer Is BetterApache HBase 2.2.3Test: Increment - Clients: 1Defaultmitigations=off400800120016002000SE +/- 19.34, N = 3SE +/- 8.21, N = 317621767

Apache HBase

Test: Increment - Clients: 1

OpenBenchmarking.orgRows Per Second, More Is BetterApache HBase 2.2.3Test: Increment - Clients: 1Defaultmitigations=off120240360480600SE +/- 6.03, N = 3SE +/- 2.85, N = 3566565

Apache HBase

Test: Sequential Read - Clients: 1

OpenBenchmarking.orgMicroseconds - Average Latency, Fewer Is BetterApache HBase 2.2.3Test: Sequential Read - Clients: 1Defaultmitigations=off2004006008001000SE +/- 12.77, N = 3SE +/- 11.92, N = 410811096

Apache HBase

Test: Sequential Read - Clients: 1

OpenBenchmarking.orgRows Per Second, More Is BetterApache HBase 2.2.3Test: Sequential Read - Clients: 1Defaultmitigations=off2004006008001000SE +/- 10.74, N = 3SE +/- 9.65, N = 4922909

Renaissance

Test: Random Forest

OpenBenchmarking.orgms, Fewer Is BetterRenaissance 0.10.0Test: Random Forestmitigations=offDefault6001200180024003000SE +/- 20.17, N = 5SE +/- 24.69, N = 52925.442991.85

Apache HBase

Test: Random Read - Clients: 1

OpenBenchmarking.orgMicroseconds - Average Latency, Fewer Is BetterApache HBase 2.2.3Test: Random Read - Clients: 1Defaultmitigations=off2004006008001000SE +/- 7.75, N = 3SE +/- 9.68, N = 310971132

Apache HBase

Test: Random Read - Clients: 1

OpenBenchmarking.orgRows Per Second, More Is BetterApache HBase 2.2.3Test: Random Read - Clients: 1Defaultmitigations=off2004006008001000SE +/- 6.56, N = 3SE +/- 7.51, N = 3907879

KeyDB

OpenBenchmarking.orgOps/sec, More Is BetterKeyDB 6.0.16Defaultmitigations=off110K220K330K440K550KSE +/- 3720.26, N = 3SE +/- 7304.04, N = 15516052.16508188.291. (CXX) g++ options: -O2 -levent_openssl -levent -lcrypto -lssl -lpthread -lz -lpcre

Renaissance

Test: Apache Spark ALS

OpenBenchmarking.orgms, Fewer Is BetterRenaissance 0.10.0Test: Apache Spark ALSmitigations=offDefault7001400210028003500SE +/- 30.59, N = 5SE +/- 26.99, N = 53422.743448.86

PostgreSQL pgbench

Scaling Factor: 100 - Clients: 250 - Mode: Read Only - Average Latency

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL pgbench 13.0Scaling Factor: 100 - Clients: 250 - Mode: Read Only - Average Latencymitigations=offDefault0.05920.11840.17760.23680.296SE +/- 0.002, N = 15SE +/- 0.005, N = 130.2620.2631. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

PostgreSQL pgbench

Scaling Factor: 100 - Clients: 250 - Mode: Read Only

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 13.0Scaling Factor: 100 - Clients: 250 - Mode: Read Onlymitigations=offDefault200K400K600K800K1000KSE +/- 8471.25, N = 15SE +/- 17109.96, N = 139583459579661. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

Renaissance

Test: Twitter HTTP Requests

OpenBenchmarking.orgms, Fewer Is BetterRenaissance 0.10.0Test: Twitter HTTP RequestsDefaultmitigations=off13002600390052006500SE +/- 57.24, N = 5SE +/- 36.16, N = 56118.786172.60

InfluxDB

Concurrent Streams: 4 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000

OpenBenchmarking.orgval/sec, More Is BetterInfluxDB 1.8.2Concurrent Streams: 4 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000mitigations=offDefault160K320K480K640K800KSE +/- 2881.97, N = 3SE +/- 1793.11, N = 3727924.4721644.5

Timed Linux Kernel Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Linux Kernel Compilation 5.10.20Time To Compilemitigations=offDefault612182430SE +/- 0.37, N = 15SE +/- 0.34, N = 1324.7624.78

Timed Node.js Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Node.js Compilation 15.11Time To CompileDefaultmitigations=off20406080100SE +/- 0.54, N = 3SE +/- 0.24, N = 3101.31102.11

InfluxDB

Concurrent Streams: 64 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000

OpenBenchmarking.orgval/sec, More Is BetterInfluxDB 1.8.2Concurrent Streams: 64 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000mitigations=offDefault300K600K900K1200K1500KSE +/- 1536.16, N = 3SE +/- 4126.24, N = 31196123.71188389.4

Timed Godot Game Engine Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Godot Game Engine Compilation 3.2.3Time To CompileDefaultmitigations=off20406080100SE +/- 0.55, N = 3SE +/- 0.45, N = 374.8874.94

PJSIP

Method: INVITE

OpenBenchmarking.orgResponses Per Second, More Is BetterPJSIP 2.11Method: INVITEmitigations=offDefault6001200180024003000SE +/- 3.79, N = 3SE +/- 31.32, N = 3262325881. (CC) gcc options: -lstdc++ -lssl -lcrypto -luuid -lm -lrt -lpthread -lasound -O2

PJSIP

Method: OPTIONS, Stateful

OpenBenchmarking.orgResponses Per Second, More Is BetterPJSIP 2.11Method: OPTIONS, StatefulDefaultmitigations=off8001600240032004000SE +/- 10.39, N = 3SE +/- 3.79, N = 3382138131. (CC) gcc options: -lstdc++ -lssl -lcrypto -luuid -lm -lrt -lpthread -lasound -O2

Mobile Neural Network

Model: inception-v3

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 1.1.3Model: inception-v3mitigations=offDefault612182430SE +/- 0.20, N = 3SE +/- 0.12, N = 325.4725.90MIN: 23.79 / MAX: 74.89MIN: 24.95 / MAX: 57.751. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl

Mobile Neural Network

Model: SqueezeNetV1.0

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 1.1.3Model: SqueezeNetV1.0mitigations=offDefault1.09532.19063.28594.38125.4765SE +/- 0.016, N = 3SE +/- 0.007, N = 34.7434.868MIN: 4.53 / MAX: 11.47MIN: 4.49 / MAX: 18.51. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl

TensorFlow Lite

Model: Inception V4

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2020-08-23Model: Inception V4mitigations=offDefault150K300K450K600K750KSE +/- 8121.20, N = 3SE +/- 4490.09, N = 3680103691668

TensorFlow Lite

Model: SqueezeNet

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2020-08-23Model: SqueezeNetmitigations=offDefault10K20K30K40K50KSE +/- 403.24, N = 3SE +/- 202.00, N = 347385.947669.7

SQLite Speedtest

Timed Time - Size 1,000

OpenBenchmarking.orgSeconds, Fewer Is BetterSQLite Speedtest 3.30Timed Time - Size 1,000Defaultmitigations=off1326395265SE +/- 0.21, N = 3SE +/- 0.09, N = 359.5660.181. (CC) gcc options: -O2 -ldl -lz -lpthread

Apache HBase

Test: Random Write - Clients: 1

OpenBenchmarking.orgMicroseconds - Average Latency, Fewer Is BetterApache HBase 2.2.3Test: Random Write - Clients: 1Defaultmitigations=off918273645SE +/- 0.43, N = 15SE +/- 0.57, N = 143738

Apache HBase

Test: Random Write - Clients: 1

OpenBenchmarking.orgRows Per Second, More Is BetterApache HBase 2.2.3Test: Random Write - Clients: 1mitigations=offDefault6K12K18K24K30KSE +/- 320.28, N = 14SE +/- 345.85, N = 152580425670

PJSIP

Method: OPTIONS, Stateless

OpenBenchmarking.orgResponses Per Second, More Is BetterPJSIP 2.11Method: OPTIONS, Statelessmitigations=offDefault9K18K27K36K45KSE +/- 329.47, N = 3SE +/- 197.93, N = 341291404821. (CC) gcc options: -lstdc++ -lssl -lcrypto -luuid -lm -lrt -lpthread -lasound -O2

DaCapo Benchmark

Java Test: Tradebeans

OpenBenchmarking.orgmsec, Fewer Is BetterDaCapo Benchmark 9.12-MR1Java Test: Tradebeansmitigations=offDefault4K8K12K16K20KSE +/- 139.58, N = 4SE +/- 156.28, N = 71679617045

Redis

Test: SET

OpenBenchmarking.orgRequests Per Second, More Is BetterRedis 6.0.9Test: SETmitigations=offDefault400K800K1200K1600K2000KSE +/- 21191.99, N = 15SE +/- 10529.94, N = 31762944.931720526.951. (CXX) g++ options: -MM -MT -g3 -fvisibility=hidden -O3

Apache HBase

Test: Sequential Write - Clients: 1

OpenBenchmarking.orgMicroseconds - Average Latency, Fewer Is BetterApache HBase 2.2.3Test: Sequential Write - Clients: 1Defaultmitigations=off714212835SE +/- 0.52, N = 15SE +/- 0.51, N = 153131

Apache HBase

Test: Sequential Write - Clients: 1

OpenBenchmarking.orgRows Per Second, More Is BetterApache HBase 2.2.3Test: Sequential Write - Clients: 1mitigations=offDefault7K14K21K28K35KSE +/- 418.18, N = 15SE +/- 432.13, N = 153060230564

PostMark

Disk Transaction Performance

OpenBenchmarking.orgTPS, More Is BetterPostMark 1.51Disk Transaction PerformanceDefaultmitigations=off13002600390052006500SE +/- 74.77, N = 4SE +/- 51.00, N = 3621461991. (CC) gcc options: -O3

Sockperf

Test: Throughput

OpenBenchmarking.orgMessages Per Second, More Is BetterSockperf 3.4Test: Throughputmitigations=offDefault70K140K210K280K350KSE +/- 2916.53, N = 5SE +/- 5074.55, N = 253209393155571. (CXX) g++ options: --param -O3 -rdynamic -ldl -lpthread

Chia Blockchain VDF

Test: Square Plain C++

OpenBenchmarking.orgIPS, More Is BetterChia Blockchain VDF 1.0.1Test: Square Plain C++mitigations=offDefault30K60K90K120K150KSE +/- 400.00, N = 3SE +/- 835.33, N = 31402001399671. (CXX) g++ options: -flto -no-pie -lgmpxx -lgmp -lboost_system -pthread

PostgreSQL pgbench

Scaling Factor: 100 - Clients: 250 - Mode: Read Write - Average Latency

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL pgbench 13.0Scaling Factor: 100 - Clients: 250 - Mode: Read Write - Average LatencyDefaultmitigations=off246810SE +/- 0.022, N = 3SE +/- 0.018, N = 38.6828.6951. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

PostgreSQL pgbench

Scaling Factor: 100 - Clients: 250 - Mode: Read Write

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 13.0Scaling Factor: 100 - Clients: 250 - Mode: Read WriteDefaultmitigations=off6K12K18K24K30KSE +/- 79.55, N = 3SE +/- 54.70, N = 328904288651. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

VOSK Speech Recognition Toolkit

OpenBenchmarking.orgSeconds, Fewer Is BetterVOSK Speech Recognition Toolkit 0.3.21mitigations=offDefault714212835SE +/- 0.40, N = 3SE +/- 0.38, N = 328.4028.62

DaCapo Benchmark

Java Test: H2

OpenBenchmarking.orgmsec, Fewer Is BetterDaCapo Benchmark 9.12-MR1Java Test: H2mitigations=offDefault2K4K6K8K10KSE +/- 34.81, N = 4SE +/- 64.56, N = 41071010754

Timed Mesa Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Mesa Compilation 21.0Time To CompileDefaultmitigations=off510152025SE +/- 0.08, N = 3SE +/- 0.06, N = 318.9619.01

Sockperf

Test: Latency Ping Pong

OpenBenchmarking.orgusec, Fewer Is BetterSockperf 3.4Test: Latency Ping PongDefaultmitigations=off246810SE +/- 0.085, N = 5SE +/- 0.091, N = 58.3478.3621. (CXX) g++ options: --param -O3 -rdynamic -ldl -lpthread

ctx_clock

Context Switch Time

OpenBenchmarking.orgClocks, Fewer Is Betterctx_clockContext Switch TimeDefaultmitigations=off50100150200250228228

Phoronix Test Suite v10.8.5