API tools faq
paste
xlazom00

Untitled

xlazom00
Mar 13th, 2023
297
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 7.22 KB | None | 0 0
raw download clone embed print report
  1. misko@uefi-arm64:~$ sudo /bin/bash ./sbc-bench.sh -r
  2. Starting to examine hardware/software for review purposes...
  3.  
  4. Average load and/or CPU utilization too high (too much background activity). Waiting...
  5.  
  6. Too busy for benchmarking: 13:55:10 up 10 min, 2 users, load average: 0.56, 0.82, 0.52, cpu: 11%
  7. Too busy for benchmarking: 13:55:15 up 10 min, 2 users, load average: 0.51, 0.81, 0.52, cpu: 0%
  8. Too busy for benchmarking: 13:55:20 up 10 min, 2 users, load average: 0.47, 0.80, 0.52, cpu: 0%
  9. Too busy for benchmarking: 13:55:25 up 10 min, 2 users, load average: 0.43, 0.78, 0.51, cpu: 0%
  10. Too busy for benchmarking: 13:55:30 up 10 min, 2 users, load average: 0.40, 0.77, 0.51, cpu: 0%
  11. Too busy for benchmarking: 13:55:35 up 10 min, 2 users, load average: 0.37, 0.76, 0.51, cpu: 0%
  12. Too busy for benchmarking: 13:55:40 up 10 min, 2 users, load average: 0.34, 0.74, 0.50, cpu: 0%
  13.  
  14. sbc-bench v0.9.38
  15.  
  16. Installing needed tools: Done.
  17. Checking cpufreq OPP. Done.
  18. Executing tinymembench. Done.
  19. Executing RAM latency tester. Done.
  20. Executing OpenSSL benchmark. Done.
  21. Executing 7-zip benchmark. Done.
  22. Throttling test: heating up the device, 5 more minutes to wait. Done.
  23. Checking cpufreq OPP again. Done (15 minutes elapsed).
  24.  
  25. Results validation:
  26.  
  27. * No mismatch between advertised and measured max CPU clockspeed
  28. * Background activity (%system) OK
  29. * No throttling
  30.  
  31. Full results uploaded to http://ix.io/4qIj
  32.  
  33.  
  34.  
  35.  
  36.  
  37. # Microsoft Devkit 2023
  38.  
  39. Tested with sbc-bench v0.9.38 on Mon, 13 Mar 2023 14:10:27 +0100. Full info: [http://ix.io/4qIj](http://ix.io/4qIj)
  40.  
  41. ### General information:
  42.  
  43. The CPU features 2 clusters of different core types:
  44.  
  45. Snapdragon 449 rev 1.1, Qualcomm Snapdragon 8cx Gen 3, Kernel: aarch64, Userland: arm64
  46.  
  47. CPU sysfs topology (clusters, cpufreq members, clockspeeds)
  48. cpufreq min max
  49. CPU cluster policy speed speed core type
  50. 0 0 0 300 2438 Cortex-A78C / r0p0
  51. 1 0 0 300 2438 Cortex-A78C / r0p0
  52. 2 0 0 300 2438 Cortex-A78C / r0p0
  53. 3 0 0 300 2438 Cortex-A78C / r0p0
  54. 4 0 4 826 2995 Cortex-X1C / r0p0
  55. 5 0 4 826 2995 Cortex-X1C / r0p0
  56. 6 0 4 826 2995 Cortex-X1C / r0p0
  57. 7 0 4 826 2995 Cortex-X1C / r0p0
  58.  
  59. 31420 KB available RAM
  60.  
  61. ### Governors/policies (performance vs. idle consumption):
  62.  
  63. Original governor settings:
  64.  
  65. cpufreq-policy0: performance / 2438 MHz (conservative ondemand userspace powersave performance schedutil / 300 403 499 595 691 806 902 1018 1114 1210 1325 1440 1555 1670 1786 1882 1997 2112 2227 2342 2438)
  66. cpufreq-policy4: performance / 2995 MHz (conservative ondemand userspace powersave performance schedutil / 826 941 1056 1171 1286 1402 1517 1632 1747 1862 1978 2074 2170 2285 2400 2496 2592 2688 2803 2899 2995)
  67.  
  68. Tuned governor settings:
  69.  
  70. cpufreq-policy0: performance / 2438 MHz
  71. cpufreq-policy4: performance / 2995 MHz
  72.  
  73. Status of performance related policies found below /sys:
  74.  
  75. /sys/module/pcie_aspm/parameters/policy: default [performance] powersave powersupersave
  76.  
  77. ### Clockspeeds (idle vs. heated up):
  78.  
  79. Before at 35.2°C:
  80.  
  81. cpu0-cpu3 (Cortex-A78C): OPP: 2438, Measured: 2433
  82. cpu4-cpu7 (Cortex-X1C): OPP: 2995, Measured: 2991
  83.  
  84. After at 59.1°C:
  85.  
  86. cpu0-cpu3 (Cortex-A78C): OPP: 2438, Measured: 2433
  87. cpu4-cpu7 (Cortex-X1C): OPP: 2995, Measured: 2991
  88.  
  89. ### Performance baseline
  90.  
  91. * cpu0 (Cortex-A78C): memcpy: 15206.9 MB/s, memchr: 23040.3 MB/s, memset: 38607.5 MB/s
  92. * cpu4 (Cortex-X1C): memcpy: 17704.5 MB/s, memchr: 27147.2 MB/s, memset: 42187.9 MB/s
  93. * cpu0 (Cortex-A78C) 16M latency: 64.26 47.70 68.82 47.89 63.67 36.73 31.92 51.79
  94. * cpu4 (Cortex-X1C) 16M latency: 44.93 30.58 39.82 30.58 41.85 25.39 27.55 48.23
  95. * 7-zip MIPS (3 consecutive runs): 35268, 35481, 35392 (35380 avg), single-threaded: 4278
  96. * `aes-256-cbc 822719.49k 1181035.69k 1317447.08k 1359060.31k 1372359.34k 1373170.35k (Cortex-A78C)`
  97. * `aes-256-cbc 1161465.83k 1541332.95k 1644596.65k 1683085.31k 1693589.50k 1694493.35k (Cortex-X1C)`
  98.  
  99. ### Storage devices:
  100.  
  101. * 476.9GB "KBG40ZNS512G BG4A KIOXIA" SSD as /dev/nvme0: Speed 8GT/s, Width x4, 0% worn out, drive temp: 51°C
  102. * 59.5GB "Realtek Card Reader" as /dev/sda: USB, Driver=usb-storage, 480Mbps (capable of 12Mbps, 480Mbps, 5Gbps)
  103.  
  104. ### Challenging filesystems:
  105.  
  106. The following partitions are NTFS: nvme0n1p4
  107.  
  108. When this OS uses FUSE/userland methods to access NTFS filesystems performance
  109. will be significantly harmed or at least likely be bottlenecked by maxing out
  110. one or more CPU cores. It is highly advised when benchmarking with any NTFS to
  111. monitor closely CPU utilization or better switch to a 'Linux native' filesystem
  112. like ext4 since representing 'storage performance' a lot more than 'somewhat
  113. dealing with a foreign filesystem' as with NTFS.
  114.  
  115. ### Software versions:
  116.  
  117. * Ubuntu 22.10
  118. * Build scripts: [email protected]:armbian/build.git, 23.02.0-trunk, UEFI arm64, uefi-arm64, arm64
  119. * Compiler: /usr/bin/gcc (Ubuntu 12.2.0-3ubuntu1) 12.2.0 / aarch64-linux-gnu
  120.  
  121. ### Kernel info:
  122.  
  123. * `/proc/cmdline: BOOT_IMAGE=/boot/vmlinuz-6.2.2-arm64 root=UUID=0a2efc89-1705-459e-ad2b-a6da7fe27af3 ro console=tty1 clk_ignore_unused efi=novamap earlycon=efifb console=efifb loglevel=9`
  124. * Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
  125. * Vulnerability Spectre v1: Mitigation; __user pointer sanitization
  126. * Vulnerability Spectre v2: Mitigation; CSV2, BHB
  127. * Kernel 6.2.2-arm64 / CONFIG_HZ=1000
  128.  
  129. All known settings adjusted for performance. Device now ready for benchmarking.
  130. Once finished stop with [ctrl]-[c] to get info about throttling, frequency cap
  131. and too high background activity all potentially invalidating benchmark scores.
  132. All changes with storage and PCIe devices as well as suspicious dmesg contents
  133. will be reported too.
  134.  
  135. Time big.LITTLE load %cpu %sys %usr %nice %io %irq Temp
  136. 14:10:30: 2995/2438MHz 5.74 33% 0% 31% 0% 1% 0% 50.7°C
  137. 14:11:30: 2995/2438MHz 2.11 0% 0% 0% 0% 0% 0% 45.3°C
  138. 14:12:30: 2995/2438MHz 0.77 0% 0% 0% 0% 0% 0% 42.9°C
  139. 14:13:30: 2995/2438MHz 0.28 0% 0% 0% 0% 0% 0% 41.9°C
  140. 14:14:30: 2995/2438MHz 0.15 0% 0% 0% 0% 0% 0% 41.9°C
  141. 14:15:30: 2995/2438MHz 0.05 0% 0% 0% 0% 0% 0% 41.9°C
  142. 14:16:30: 2995/2438MHz 0.02 0% 0% 0% 0% 0% 0% 41.9°C
  143. 14:17:30: 2995/2438MHz 0.00 0% 0% 0% 0% 0% 0% 41.9°C
  144. 14:18:31: 2995/2438MHz 0.03 0% 0% 0% 0% 0% 0% 41.6°C
  145. 14:19:31: 2995/2438MHz 0.01 0% 0% 0% 0% 0% 0% 41.3°C
  146. 14:20:31: 2995/2438MHz 0.00 0% 0% 0% 0% 0% 0% 40.9°C
  147.  
  148. 14:21:31: 2995/2438MHz 0.00 0% 0% 0% 0% 0% 0% 40.9°C
  149. ^C
  150.  
  151. Cleaning up. Done.
  152. Checking cpufreq OPP again. Done.
  153.  
  154. Clockspeeds now at 41.3°C:
  155.  
  156. cpu0-cpu3 (Cortex-A78C): OPP: 2438, Measured: 2433
  157. cpu4-cpu7 (Cortex-X1C): OPP: 2995, Measured: 2991
  158.  
  159. Results validation:
  160.  
  161. * No mismatch between advertised and measured max CPU clockspeed
  162. * Background activity (%system) OK
  163. * No throttling
Advertisement
Add Comment
Please, Sign In to add comment
Public Pastes
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!