Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- Orange Pi Zero Plus with vendor's own Ubuntu Xenial arm64 image (64-bit kernel 3.10.65, settings limiting cpufreq to 1008 MHz and clocking DRAM at 624 MHz):
- root@Orangepi:~/tinymembench# ./tinymembench
- tinymembench v0.4.9 (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 : 851.7 MB/s (2.0%)
- C copy backwards (32 byte blocks) : 861.5 MB/s (1.2%)
- C copy backwards (64 byte blocks) : 867.8 MB/s (1.1%)
- C copy : 861.7 MB/s (1.2%)
- C copy prefetched (32 bytes step) : 693.5 MB/s
- C copy prefetched (64 bytes step) : 775.7 MB/s (0.5%)
- C 2-pass copy : 863.1 MB/s (0.5%)
- C 2-pass copy prefetched (32 bytes step) : 646.7 MB/s
- C 2-pass copy prefetched (64 bytes step) : 344.1 MB/s (0.3%)
- C fill : 2037.2 MB/s (0.6%)
- C fill (shuffle within 16 byte blocks) : 2037.8 MB/s
- C fill (shuffle within 32 byte blocks) : 2038.4 MB/s (0.6%)
- C fill (shuffle within 64 byte blocks) : 2036.8 MB/s
- ---
- standard memcpy : 887.9 MB/s (0.6%)
- standard memset : 2037.9 MB/s (0.6%)
- ---
- NEON LDP/STP copy : 871.7 MB/s (0.8%)
- NEON LDP/STP copy pldl2strm (32 bytes step) : 670.7 MB/s (0.6%)
- NEON LDP/STP copy pldl2strm (64 bytes step) : 781.7 MB/s (0.3%)
- NEON LDP/STP copy pldl1keep (32 bytes step) : 932.4 MB/s
- NEON LDP/STP copy pldl1keep (64 bytes step) : 934.4 MB/s (0.5%)
- NEON LD1/ST1 copy : 860.8 MB/s (1.2%)
- NEON STP fill : 2035.9 MB/s (2.0%)
- NEON STNP fill : 1772.4 MB/s (3.5%)
- ARM LDP/STP copy : 871.1 MB/s (0.7%)
- ARM STP fill : 2036.1 MB/s (0.6%)
- ARM STNP fill : 1780.6 MB/s (2.1%)
- ==========================================================================
- == Framebuffer read tests. ==
- == ==
- == Many ARM devices use a part of the system memory as the framebuffer, ==
- == typically mapped as uncached but with write-combining enabled. ==
- == Writes to such framebuffers are quite fast, but reads are much ==
- == slower and very sensitive to the alignment and the selection of ==
- == CPU instructions which are used for accessing memory. ==
- == ==
- == Many x86 systems allocate the framebuffer in the GPU memory, ==
- == accessible for the CPU via a relatively slow PCI-E bus. Moreover, ==
- == PCI-E is asymmetric and handles reads a lot worse than writes. ==
- == ==
- == If uncached framebuffer reads are reasonably fast (at least 100 MB/s ==
- == or preferably >300 MB/s), then using the shadow framebuffer layer ==
- == is not necessary in Xorg DDX drivers, resulting in a nice overall ==
- == performance improvement. For example, the xf86-video-fbturbo DDX ==
- == uses this trick. ==
- ==========================================================================
- NEON LDP/STP copy (from framebuffer) : 172.1 MB/s (0.2%)
- NEON LDP/STP 2-pass copy (from framebuffer) : 164.4 MB/s (2.5%)
- NEON LD1/ST1 copy (from framebuffer) : 45.6 MB/s (0.8%)
- NEON LD1/ST1 2-pass copy (from framebuffer) : 45.0 MB/s (0.1%)
- ARM LDP/STP copy (from framebuffer) : 89.9 MB/s (0.4%)
- ARM LDP/STP 2-pass copy (from framebuffer) : 87.6 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
- 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 : 6.9 ns / 11.7 ns
- 131072 : 10.7 ns / 16.3 ns
- 262144 : 13.7 ns / 19.4 ns
- 524288 : 74.2 ns / 99.8 ns
- 1048576 : 122.2 ns / 177.3 ns
- 2097152 : 168.2 ns / 212.1 ns
- 4194304 : 184.3 ns / 229.6 ns
- 8388608 : 197.1 ns / 246.4 ns
- 16777216 : 205.9 ns / 248.3 ns
- 33554432 : 210.5 ns / 250.2 ns
- 67108864 : 218.7 ns / 255.3 ns
- root@Orangepi:~/tinymembench# /bin/bash /tmp/sysbench.sh
- 480: execution time (avg/stddev): 77.1597/0.00
- 648: execution time (avg/stddev): 56.9540/0.00
- 720: execution time (avg/stddev): 51.3270/0.00
- 816: execution time (avg/stddev): 45.2233/0.00
- 912: execution time (avg/stddev): 40.4356/0.00
- 1008: execution time (avg/stddev): 36.5771/0.00
- 1104: execution time (avg/stddev): 36.5767/0.00
- 1152: execution time (avg/stddev): 36.5724/0.00
- 1200: execution time (avg/stddev): 36.5786/0.00
- 480: execution time (avg/stddev): 19.5235/0.01
- 648: execution time (avg/stddev): 14.4798/0.01
- 720: execution time (avg/stddev): 13.0305/0.00
- 816: execution time (avg/stddev): 11.4151/0.01
- 912: execution time (avg/stddev): 10.3376/0.00
- 1008: execution time (avg/stddev): 9.2395/0.01
- 1104: execution time (avg/stddev): 9.2420/0.01
- 1152: execution time (avg/stddev): 9.2317/0.01
- 1200: execution time (avg/stddev): 9.2291/0.01
- root@Orangepi:~/tinymembench# 7zr b
- 7-Zip (A) 9.20 Copyright (c) 1999-2010 Igor Pavlov 2010-11-18
- p7zip Version 9.20 (locale=C,Utf16=off,HugeFiles=on,4 CPUs)
- RAM size: 468 MB, # CPU hardware threads: 4
- RAM usage: 434 MB, # Benchmark threads: 4
- Dict Compressing | Decompressing
- Speed Usage R/U Rating | Speed Usage R/U Rating
- KB/s % MIPS MIPS | KB/s % MIPS MIPS
- 22: 1324 281 457 1288 | 36532 382 862 3296
- 23: 1319 291 462 1344 | 35139 373 861 3215
- Killed
- for i in 128 192 256 ; do openssl speed -elapsed -evp aes-${i}-cbc ; done
- type 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
- aes-128-cbc 126688.02k 339073.43k 566900.48k 704829.78k 758764.89k
- aes-192-cbc 118457.32k 291691.16k 453099.78k 539112.45k 570556.42k
- aes-256-cbc 113399.49k 262801.51k 386358.36k 448731.82k 462544.90k
- root@Orangepi:~/StabilityTester# ./stabilityTester.sh
- Testing frequency 648000
- Cooling down CPU Freq: 648000 CPU Core: 1200000
- Testing frequency 720000
- Cooling down CPU Freq: 720000 CPU Core: 1200000
- Testing frequency 816000
- Cooling down CPU Freq: 816000 CPU Core: 1200000
- Testing frequency 912000
- Cooling down CPU Freq: 912000 CPU Core: 1200000
- Testing frequency 1008000
- Cooling down CPU Freq: 1008000 CPU Core: 1200000
- Testing frequency 1104000
- ./stabilityTester.sh: line 55: echo: write error: Invalid argument
- root@Orangepi:~/StabilityTester# cat /sys/devices/1c62000.dramfreq/devfreq/dramfreq/cur_freq
- 624000
- root@Orangepi:~/StabilityTester# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq
- 1008000
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement