Linux 5.0 To Linux 5.9 On AMD EPYC
AMD EPYC Linux kernel benchmarks by Michael Larabel for a future article.
HTML result view exported from: https://openbenchmarking.org/result/2009078-FI-KERNELCOM02&grs.
Flexible IO Tester
Type: Random Read - Engine: IO_uring - Buffered: Yes - Direct: No - Block Size: 4KB - Disk Target: Default Test Directory
Flexible IO Tester
Type: Random Read - Engine: IO_uring - Buffered: Yes - Direct: No - Block Size: 4KB - Disk Target: Default Test Directory
Stress-NG
Test: System V Message Passing
Apache Siege
Concurrent Users: 200
Stress-NG
Test: MMAP
Apache Siege
Concurrent Users: 250
Stress-NG
Test: Forking
Flexible IO Tester
Type: Random Write - Engine: IO_uring - Buffered: Yes - Direct: No - Block Size: 4KB - Disk Target: Default Test Directory
Flexible IO Tester
Type: Random Write - Engine: IO_uring - Buffered: Yes - Direct: No - Block Size: 4KB - Disk Target: Default Test Directory
Flexible IO Tester
Type: Sequential Write - Engine: Linux AIO - Buffered: Yes - Direct: No - Block Size: 2MB - Disk Target: Default Test Directory
Flexible IO Tester
Type: Sequential Write - Engine: Linux AIO - Buffered: Yes - Direct: No - Block Size: 2MB - Disk Target: Default Test Directory
TensorFlow Lite
Model: Inception V4
Memcached mcperf
Method: Set - Connections: 1
Apache Siege
Concurrent Users: 100
TensorFlow Lite
Model: Inception ResNet V2
Flexible IO Tester
Type: Random Read - Engine: Linux AIO - Buffered: Yes - Direct: No - Block Size: 4KB - Disk Target: Default Test Directory
Flexible IO Tester
Type: Random Read - Engine: Linux AIO - Buffered: Yes - Direct: No - Block Size: 4KB - Disk Target: Default Test Directory
KeyDB
FS-Mark
Test: 4000 Files, 32 Sub Dirs, 1MB Size
FS-Mark
Test: 1000 Files, 1MB Size
Flexible IO Tester
Type: Sequential Write - Engine: IO_uring - Buffered: Yes - Direct: No - Block Size: 2MB - Disk Target: Default Test Directory
Flexible IO Tester
Type: Sequential Write - Engine: IO_uring - Buffered: Yes - Direct: No - Block Size: 2MB - Disk Target: Default Test Directory
Memcached mcperf
Method: Get - Connections: 32
Redis
Test: LPUSH
Memcached mcperf
Method: Get - Connections: 1
OSBench
Test: Create Processes
OSBench
Test: Launch Programs
Memcached mcperf
Method: Delete - Connections: 1
Stress-NG
Test: MEMFD
SQLite
Threads / Copies: 1
Flexible IO Tester
Type: Sequential Read - Engine: Linux AIO - Buffered: Yes - Direct: No - Block Size: 2MB - Disk Target: Default Test Directory
Flexible IO Tester
Type: Sequential Read - Engine: Linux AIO - Buffered: Yes - Direct: No - Block Size: 2MB - Disk Target: Default Test Directory
Stress-NG
Test: Socket Activity
Memcached mcperf
Method: Set - Connections: 32
OSBench
Test: Create Files
Flexible IO Tester
Type: Random Write - Engine: Linux AIO - Buffered: Yes - Direct: No - Block Size: 4KB - Disk Target: Default Test Directory
OSBench
Test: Create Threads
Flexible IO Tester
Type: Random Write - Engine: Linux AIO - Buffered: Yes - Direct: No - Block Size: 4KB - Disk Target: Default Test Directory
Stress-NG
Test: Semaphores
Stress-NG
Test: NUMA
BlogBench
Test: Read
Geekbench
Test: CPU Multi Core - Face Detection
Facebook RocksDB
Test: Random Fill Sync
Compile Bench
Test: Initial Create
RawTherapee
Total Benchmark Time
BlogBench
Test: Write
WireGuard + Linux Networking Stack Stress Test
Stress-NG
Test: SENDFILE
SQLite
Threads / Copies: 8
Renaissance
Test: In-Memory Database Shootout
Compile Bench
Test: Compile
TensorFlow Lite
Model: Mobilenet Float
Geekbench
Test: CPU Multi Core
TensorFlow Lite
Model: Mobilenet Quant
AI Benchmark Alpha
Device Inference Score
FS-Mark
Test: 5000 Files, 1MB Size, 4 Threads
LevelDB
Benchmark: Fill Sync
Zstd Compression
Compression Level: 19
LevelDB
Benchmark: Fill Sync
Renaissance
Test: Genetic Algorithm Using Jenetics + Futures
dav1d
Video Input: Summer Nature 4K
AI Benchmark Alpha
Device AI Score
OpenVKL
Benchmark: vklBenchmark
DaCapo Benchmark
Java Test: Tradebeans
dav1d
Video Input: Summer Nature 1080p
Facebook RocksDB
Test: Random Fill
PostMark
Disk Transaction Performance
DaCapo Benchmark
Java Test: Tradesoap
OSBench
Test: Memory Allocations
SQLite Speedtest
Timed Time - Size 1,000
DaCapo Benchmark
Java Test: H2
Radiance Benchmark
Test: SMP Parallel
Facebook RocksDB
Test: Sequential Fill
Intel Open Image Denoise
Scene: Memorial
Swet
Average
Stress-NG
Test: Glibc C String Functions
TensorFlow Lite
Model: SqueezeNet
Facebook RocksDB
Test: Read While Writing
SVT-AV1
Encoder Mode: Enc Mode 4 - Input: 1080p
TensorFlow Lite
Model: NASNet Mobile
LevelDB
Benchmark: Sequential Fill
SVT-AV1
Encoder Mode: Enc Mode 8 - Input: 1080p
LevelDB
Benchmark: Sequential Fill
YafaRay
Total Time For Sample Scene
LevelDB
Benchmark: Random Read
dav1d
Video Input: Chimera 1080p
MariaDB
Clients: 64
dav1d
Video Input: Chimera 1080p 10-bit
Stress-NG
Test: Glibc Qsort Data Sorting
Zstd Compression
Compression Level: 3
LevelDB
Benchmark: Seek Random
Renaissance
Test: Twitter HTTP Requests
Open Porous Media
OPM Benchmark: Flow MPI Norne - Threads: 8
Coremark
CoreMark Size 666 - Iterations Per Second
AI Benchmark Alpha
Device Training Score
LevelDB
Benchmark: Random Fill
Renaissance
Test: Apache Spark ALS
DaCapo Benchmark
Java Test: Jython
LevelDB
Benchmark: Random Fill
Timed Apache Compilation
Time To Compile
DeepSpeech
Acceleration: CPU
Hugin
Panorama Photo Assistant + Stitching Time
asmFish
1024 Hash Memory, 26 Depth
OCRMyPDF
Processing 60 Page PDF Document
Facebook RocksDB
Test: Random Read
Flexible IO Tester
Type: Sequential Read - Engine: IO_uring - Buffered: Yes - Direct: No - Block Size: 2MB - Disk Target: Default Test Directory
Flexible IO Tester
Type: Sequential Read - Engine: IO_uring - Buffered: Yes - Direct: No - Block Size: 2MB - Disk Target: Default Test Directory
LevelDB
Benchmark: Hot Read
LevelDB
Benchmark: Random Delete
LevelDB
Benchmark: Overwrite
MariaDB
Clients: 512
Timed Linux Kernel Compilation
Time To Compile
LuxCoreRender
Scene: Rainbow Colors and Prism
Geekbench
Test: CPU Single Core - Gaussian Blur
Stockfish
Total Time
LuxCoreRender
Scene: DLSC
Stress-NG
Test: Crypto
LevelDB
Benchmark: Overwrite
Timed LLVM Compilation
Time To Compile
BRL-CAD
VGR Performance Metric
PyBench
Total For Average Test Times
Stress-NG
Test: Atomic
Geekbench
Test: CPU Single Core - Horizon Detection
Geekbench
Test: CPU Single Core
Build2
Time To Compile
Geekbench
Test: CPU Single Core - Face Detection
Geekbench
Test: CPU Multi Core - Horizon Detection
miniFE
Problem Size: Small
Stress-NG
Test: CPU Stress
High Performance Conjugate Gradient
NAMD
ATPase Simulation - 327,506 Atoms
Geekbench
Test: CPU Multi Core - Gaussian Blur
Stress-NG
Test: Context Switching
Redis
Test: SET
Redis
Test: GET
Redis
Test: SADD
Phoronix Test Suite v10.8.5