CPU Benchmark

Latest Phoronix CPU benchmark 10.8.4 under Ubuntu 20.04.LTS and 5.13.0-28-generic kernel.
Previous Phoronix CPU benchmark 10.8 under Ubuntu 20.04.LTS and 5.13.0-28-generic kernel.
Previous Phoronix CPU benchmark under Ubuntu 20.04.LTS and 5.4.0-53-generic kernel.

Older benchmark data are available here. Since the initial publication below, AMD latest Ryzen 3 platforms (Marquis K836 and K837) have reached a new performance plateau. With high core count and impressive clock speed, Ryzen 3 processors deliver tremendous computing power and value for mid-range workstation.

This Phoronix CPU benchmark suite was published on Feb 12, 2018. The test platform uses 4.15 kernel which include the Spectre/Meltdown patches. Additional benchmark data are provided by ASL for analysis on the processor performance and cost consideration.

This table summarizes relative performance of various CPU models (uni-processor configuration) using AMD Epyc 7601 as the base index of one. The fastest performer for each test is highlighted. Here are some notable performers:

  • Ryzen Threadripper 2990WX beats Epyc 7601 in 30 out of 31 tests at less than half the cost.
  • Core i9-7940X beats Epyc 7601 in 18 out of 31 tests at less than 1/3rd the cost.
  • Core i7-9700K beats Epyc 7601 in 13 out of 31 tests at less than 1/10th the cost.

    Marquis Model
    CPU Model
    Core/Thread/Clock
    MSRP
    Faster runs (31)
    K834
    Epyc 7601
    32/64/2.2G
    $4,265
    (-)
    C535
    i7-9700K
    8/8/3.6G
    $385
    (13)
    C535
    i7-8086K
    6/12/4.0G
    $425
    (13)
    K833
    TR 2990WX
    32/64/3.0G
    $1699
    (30)
    C532
    i9-7960X
    16/32/2.8G
    $1699
    (19)
    C532
    i7-7820X
    8/16/3.6G
    $599
    (12)
    K833
    TR 1900X
    8/16/3.8G
    $449
    (15)
    C532
    i9-7900X
    10/20/3.3G
    $999
    (13)
    C532
    i7-7800X
    6/12/3.5G
    $383
    (13)
    C532
    i9-7940X
    14/28/3.1G
    $1399
    (18)
    C535
    i7-8700K
    6/12/3.7G
    $359
    (13)
    NAS Parallel Benchmarks v3.3 1 0.3 0.4 1.3 0.8 0.4 0.4 0.5 0.3 0.7 0.4
    Parboil v2.5 (OpenMP CUTCP) 1 0.18 0.23 1.13 0.43 0.25 0.46 0.31 0.19 0.4 0.23
    Parboil v2.5 (OpenMP Stencil) 1 0.61 0.65 0.85 1.85 0.9 0.66 1.31 0.82 1.74 0.65
    Rodinia v2.4 (OpenMP LavaMD) 1 0.35 0.31 1.35 0.68 0.38 0.42 0.48 0.29 0.63 0.31
    Timed MAFFT Alignmen v6.864 1 1.54 1.4 1.22 1.11 1.28 1.49 1.42 1.21 1.4 1.38
    Timed MrBayes Analysis v3.1.2 1 0.64 0.6 1.04 1.16 0.71 0.7 0.84 0.54 1.05 0.6
    Botan v2.4.0 (KASUMI - Encrypt) 1 1.5 1.5 1.3 1.3 1.3 1.3 1.3 1.2 1.3 1.4
    Botan v2.4.0 (AES-256 - Encrypt) 1 0.9 0.9 1.2 0.3 0.3 1.3 0.3 0.3 0.3 0.9
    Botan v2.4.0 (Twofish - Encrypt) 1 1.5 1.5 1.1 1.3 1.3 1.2 1.3 1.2 1.3 1.5
    John The Ripper v1.8.0 (Blowfish) 1 0.5 0.6 1.7 1.2 0.7 0.6 0.9 0.5 1.1 0.6
    Open FMM Nero2D v2.0.2 1 0.40 0.31 1.12 0.51 0.46 0.46 0.28 0.45 0.31
    TTSIOD 3D Renderer v2.3a 1 0.7 0.7 1.2 1.2 0.7 0.7 0.9 0.5 1.1 0.7
    H.264 Video Encoding v2018-02-05 1 0.7 0.7 1.3 1.3 0.8 0.7 1 0.6 1.2 0.7
    Himeno Benchmark v3. 1 3.1 3.2 1.3 2.8 2.7 1.3 2.8 2.6 2.9 3.0
    7-Zip Compression v9.20.1 1 0.4 0.4 1.1 0.9 0.5 0.4 0.6 0.4 0.8 0.4
    asmFish v2017-09-19 1 0.4 0.4 1.2 0.7 0.4 0.3 0.5 0.3 0.7 0.3
    Timed Linux Kernel Compilation v4.13 1 0.61 0.58 1.46 1.1 0.67 0.6 .81 .52 1.03 0.58
    C-Ray v1.1 1 0.2 0.23 1.52 0.35 0.2 0.46 0.25 0.15 0.33 0.23
    POV-Ray v3.7.0.7 1 0.38 0.36 1.27 0.76 0.43 0.43 0.53 0.33 0.71 0.36
    AOBench 1 1.36 1.37 1.14 1.14 1.17 1.12 1.14 1.08 1.16 1.30
    LAME MP3 Encoding v3.100 1 1.78 1.73 1.31 1.52 1.56 1.31 1.54 1.45 1.53 1.7
    FFmpeg v3.4.1 (H.264 HD to NTSC DV) 1 3.13 2.36 1.27 0.77 0.81 1.37 0.79 1.96 0.77 2.37
    Hackbench (Count:8 - Type:Process) 1 0.52 0.64 1.13 0.98 0.62 0.69 0.73 0.48 0.86 0.6
    OpenSSL v1.1.0f (RSA 4096-bit Performance) 1 0.6 0.4 1.3 1 0.5 0.4 0.7 0.4 0.9 0.4
    PyBench v2008-08-14 1 1.77 1.8 1.24 1.5 1.54 1.22 1.53 1.43 1.54 1.7
    NGONX Benchmark v1.9.9 1 2.0 1.9 1.2 1.8 1.8 1.2 1.7 1.7 1.8 1.9
    Apache Benchmark v2.4.29 1 2.4 2.1 1.1 1.6 1.6 1.4 1.8 1.7 1.8 2.1
    PHPBench v.0.8.1 1 1.9 1.9 1.2 1.6 1.6 1.2 1.6 1.5 1.6 1.8
    Botan v2.4.0 (KASUMI - Encrypt - Decrypt) 1 1.5 1.5 1.3 1.3 1.3 1.3 1.3 1.2 1.3 1.4
    Botan v2.4.0 (AES-256 - Encrypt - Decrypt) 1 0.9 0.9 1.3 0.3 0.3 1.3 0.3 0.3 0.3 0.8
    Botan v2.4.0 (Two - Encrypt - Decrypt) 1 1.5 1.5 1.4 1.3 1.3 1.3 1.3 1.2 1.3 1.4

    Fastest Performance with Extreme Core Count

    For highly parallelized applications such as 3D rendering, highest performance can be obtained using multi-processor system with extreme high core CPU. For example, the Marquis C946 configured with four Xeon MP E7-4830v4 processors (56 cores/112 threads) obtains the fastest Povray rendering time.

    The Marquis K833 configured with one AMD Threadripper 2990WX processor (32 cores/64 threads) delivers the 2nd fastest Povray rendering time. Using the Marquis K918 with two AMD Epyc 7601 processors (64 cores/128 threads), Povray rendering speed will surpass the four-way Intel Xeon MP configuration.

    Best Performance and Cost Ratio

    The Marquis C532 configured with one Core i9 processor delivers the best performance and cost ratio due to its high instructions per cycle (IPC) and moderate high core count.

    Single Threaded Application on High CPU Clock Speed

    The Marquis C535 configured with one Intel Core i7-9700K processor delivers fastest performance on single threaded applications due to its high instructions per cycle (IPC) and advanced 9th generation architecture.

  • Multi-GPU Redshift (v3.0.31) Benchmark

    Ages of Vultures rendering speed (lower is faster) under Ubuntu 20.04 LTS using Cuda 11.1 Update 1. Past results are recorded here

    System Model
    CPU Model
    Core/Thread/Clock
    Marquis C532
    Core i9-10980XE
    18/36/3.0Ghz
    Lancelot 4884-GSR
    Silver 4214R
    24/48/2.4Ghz
    One EVGA RTX 3090 (24G-P5-3975-KR) 2m 24s 3m 4s
    One PNY RTX 3080 (VCG308010TFXPPB) 2m 48s -
    One EVGA RTX 3070 (8G-P5-3751-KR) 3m 48s -
    Two EVGA RTX 3090 (24G-P5-3975-KR) 1m 15s 1m 34s
    Two EVGA RTX 3070 (8G-P5-3751-KR) 1m 58s -

    Single GPU Blender (v2.90.1) Benchmark

    The rendering speed (lower is faster) utilizes Nvidia Optix Ray Tracing Engine under Windows 10 Pro.

    Scene
    bmw27 classroom fishy_cat koro pavillion victor
    EVGA RTX 3090 (24G-P5-3975-KR) 9s 36s 18s 39s 46s 1m 19s
    PNY RTX 3080 (VCG308010TFXPPB) 11s 43s 21s 46s 54s 1m 29s
    EVGA RTX 3070 (8G-P5-3751-KR) 16s 53s 31s 1m 6s 1m 13s 1m 49s

    Single GPU OpenCL and CUDA performance

    Phoronix 10.2 GPU benchmark featuring Nvidia RTX 3000 series (Ampere) GPU cards. The benchmarks are performed under Ubuntu 20.04 LTS using Cuda 11.1 Update 1.