1234
返回列表 發帖
MSI_R9_280_Gami

Rank: 2Rank: 2
1# 跳轉到 » 倒序看帖
打印
tT
發表於 2026-1-23 00:01 | 只看該作者

[實用軟件] super pi & sysbench

本帖最後由 MSI_R9_280_Gami 於 2026-1-23 00:54 編輯

super_pi同pi run唔到，似乎因為 ./Makefile入面（found @ a github page）

CC = gcc
#Set this to the cpu to target
-march=i686改做x86-64或amd64，或3代ivybridge 5代broadwell

russo-altlinux apt-get install gcc，make all都唔得，compiled failed。唔識Using the musl-gcc wrapper to compile musl packages所以放棄。void musl唔同alpinelinux唔係用純正musl，底層都係調用glibc2

super_pi係2007年左右既windows software，俾人port左做linux version，不過編譯架構太舊，binary ball要重新compile過 from source code。super_pi係計算圓周率測試CPU單綫程性能，多核並行用sysbench
收藏 分享

0

MSI_R9_280_Gami

Rank: 2Rank: 2
2#
發表於 2026-1-23 00:01 | 只看該作者
本帖最後由 MSI_R9_280_Gami 於 2026-1-23 17:26 編輯

p1 p2 installation of sysbench on void musl & debian
p3 sysbench cpu 計質數（素數）結果 on debian 3代ivybridge X79
p4 sysbench cpu 計質數結果 on void musl 3代ivybridge X79
p5 sysbench cpu 計質數結果 on void musl 5代broadwell Z97，似乎因為高頻率，加上128MB vram for iGP was set as L4 cache，E3-1285L v4 4C8T 比E5-2697 v2 12C24T快左一倍。musl package亦run得比glibc2 libc6快

#sysbench cpu --cpu-max-prime=? --threads=? --events=1 run

--events=?，run幾多次質數篩查。另一個替代，交替性選項係--time=?，規定時間內run到幾多次

--cpu-max-prime=?質數範圍，我選左5000000或50000000。--threads=?，4C8T定12C24T，綫程數 threads-counts

两台服务器进行CPU性能比较，当素数上限和线程数一致时：

相同时间，比较events；
相同events，比较时间；
时间和events都相同，比较标准差。

https://imgur.com/a/2TeKPJ2










https://zhuanlan.zhihu.com/p/691766392
使用SuperPi和Sysbench测试CPU性能 - 知乎
https://zhuanlan.zhihu.com/p/689586190
分布式存储性能调优 - sysbench内存带宽测试详解 - 知乎
https://www.zhihu.com/question/604623646/answer/3064216963
如何评估计算机的算力性能？ - 知乎
https://blog.csdn.net/weixin_43494937/article/details/106997308
Super PI linux 版本最新可用_super pi arm-CSDN博客

sysbench除左run cpu test仲可以test numa bandwidth（而不是SMP），zhihu大陸仔部機係server板雙路3647，6通道 x2：

6-channels DDR4 6-sticks=CPU1=CPU2=6-channels DDR4 6-sticks
6=1=QPI / UPI=1=6

所以佢提到「本地内存带宽测试」同「远程内存带宽测试」、「混合内存带宽测试」。XEONs之間通過北橋QPI / UPI互聯。（DMI係南橋同北橋互聯），即係一個CPU訪問另一個CPU的DDR4 recc

TOP

MSI_R9_280_Gami

Rank: 2Rank: 2
3#
發表於 2026-1-23 00:32 | 只看該作者
as a result

  1. void@void ~> sysbench cpu --cpu-max-prime=5000000 --threads=8 --events=1 run
  2. sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

  4. Running the test with following options:
  5. Number of threads: 8
  6. Initializing random number generator from current time


  9. Prime numbers limit: 5000000

  11. Initializing worker threads...

  13. Threads started!

  15. CPU speed:
  16.     events per second:     0.22

  18. General statistics:
  19.     total time:                          4.4461s
  20.     total number of events:              1

  22. Latency (ms):
  23.          min:                                 4446.04
  24.          avg:                                 4446.04
  25.          max:                                 4446.04
  26.          95th percentile:                     4437.27
  27.          sum:                                 4446.04

  29. Threads fairness:
  30.     events (avg/stddev):           0.1250/0.33
  31.     execution time (avg/stddev):   0.5558/1.47
複製代碼

TOP

MSI_R9_280_Gami

Rank: 2Rank: 2
4#
發表於 2026-1-23 14:59 | 只看該作者
本帖最後由 MSI_R9_280_Gami 於 2026-1-24 02:31 編輯

sysbench memory子命令用于测试内存性能，包括时延和带宽数据，本文只关注带宽数据。sysbench memory有6个可以配置的参数：

--memory-block-size=SIZE， 指定分配的buffer大小，不要分配得太小，否则会出现数据全从CPU缓存命中，达不到测试内存带宽的目的。本文测试用例中每个sysbench进程都设置为1GB。

--memory-total-size=SIZE，指定总共要测试多少数据量，例如100GB，那么便会循环操作--memory-block-size（例如1G）100次。每次操作以8字节对1GB buffer进行读写，直到1GB读写完毕。

--memory-scope=STRING，指定访问本地内存还是全局内存。本文的测试中使用numactl控制了内存访问范围，所以这个参数可以直接忽略。

--memory-hugetlb[=on|off]，是否从分配大页内存。本文的测试中设置为on。

--memory-oper=STRING，指定read/wirte操作，默认是write。

--memory-access-mode=STRING，指定seq/rnd，即顺序或随机，默认是seq。

void@void ~> sysbench memory help
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

memory options:
  --memory-block-size=SIZE    size of memory block for test [1K]
  --memory-total-size=SIZE    total size of data to transfer [100G]
  --memory-scope=STRING       memory access scope {global,local} [global]
  --memory-hugetlb[=on|off]   allocate memory from HugeTLB pool [off]
  --memory-oper=STRING        type of memory operations {read, write, none} [write]
  --memory-access-mode=STRING memory access mode {seq,rnd} [seq]
void @void ~ > sysbench  memory --memory-block-size=5GB --memory-total-size=200GB run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

FATAL: Invalid value for memory-block-size: 5GB

void@void ~ [1]> sysbench memory --memory-block-size=1GB --memory-total-size=200GB run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 1048576KiB
  total size: 204800MiB
  operation: write
  scope: global

Initializing worker threads...

Threads started!

Total operations: 103 (   10.20 per second)

105472.00 MiB transferred (10445.16 MiB/sec)


General statistics:
    total time:                          10.0961s
    total number of events:              103

Latency (ms):
         min:                                   96.24
         avg:                                   98.01
         max:                                  101.50
         95th percentile:                       99.33
         sum:                                10095.44

Threads fairness:
    events (avg/stddev):           103.0000/0.00
    execution time (avg/stddev):   10.0954/0.00

void@void ~> sysbench memory --memory-block-size=2GB --memory-total-size=200GB run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 2097152KiB
  total size: 204800MiB
  operation: write
  scope: global

Initializing worker threads...

Threads started!

Total operations: 51 (    5.04 per second)

104448.00 MiB transferred (10314.30 MiB/sec)


General statistics:
    total time:                          10.1249s
    total number of events:              51

Latency (ms):
         min:                                  193.10
         avg:                                  198.52
         max:                                  204.63
         95th percentile:                      204.11
         sum:                                10124.50

Threads fairness:
    events (avg/stddev):           51.0000/0.00
    execution time (avg/stddev):   10.1245/0.00

void@void ~> sysbench memory --memory-block-size=3GB --memory-total-size=200GB run
sysbench 1.0.20 (using system LuaJIT 2.1.0-beta3)

FATAL: Invalid value for memory-block-size: 3GB

以上乃E3-1285L v4 + GA-Z97-HD3 REV:2.0 + DDR3-2000mt/s 8GB x4 2通道4條 CL11-11-11-28

以下乃E5-2697 v2 + ASUS Sabertooth X79 + DDR3-1866mt/s 8GB x8 4通道8條 CL10-10-10-27

由于Xeon E5 TDP限制，無法超頻CPU & memory，所以我的ram步战士2200mt/s CL13-13-13-33（uefi睇係CL11，memtest86+睇係CL13）只可以stay @1866mt/s CL10-10-10-27 on Sabertooth X79 & Xeon E5-2697 v2，好奇怪，X79 4-ch得8.5GB/s，反而Z97 2-ch有10.3GB/s


debian@debian:~/Videos/SuperPI-main$ sysbench memory --memory-block-size=2GB --memory-total-size=200GB run
sysbench 1.0.20 (using system LuaJIT 2.1.1700206165)

Running the test with following options:
Number of threads: 1
Initializing random number generator from current time


Running memory speed test with the following options:
  block size: 2097152KiB
  total size: 204800MiB
  operation: write
  scope: global

Initializing worker threads...

Threads started!

Total operations: 42 (    4.18 per second)

86016.00 MiB transferred (8550.65 MiB/sec)


General statistics:
    total time:                          10.0575s
    total number of events:              42

Latency (ms):
         min:                                  238.78
         avg:                                  239.45
         max:                                  246.02
         95th percentile:                      240.02
         sum:                                10056.70

Threads fairness:
    events (avg/stddev):           42.0000/0.00
    execution time (avg/stddev):   10.0567/0.00

TOP

MSI_R9_280_Gami

Rank: 2Rank: 2
5#
發表於 2026-1-23 15:47 | 只看該作者
本帖最後由 MSI_R9_280_Gami 於 2026-1-26 23:25 編輯

上左鬼佬壇問鬼佬，super_pi compilation方法est on below，worked on Void Musl & debian glibc2 libc6，人生第二次compiling成功，第一次係brave

binaric ball pi means it was compiled as -march=ivybridge，pi_css5 as broadwell

https://github.com/Fibonacci43/SuperPI?tab=readme-ov-file#build-on-linux
#xbps-install gcc
$cd ./;gcc -O -funroll-loops -fomit-frame-pointer pi_fftcs.c fftsg_h.c -lm -o pi_css5;./pi_css5

void@void ~/V/SuperPI-main> gcc -O -funroll-loops -fomit-frame-pointer pi_fftcs.c fftsg_h.c -lm -o pi_css5
void@void ~/V/SuperPI-main> /home/void/Videos/SuperPI-main/pi_css5
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave

Usage: /home/void/Videos/SuperPI-main/pi_css5 digits

Number of digits of pi to calculate?
100
initializing...
nfft= 32
radix= 10000
error_margin= 5.15588e-07
calculating 128 digits of PI...
AGM iteration
precision= 48: 0.00 sec
precision= 80: 0.00 sec
precision= 176: 0.00 sec
writing pi128.txt...
0.00 sec. (real time)
Hit RETURN to exit.


void@void ~/V/SuperPI-main [SIGINT]> /home/void/Videos/SuperPI-main/pi_css5 500000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 131072
radix= 10000
error_margin= 0.0029395
calculating 524288 digits of PI...
AGM iteration
precision= 48: 0.03 sec
precision= 80: 0.03 sec
precision= 176: 0.03 sec
precision= 352: 0.03 sec
precision= 688: 0.03 sec
precision= 1392: 0.03 sec
precision= 2784: 0.03 sec
precision= 5584: 0.03 sec
precision= 11168: 0.03 sec
precision= 22336: 0.03 sec
precision= 44688: 0.03 sec
precision= 89408: 0.03 sec
precision= 178816: 0.03 sec
precision= 357648: 0.03 sec
precision= 715312: 0.03 sec
writing pi524288.txt...
0.60 sec. (real time)

void@void ~/M/SuperPI-main> ./pi_css5 50000000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 16777216
radix= 10000
error_margin= 0.365078
calculating 67108864 digits of PI...
AGM iteration
precision= 48: 6.38 sec
precision= 80: 6.31 sec
precision= 176: 6.52 sec
precision= 352: 6.53 sec
precision= 688: 6.49 sec
precision= 1392: 6.52 sec
precision= 2784: 6.54 sec
precision= 5584: 6.42 sec
precision= 11168: 6.42 sec
precision= 22336: 6.47 sec
precision= 44688: 6.34 sec
precision= 89408: 6.31 sec
precision= 178816: 6.24 sec
precision= 357648: 6.25 sec
precision= 715312: 6.25 sec
precision= 1430640: 6.23 sec
precision= 2861280: 6.24 sec
precision= 5722592: 6.20 sec
precision= 11445200: 6.21 sec
precision= 22890416: 6.21 sec
precision= 45780848: 6.24 sec
precision= 91561728: 6.22 sec
writing pi67108864.txt...
156.92 sec. (real time)

#xbps-remove binutils-doc binutils-libs libdebuginfod binutils libgcc-devel libstdc++-devel libatomic libatomic-devel kernel-libc-headers musl-devel gcc unzip





debian@debian:~/Videos/SuperPI-main$ ./pi 10000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 4096
radix= 10000
error_margin= 7.41633e-05
calculating 16384 digits of PI...
AGM iteration
precision= 48: 0.00 sec
precision= 80: 0.00 sec
precision= 176: 0.00 sec
precision= 352: 0.00 sec
precision= 688: 0.00 sec
precision= 1392: 0.00 sec
precision= 2784: 0.00 sec
precision= 5584: 0.00 sec
precision= 11168: 0.00 sec
precision= 22336: 0.00 sec
writing pi16384.txt...
0.02 sec. (real time)
debian@debian:~/Videos/SuperPI-main$ ./pi 500000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 131072
radix= 10000
error_margin= 0.0029395
calculating 524288 digits of PI...
AGM iteration
precision= 48: 0.05 sec
precision= 80: 0.05 sec
precision= 176: 0.05 sec
precision= 352: 0.05 sec
precision= 688: 0.05 sec
precision= 1392: 0.05 sec
precision= 2784: 0.05 sec
precision= 5584: 0.05 sec
precision= 11168: 0.05 sec
precision= 22336: 0.05 sec
precision= 44688: 0.05 sec
precision= 89408: 0.05 sec
precision= 178816: 0.05 sec
precision= 357648: 0.05 sec
precision= 715312: 0.05 sec
writing pi524288.txt...
0.91 sec. (real time)


debian@debian:~/Videos/SuperPI-main$ ./pi 50000000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 16777216
radix= 10000
error_margin= 0.365078
calculating 67108864 digits of PI...
AGM iteration
precision= 48: 10.12 sec
precision= 80: 10.07 sec
precision= 176: 10.10 sec
precision= 352: 10.12 sec
precision= 688: 10.12 sec
precision= 1392: 10.12 sec
precision= 2784: 10.13 sec
precision= 5584: 10.10 sec
precision= 11168: 10.05 sec
precision= 22336: 10.10 sec
precision= 44688: 10.11 sec
precision= 89408: 10.10 sec
precision= 178816: 10.07 sec
precision= 357648: 10.06 sec
precision= 715312: 10.08 sec
precision= 1430640: 10.11 sec
precision= 2861280: 10.11 sec
precision= 5722592: 10.10 sec
precision= 11445200: 10.15 sec
precision= 22890416: 10.08 sec
precision= 45780848: 10.08 sec
precision= 91561728: 10.10 sec
writing pi67108864.txt...
250.08 sec. (real time)

intel turbo boost對core i5 i7 Xeon E3效能提升影響好大（亦有可能係Broadwell vram for iGP as L4 cache有關），E5限定TDP，DDR bandwidth同算力都差d

Super PI是一個計算圓周率的電腦程式，最大精確到小數點以後的3200萬位數。該程式採用高斯-勒讓德演算法，並且於1995年被日本電腦科學家金田康正使用在Windows環境下計算出精確到232次方的圓周率。

高斯-勒讓德演算法是一種用於計算圓周率（π）的演算法。它以迅速收斂著稱，只需25次迭代即可產生π的4500萬位正確數字。不過，它的缺點是主記憶體密集，因此有時它不如梅欽類公式使用廣泛。

該演算法具有二階收斂性，本質上說就是演算法每執行一步正確位數就會加倍。

Cache L1:         64 KB (per core)
Cache L2:         256 KB (per core)
Cache L3:         30 MB (shared)

E5-2697 v2似乎cache輸蝕俾E3-1285L v4的128MB L4 cache，難怪之前睇過有篇文話有d使用情況下5代CPU可以打平8代9代甚至10代U，類似AMD X3D咁大cache advantage。比起E5-4650 4657L 2697 v2，E3-1285L v4高時脈，大L4 cache，新架構新指令集，run super-pi計圓周率，sysbench計質數，memory bandwidth都有優勢

debian@debian:~/Music/SuperPI-main$ gcc -O -funroll-loops -fomit-frame-pointer pi_fftcs.c fftsg_h.c -lm -o pi-libc6-3ʳᵈivybridge
debian@debian:~/Music/SuperPI-main$ ./pi-libc6-3ʳᵈivybridge 300000000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 134217728
radix= 1000
error_margin= 0.0560284
calculating 402653184 digits of PI...
AGM iteration
precision= 36: 87.94 sec
precision= 84: 88.21 sec
precision= 168: 87.79 sec
precision= 348: 87.65 sec
precision= 696: 87.95 sec
precision= 1392: 88.12 sec
precision= 2784: 87.79 sec
precision= 5580: 87.51 sec
precision= 11160: 87.90 sec
precision= 22344: 87.93 sec
precision= 44688: 87.91 sec
precision= 89400: 88.27 sec
precision= 178812: 87.97 sec
precision= 357648: 87.70 sec
precision= 715308: 87.76 sec
precision= 1430628: 87.89 sec
precision= 2861280: 87.86 sec
precision= 5722584: 87.97 sec
precision= 11445192: 88.02 sec
precision= 22890408: 87.72 sec
precision= 45780852: 87.38 sec
precision= 91561716: 87.35 sec
precision= 183123468: 87.90 sec
precision= 366246960: 87.87 sec
precision= 732493944: 87.86 sec
writing pi402653184.txt...
2435.97 sec. (real time)


void@void ~/M/SuperPI-main> ./pi-voidMUSL-5ᵗʰbroadwell 300000000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 134217728
radix= 1000
error_margin= 0.0561774
calculating 402653184 digits of PI...
AGM iteration
precision= 36: 59.66 sec
precision= 84: 58.57 sec
precision= 168: 59.34 sec
precision= 348: 59.91 sec
precision= 696: 58.97 sec
precision= 1392: 58.43 sec
precision= 2784: 59.67 sec
precision= 5580: 59.88 sec
precision= 11160: 59.69 sec
precision= 22344: 59.08 sec
precision= 44688: 59.83 sec
precision= 89400: 58.58 sec
precision= 178812: 58.94 sec
precision= 357648: 59.26 sec
precision= 715308: 59.20 sec
precision= 1430628: 60.34 sec
precision= 2861280: 58.43 sec
precision= 5722584: 57.23 sec
precision= 11445192: 57.27 sec
precision= 22890408: 57.52 sec
precision= 45780852: 59.47 sec
precision= 91561716: 58.69 sec
precision= 183123468: 58.90 sec
precision= 366246960: 58.73 sec
precision= 732493944: 58.93 sec
writing pi402653184.txt...
1635.52 sec. (real time)


void@void ~/M/SuperPI-main> gcc -O -funroll-loops -fomit-frame-pointer pi_fftcs.c fftsg_h.c -lm -o pi-voidMUSL-3ʳᵈivybridge
void@void ~/M/SuperPI-main> ./pi-voidMUSL-3ʳᵈivybridge 300000000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 134217728
radix= 1000
error_margin= 0.0561774
calculating 402653184 digits of PI...
AGM iteration
precision= 36: 91.13 sec
precision= 84: 88.87 sec
precision= 168: 88.87 sec
precision= 348: 89.33 sec
precision= 696: 89.20 sec
precision= 1392: 89.02 sec
precision= 2784: 89.24 sec
precision= 5580: 89.20 sec
precision= 11160: 89.16 sec
precision= 22344: 89.04 sec
precision= 44688: 89.05 sec
precision= 89400: 89.03 sec
precision= 178812: 89.03 sec
precision= 357648: 89.02 sec
precision= 715308: 89.11 sec
precision= 1430628: 89.21 sec
precision= 2861280: 89.21 sec
precision= 5722584: 89.23 sec
precision= 11445192: 89.30 sec
precision= 22890408: 89.19 sec
precision= 45780852: 89.21 sec
precision= 91561716: 89.22 sec
precision= 183123468: 89.20 sec
precision= 366246960: 89.23 sec
precision= 732493944: 89.24 sec
writing pi402653184.txt...
2476.21 sec. (real time)

X79 + E5-2697 v2
debian@debian:~/Videos$ /home/debian/Videos/pi-libc6-3ʳᵈivybridge 1000000000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 268435456
radix= 1000
error_margin= 0.0882149
calculating 805306368 digits of PI...
AGM iteration
precision= 36: 192.16 sec
precision= 84: 188.46 sec
precision= 168: 188.19 sec
precision= 348: 188.84 sec
precision= 696: 190.10 sec
precision= 1392: 187.38 sec
precision= 2784: 187.40 sec
precision= 5580: 187.47 sec
precision= 11160: 187.48 sec
precision= 22344: 187.54 sec
precision= 44688: 187.46 sec
precision= 89400: 187.50 sec
precision= 178812: 187.55 sec
precision= 357648: 187.69 sec
precision= 715308: 187.79 sec
precision= 1430628: 190.35 sec
precision= 2861280: 187.80 sec
precision= 5722584: 188.33 sec
precision= 11445192: 187.73 sec
precision= 22890408: 187.60 sec
precision= 45780852: 187.78 sec
precision= 91561716: 187.50 sec
precision= 183123468: 187.72 sec
precision= 366246960: 191.09 sec
precision= 732493944: 187.33 sec
precision= 1464987924: 187.46 sec
writing pi805306368.txt...
5407.86 sec. (real time)

H81M-K + i7-4770 + 8GB DDR3-1600mt/s x2 VALUE SELECT CL11-11-11-30 altlinux
$make all=$gcc -Wall -pedantic -O -fomit-frame-pointer -funroll-loops -march=haswell -malign-double pi_fftcs.o fftsg_h.o -lm -static -o pi_css5，但會failed

[wery245@wery245 Music]$ ./pi-libc6-4ᵗʰhaswell 1000000000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 268435456
radix= 1000
error_margin= 0.0929832
calculating 805306368 digits of PI...
AGM iteration
precision= 36: 181.45 sec
precision= 84: 181.62 sec
precision= 168: 179.28 sec
precision= 348: 182.41 sec
precision= 696: 180.27 sec
precision= 1392: 182.02 sec
precision= 2784: 182.74 sec
precision= 5580: 185.80 sec
precision= 11160: 181.03 sec
precision= 22344: 176.05 sec
precision= 44688: 174.54 sec
precision= 89400: 172.71 sec
precision= 178812: 178.56 sec
precision= 357648: 179.75 sec
precision= 715308: 172.14 sec
precision= 1430628: 171.36 sec
precision= 2861280: 181.77 sec
precision= 5722584: 182.58 sec
precision= 11445192: 183.21 sec
precision= 22890408: 182.10 sec
precision= 45780852: 182.92 sec
precision= 91561716: 182.84 sec
precision= 183123468: 182.37 sec
precision= 366246960: 177.46 sec
precision= 732493944: 170.02 sec
precision= 1464987924: 176.45 sec
writing pi805306368.txt...
5325.24 sec. (real time)

同上 void musl
[void@void Music]$ ./pi-voidMUSL-4ᵗʰhaswell 1000000000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 268435456
radix= 1000
error_margin= 0.0882149
calculating 805306368 digits of PI...
AGM iteration
precision= 36: 145.07 sec
precision= 84: 144.53 sec
precision= 168: 145.11 sec
precision= 348: 142.62 sec
precision= 696: 142.96 sec
precision= 1392: 142.71 sec
precision= 2784: 144.68 sec
precision= 5580: 144.27 sec
precision= 11160: 144.80 sec
precision= 22344: 145.41 sec
precision= 44688: 143.26 sec
precision= 89400: 142.73 sec
precision= 178812: 142.91 sec
precision= 357648: 142.73 sec
precision= 715308: 142.98 sec
precision= 1430628: 142.87 sec
precision= 2861280: 143.06 sec
precision= 5722584: 142.82 sec
precision= 11445192: 143.80 sec
precision= 22890408: 143.07 sec
precision= 45780852: 142.97 sec
precision= 91561716: 143.07 sec
precision= 183123468: 142.96 sec
precision= 366246960: 142.86 sec
precision= 732493944: 143.09 sec
precision= 1464987924: 142.98 sec
writing pi805306368.txt...
4129.69 sec. (real time)


兩張奇怪的網圖

TOP

MSI_R9_280_Gami

Rank: 2Rank: 2
6#
發表於 2026-1-24 18:33 | 只看該作者
本帖最後由 MSI_R9_280_Gami 於 2026-1-28 09:30 編輯

原先想compile scim用cangjie5，發覺太複雜，唔曉搞，玩返zram

root@void ~# xbps-install zram
ERROR: Package 'zram' not found in repository pool.
root@void ~ [2]# xbps-install zramctl
ERROR: Package 'zramctl' not found in repository pool.
root@void ~ [2]# zramctl
root@void ~# cat /sys/block/zram0/comp_algorithm
cat: /sys/block/zram0/comp_algorithm: No such file or directory
root@void ~ [1]# cat /sys/block/zram/comp_algorithm
cat: /sys/block/zram/comp_algorithm: No such file or directory
root@void ~ [1]# zramctl -s 64GiB -a zstd lz4 /dev/zram0
zramctl: only one <device> at a time is allowed
root@void ~ [1]# zramctl -s 64GiB -a zstd /dev/zram0
zramctl: /dev/zram0: No such device
root@void ~ [1]# modprobe zram
root@void ~# cat /sys/block/zram0/comp_algorithm
lzo-rle [lzo] lz4 lz4hc zstd 842
root@void ~# zramctl -s 64GiB -a 842 /dev/zram0
root@void ~# mkswap -U clear /dev/zram0
Setting up swapspace version 1, size = 64 GiB (68719472640 bytes)
no label, UUID=00000000-0000-0000-0000-000000000000
root@void ~# swapon --priority 100 /dev/zram0
root@void ~# echo 1 > /sys/module/zswap/parameters/enabled
root@void ~# sysctl vm.swappiness=200
vm.swappiness = 200

a brief guide
specify algorithm, supported are:
lzo, lz4, lz4hc, deflate, 842 and zstd

1) Load Module:
modprobe zram num_devices=4
This creates 4 devices: /dev/zram{0,1,2,3}

9) Deactivate:
swapoff /dev/zram0
umount /dev/zram1❌

10) Reset:
Write any positive value to 'reset' sysfs node
echo 1 > /sys/block/zram0/reset
echo 1 > /sys/block/zram1/reset

This frees all the memory allocated for the given device and
resets the disksize to zero. You must set the disksize again
before reusing the device.

to detach

#swapoff /dev/zram0
#modprobe -r zram
#zramctl -r /dev/zram0
#echo 0>/sys/block/zram0/reset

apk add or del an input method
~ # apk add ibus ibus-rime
( 1/15) Installing dconf (0.49.0-r1)
( 2/15) Installing ibus (1.5.33-r0)
  Executing ibus-1.5.33-r0.post-install
( 3/15) Installing ibus-gtk3 (1.5.33-r0)
( 4/15) Installing ibus-pyc (1.5.33-r0)
( 5/15) Installing ibus-gtk4 (1.5.33-r0)
( 6/15) Installing gflags (2.2.2-r2)
( 7/15) Installing libunwind (1.8.1-r0)
( 8/15) Installing glog (0.7.1-r2)
( 9/15) Installing snappy (1.2.2-r0)
(10/15) Installing leveldb (1.23-r1)
(11/15) Installing libmarisa (0.2.6-r7)
(12/15) Installing opencc-libs (1.1.9-r2)
(13/15) Installing yaml-cpp (0.8.0-r1)
(14/15) Installing librime (1.15.0-r0)
(15/15) Installing ibus-rime (1.5.1-r0)
Executing busybox-1.37.0-r30.trigger
Executing glib-2.86.3-r0.trigger
Executing gtk-update-icon-cache-3.24.51-r0.trigger
Executing gtk+3.0-3.24.51-r0.trigger
Executing desktop-file-utils-0.28-r0.trigger
OK: 1258.3 MiB in 468 packages

奇怪，用唔到ibus-rime
apk del ibus-gtk3 ibus-gtk4 dconf ibus ibus-pyc ibus-rime librime yaml-cpp opencc-libs libmarisa leveldb snappy glog gflags libunwind

TOP

fakeman

Rank: 4Rank: 4Rank: 4Rank: 4
7#
發表於 2026-1-27 08:19 | 只看該作者
不如用 Prime95 算？

TOP

MSI_R9_280_Gami

Rank: 2Rank: 2
8#
發表於 2026-1-27 16:38 | 只看該作者
用過好多distros，但凡official repository有，咁係developers覺得有需要，compiled from source codes to binary & tested fine，然後將binary packages放入套件庫

好多distros收左sysbench，冇收到 pi / super_pi 及GIMPS-mprime，已經係答案

互聯網梅森質數大搜尋（英語：Great Internet Mersenne Prime Search，簡稱：GIMPS），是一個由志願者團隊協同運作的專案，從互聯網免費下載開放原始碼的Prime95和MPrime軟件來搜尋梅森質數。

Prime95的Linux及FreeBSD版本稱為MPrime。

TOP

fakeman

Rank: 4Rank: 4Rank: 4Rank: 4
9#
發表於 2026-1-29 02:36 | 只看該作者
用過好多distros，但凡official repository有，咁係developers覺得有需要，compiled from source codes to  ...
MSI_R9_280_Gami 發表於 2026-1-27 16:38



    我唔係講收唔收入去 distro 嘅問題，而係你都講本身個 super_pi 架構太舊（佢本身係 for 32bit system），要做改動先整到來用，Prime95 就直接得多

TOP

MSI_R9_280_Gami

Rank: 2Rank: 2
10#
發表於 2026-1-29 05:12 | 只看該作者
回復 9 #fakeman

ok fine

TOP

1234
返回列表