Skip to content

HiFive Premier P550 #65

Open
Open
HiFive Premier P550#65
Labels
testing in progress
@geerlingguy

Description

@geerlingguy
geerlingguy
opened on Jan 24, 2025

Image

Basic information

Linux/system information

# output of `screenfetch`
                          ./+o+-       ubuntu@ubuntu
                  yyyyy- -yyyyyy+      OS: Ubuntu 24.04 noble
               ://+//////-yyyyyyo      Kernel: riscv64 Linux 6.6.21-9-premier
           .++ .:/++++++/-.+sss/`      Uptime: 1m
         .:++o:  /++++++++/:--:/-      Packages: 1735
        o:+o+:++.`..```.-/oo+++++/     Shell: dash
       .:+o:+o/.          `+sssoo+/    Disk: 5.5G / 115G (5%)
  .++/+:+oo+o:`             /sssooo.   CPU: Unknown @ 4x 1.4GHz
 /+++//+:`oo+o               /::--:.   GPU: 
 \+/+o+++`o++o               ++////.   RAM: 782MiB / 15999MiB
  .++.o+++oo+:`             /dddhhh.  
       .+.o+oo:.          `oddhhhh+   
        \+.++o+o``-````.:ohdhhhhh+    
         `:o+++ `ohhhhhhhhyo++os:     
           .o:`.syhhhhhhh/.oo++o`     
               /osyyyyyyo++ooo+++/    
                   ````` +oo+++o\:    
                          `oo++.      

# output of `uname -a`
Linux ubuntu 6.6.21-9-premier #1 SMP PREEMPT_DYNAMIC Sat Nov  9 00:21:16 UTC 2024 riscv64 riscv64 riscv64 GNU/Linux

Benchmark results

CPU

Power

  • Idle power draw (at wall): 12.5 W (7.2 W using 12V DC adapter)
  • Maximum simulated power draw (stress-ng --matrix 0): 13.5 W (8.5 W using 12V DC adapter)
  • During Geekbench multicore benchmark: 13.7 W (9.2 W using 12V DC adapter)
  • During top500 HPL benchmark: 13.9 W (8.9 W using 12V DC adapter)

Power was measured at the wall using a ThirdReality Smart Outlet, running into the included Flex ATX PSU, unless otherwise noted.

Disk

Built-in eMMC Storage (128 GB)

Benchmark Result
iozone 4K random read 21.82 MB/s
iozone 4K random write 33.07 MB/s
iozone 1M random read 207.44 MB/s
iozone 1M random write 162.38 MB/s
iozone 1M sequential read 204.20 MB/s
iozone 1M sequential write 179.72 MB/s

Kioxia XG8 (2TB - in PCIe to x4 M.2 NVMe adapter)

Benchmark Result
iozone 4K random read 47.50 MB/s
iozone 4K random write 87.84 MB/s
iozone 1M random read 703.95 MB/s
iozone 1M random write 796.03 MB/s
iozone 1M sequential read 712.22 MB/s
iozone 1M sequential write 798.01 MB/s

Network

iperf3 results:

  • iperf3 -c $SERVER_IP: 942 Mbps
  • iperf3 -c $SERVER_IP --reverse: 701 Mbps
  • iperf3 -c $SERVER_IP --bidir: 939 Mbps up, 235 Mbps down

(Be sure to test all interfaces, noting any that are non-functional.)

GPU

glmark2

glmark2-es2 / glmark2-es2-wayland results:

=======================================================
    glmark2 2023.01
=======================================================
    OpenGL Information
    GL_VENDOR:      Imagination Technologies
    GL_RENDERER:    PowerVR A-Series AXM-8-256
    GL_VERSION:     OpenGL ES 3.2 build 24.2@6643903
    Surface Config: buf=32 r=8 g=8 b=8 a=8 depth=24 stencil=0 samples=0
    Surface Size:   800x600 windowed
=======================================================
[build] use-vbo=false: FPS: 662 FrameTime: 1.512 ms
[build] use-vbo=true: FPS: 1690 FrameTime: 0.592 ms
[texture] texture-filter=nearest: FPS: 1736 FrameTime: 0.576 ms
[texture] texture-filter=linear: FPS: 1711 FrameTime: 0.585 ms
[texture] texture-filter=mipmap: FPS: 1746 FrameTime: 0.573 ms
[shading] shading=gouraud: FPS: 1554 FrameTime: 0.644 ms
[shading] shading=blinn-phong-inf: FPS: 1454 FrameTime: 0.688 ms
[shading] shading=phong: FPS: 1446 FrameTime: 0.692 ms
[shading] shading=cel: FPS: 1422 FrameTime: 0.703 ms
[bump] bump-render=high-poly: FPS: 989 FrameTime: 1.012 ms
[bump] bump-render=normals: FPS: 1679 FrameTime: 0.596 ms
[bump] bump-render=height: FPS: 1674 FrameTime: 0.597 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 1165 FrameTime: 0.859 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 584 FrameTime: 1.713 ms
[pulsar] light=false:quads=5:texture=false: FPS: 1724 FrameTime: 0.580 ms
[desktop] blur-radius=5:effect=blur:passes=1:separable=true:windows=4: FPS: 569 FrameTime: 1.759 ms
[desktop] effect=shadow:windows=4: FPS: 1180 FrameTime: 0.848 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 227 FrameTime: 4.419 ms
[buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 224 FrameTime: 4.473 ms
[buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 359 FrameTime: 2.788 ms
[ideas] speed=duration: FPS: 680 FrameTime: 1.471 ms
[jellyfish] <default>: FPS: 998 FrameTime: 1.003 ms
[terrain] <default>: FPS: 112 FrameTime: 9.000 ms
[shadow] <default>: FPS: 1121 FrameTime: 0.893 ms
[refract] <default>: FPS: 167 FrameTime: 5.998 ms
[conditionals] fragment-steps=0:vertex-steps=0: FPS: 1744 FrameTime: 0.573 ms
[conditionals] fragment-steps=5:vertex-steps=0: FPS: 1762 FrameTime: 0.568 ms
[conditionals] fragment-steps=0:vertex-steps=5: FPS: 1709 FrameTime: 0.585 ms
[function] fragment-complexity=low:fragment-steps=5: FPS: 1787 FrameTime: 0.560 ms
[function] fragment-complexity=medium:fragment-steps=5: FPS: 1769 FrameTime: 0.565 ms
[loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 1619 FrameTime: 0.618 ms
[loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 1863 FrameTime: 0.537 ms
[loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 1876 FrameTime: 0.533 ms
=======================================================
                                  glmark2 Score: 1241 
=======================================================

GravityMark

GravityMark results: DNF (no RISC-V version available)

Note: These benchmarks require an active display on the device. Not all devices may be able to run glmark2-es2, so in that case, make a note and move on!

Ollama

ollama LLM model inference results:

System CPU/GPU Model Eval Rate Power (Peak)
HiFive Premier P550 (4-core RISC-V) CPU deepseek-r1:1.5b 0.49 Tokens/s 13.5 W
HiFive Premier P550 (4-core RISC-V) CPU llama3.2:3b 0.24 Tokens/s 13.5 W
HiFive Premier P550 (4-core RISC-V) CPU llama3.1:8b 0.10 Tokens/s 13.6 W
HiFive Premier P550 (4-core RISC-V) CPU llama2:13b 0.11 Tokens/s 13.6 W

Memory

tinymembench results:

Click to expand memory benchmark result 
tinymembench v0.4.10 (simple benchmark for memory throughput and latency)

==========================================================================
== Memory bandwidth tests                                               ==
==                                                                      ==
== Note 1: 1MB = 1000000 bytes                                          ==
== Note 2: Results for 'copy' tests show how many bytes can be          ==
==         copied per second (adding together read and writen           ==
==         bytes would have provided twice higher numbers)              ==
== Note 3: 2-pass copy means that we are using a small temporary buffer ==
==         to first fetch data into it, and only then write it to the   ==
==         destination (source -> L1 cache, L1 cache -> destination)    ==
== Note 4: If sample standard deviation exceeds 0.1%, it is shown in    ==
==         brackets                                                     ==
==========================================================================

 C copy backwards                                     :   1048.6 MB/s (1.8%)
 C copy backwards (32 byte blocks)                    :   2016.1 MB/s (3.8%)
 C copy backwards (64 byte blocks)                    :   2754.2 MB/s (3.9%)
 C copy                                               :   2837.9 MB/s (3.2%)
 C copy prefetched (32 bytes step)                    :   2816.7 MB/s (2.4%)
 C copy prefetched (64 bytes step)                    :   2841.9 MB/s (3.6%)
 C 2-pass copy                                        :   1988.3 MB/s (2.9%)
 C 2-pass copy prefetched (32 bytes step)             :   1098.6 MB/s (2.1%)
 C 2-pass copy prefetched (64 bytes step)             :   1098.7 MB/s (3.2%)
 C fill                                               :  10089.3 MB/s
 C fill (shuffle within 16 byte blocks)               :  10082.1 MB/s
 C fill (shuffle within 32 byte blocks)               :  10086.7 MB/s
 C fill (shuffle within 64 byte blocks)               :  10081.9 MB/s
 ---
 standard memcpy                                      :   1126.1 MB/s (2.0%)
 standard memset                                      :  10086.3 MB/s

==========================================================================
== Memory latency test                                                  ==
==                                                                      ==
== Average time is measured for random memory accesses in the buffers   ==
== of different sizes. The larger is the buffer, the more significant   ==
== are relative contributions of TLB, L1/L2 cache misses and SDRAM      ==
== accesses. For extremely large buffer sizes we are expecting to see   ==
== page table walk with several requests to SDRAM for almost every      ==
== memory access (though 64MiB is not nearly large enough to experience ==
== this effect to its fullest).                                         ==
==                                                                      ==
== Note 1: All the numbers are representing extra time, which needs to  ==
==         be added to L1 cache latency. The cycle timings for L1 cache ==
==         latency can be usually found in the processor documentation. ==
== Note 2: Dual random read means that we are simultaneously performing ==
==         two independent memory accesses at a time. In the case if    ==
==         the memory subsystem can't handle multiple outstanding       ==
==         requests, dual random read has the same timings as two       ==
==         single reads performed one after another.                    ==
==========================================================================

block size : single random read / dual random read, [MADV_NOHUGEPAGE]
      1024 :    0.0 ns          /     0.0 ns 
      2048 :    0.0 ns          /     0.0 ns 
      4096 :    0.0 ns          /     0.0 ns 
      8192 :    0.0 ns          /     0.0 ns 
     16384 :    0.0 ns          /     0.0 ns 
     32768 :    0.0 ns          /     0.0 ns 
     65536 :    3.3 ns          /     5.0 ns 
    131072 :    5.0 ns          /     6.9 ns 
    262144 :    9.5 ns          /    13.7 ns 
    524288 :   20.2 ns          /    28.5 ns 
   1048576 :   25.8 ns          /    34.0 ns 
   2097152 :   28.8 ns          /    36.2 ns 
   4194304 :   54.4 ns          /    81.6 ns 
   8388608 :  125.4 ns          /   176.6 ns 
  16777216 :  164.9 ns          /   210.8 ns 
  33554432 :  186.9 ns          /   228.3 ns 
  67108864 :  201.8 ns          /   243.8 ns 

block size : single random read / dual random read, [MADV_HUGEPAGE]
      1024 :    0.0 ns          /     0.0 ns 
      2048 :    0.0 ns          /     0.0 ns 
      4096 :    0.0 ns          /     0.0 ns 
      8192 :    0.0 ns          /     0.0 ns 
     16384 :    0.0 ns          /     0.0 ns 
     32768 :    0.0 ns          /     0.0 ns 
     65536 :    3.3 ns          /     5.0 ns 
    131072 :    4.9 ns          /     6.8 ns 
    262144 :    6.2 ns          /     8.1 ns 
    524288 :   15.5 ns          /    21.0 ns 
   1048576 :   20.5 ns          /    24.8 ns 
   2097152 :   23.3 ns          /    25.9 ns 
   4194304 :   44.4 ns          /    61.6 ns 
   8388608 :  112.0 ns          /   153.4 ns 
  16777216 :  146.4 ns          /   179.3 ns 
  33554432 :  163.2 ns          /   186.9 ns 
  67108864 :  176.5 ns          /   196.2 ns

Core to Core Memory Latency

Image

sbc-bench results

Run sbc-bench and paste a link to the results here: https://0x0.st/8PA8.bin

Phoronix Test Suite

Results from pi-general-benchmark.sh:

  • pts/encode-mp3: N/A sec
  • pts/x264 4K: N/A fps
  • pts/x264 1080p: N/A fps
  • pts/phpbench: 81276
  • pts/build-linux-kernel (defconfig): 2488.688 sec

Metadata

Metadata

Assignees

No one assigned

    Labels

    testing in progress

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions