全志在线开发者论坛

yao0718

xtensa-hifidsp-gcc.tar.bz2
仅用于学习及研究目的，不提供任何保证，商业用途请联系全志购买candence explore

C库的头文件已经被替换为SDK中XCLIB的头文件，如要使用newlib请自行重新编译安装newlib

yao0718

自己写个RTEMS的bsp就可以了，大部分驱动在freebsd中有，稍微修改下添加到libbsd中可以直接工作

yao0718

xtensa-hifidsp-gcc.tar.bz2
仅用于学习及研究目的，不提供任何保证，商业用途请联系全志购买candence explore

C库的头文件已经被替换为SDK中XCLIB的头文件，如要使用newlib请自行重新编译安装newlib

yao0718

玩没问题，可以直接链接xcc的库文件（包含在R528的SDK中），这样可以解决math库中TIE指令和SIMD的问题，，用GCC直接编译的newlib的math库没有SIMD指令，具体版权问题需要咨询全志SDK中的二进制库文件在GCC中使用是否侵权

yao0718

ARM 双核SMP ，系统 rtems5，主频 1.08G

 [/] # coremark
2K performance run parameters for coremark.
CoreMark Size    : 666
Total ticks      : 12680
Total time (secs): 12.680000
Iterations/Sec   : 3154.574132
Iterations       : 40000
Compiler version : GCC7.5.0 20191114 (RTEMS 5, RSB 5.not_released, Newlib 7947581)
Compiler flags   : -O2
Memory location  : Please put data memory location here
                        (e.g. code in flash, data on heap etc)
seedcrc          : 0xe9f5
[0]crclist       : 0xe714
[0]crcmatrix     : 0x1fd7
[0]crcstate      : 0x8e3a
[0]crcfinal      : 0x25b5
Correct operation validated. See README.md for run and reporting rules.
CoreMark 1.0 : 3154.574132 / GCC7.5.0 20191114 (RTEMS 5, RSB 5.not_released, Newlib 7947581) -O2 / Heap
 [/] # dhry 10000000

Dhrystone Benchmark, Version 2.1 (Language: C)

Program compiled without 'register' attribute

Execution starts, 10000000 runs through Dhrystone
Execution ends

Final values of the variables used in the benchmark:

Int_Glob:            5
        should be:   5
Bool_Glob:           1
        should be:   1
Ch_1_Glob:           A
        should be:   A
Ch_2_Glob:           B
        should be:   B
Arr_1_Glob[8]:       7
        should be:   7
Arr_2_Glob[8][7]:    10000010
        should be:   Number_Of_Runs + 10
Ptr_Glob->
  Ptr_Comp:          1080941968
        should be:   (implementation-dependent)
  Discr:             0
        should be:   0
  Enum_Comp:         2
        should be:   2
  Int_Comp:          17
        should be:   17
  Str_Comp:          DHRYSTONE PROGRAM, SOME STRING
        should be:   DHRYSTONE PROGRAM, SOME STRING
Next_Ptr_Glob->
  Ptr_Comp:          1080941968
        should be:   (implementation-dependent), same as above
  Discr:             0
        should be:   0
  Enum_Comp:         1
        should be:   1
  Int_Comp:          18
        should be:   18
  Str_Comp:          DHRYSTONE PROGRAM, SOME STRING
        should be:   DHRYSTONE PROGRAM, SOME STRING
Int_1_Loc:           5
        should be:   5
Int_2_Loc:           13
        should be:   13
Int_3_Loc:           7
        should be:   7
Enum_Loc:            1
        should be:   1
Str_1_Loc:           DHRYSTONE PROGRAM, 1'ST STRING
        should be:   DHRYSTONE PROGRAM, 1'ST STRING
Str_2_Loc:           DHRYSTONE PROGRAM, 2'ND STRING
        should be:   DHRYSTONE PROGRAM, 2'ND STRING

Microseconds for one run through Dhrystone:    0.3 
Dhrystones per Second:                      3527884.4 
DMIPS:                                      2007.90 

 [/] # linpack 100
Rolled Double Precision Linpack Benchmark - PC Version in 'C/C++'

Compiler     rtems arm-rtems5-gcc 7.5.0 20191114
Optimisation -O2

norm resid      resid           machep         x[0]-1          x[n-1]-1
   1.7    7.41628980e-14   2.22044605e-16  -1.49880108e-14  -1.89848137e-14

Times are reported for matrices of order          100
1 pass times for array with leading dimension of  201

      dgefa      dgesl      total     Mflops       unit      ratio
    0.00519    0.00016    0.00535     128.35     0.0156     0.0955

Calculating matgen overhead
       100 times   0.04 seconds
      1000 times   0.37 seconds
      2000 times   0.74 seconds
      4000 times   1.48 seconds
      8000 times   2.96 seconds
     16000 times   5.92 seconds
Overhead for 1 matgen      0.00037 seconds

Calculating matgen/dgefa passes for 5 seconds
       100 times   0.55 seconds
       200 times   1.09 seconds
       400 times   2.18 seconds
       800 times   4.36 seconds
      1600 times   8.72 seconds
Passes used        917 

Times for array with leading dimension of 201

      dgefa      dgesl      total     Mflops       unit      ratio
    0.00508    0.00016    0.00524     131.12     0.0153     0.0935
    0.00508    0.00016    0.00524     131.12     0.0153     0.0935
    0.00508    0.00016    0.00524     131.11     0.0153     0.0935
    0.00508    0.00016    0.00524     131.13     0.0153     0.0935
    0.00508    0.00016    0.00524     131.13     0.0153     0.0935
Average                               131.12

Calculating matgen2 overhead
Overhead for 1 matgen      0.00037 seconds

Times for array with leading dimension of 200

      dgefa      dgesl      total     Mflops       unit      ratio
    0.00508    0.00016    0.00523     131.22     0.0152     0.0934
    0.00508    0.00016    0.00523     131.22     0.0152     0.0934
    0.00508    0.00016    0.00523     131.21     0.0152     0.0935
    0.00508    0.00016    0.00523     131.21     0.0152     0.0935
    0.00508    0.00016    0.00523     131.21     0.0152     0.0935
Average                               131.21

Rolled Double  Precision      131.12 Mflops 



**hifi4 dsp  ,主频600M，系统 freertos**

Dhrystone Benchmark, Version 2.1 (Language: C)

Program compiled without 'register' attribute

Execution starts, 10000000 runs through Dhrystone
Execution ends

Final values of the variables used in the benchmark:

Int_Glob:            5
        should be:   5
Bool_Glob:           1
        should be:   1
Ch_1_Glob:           A
        should be:   A
Ch_2_Glob:           B
        should be:   B
Arr_1_Glob[8]:       7
        should be:   7
Arr_2_Glob[8][7]:    10000010
        should be:   Number_Of_Runs + 10
Ptr_Glob->
  Ptr_Comp:          933400000
        should be:   (implementation-dependent)
  Discr:             0
        should be:   0
  Enum_Comp:         2
        should be:   2
  Int_Comp:          17
        should be:   17
  Str_Comp:          DHRYSTONE PROGRAM, SOME STRING
        should be:   DHRYSTONE PROGRAM, SOME STRING
Next_Ptr_Glob->
  Ptr_Comp:          933400000
        should be:   (implementation-dependent), same as above
  Discr:             0
        should be:   0
  Enum_Comp:         1
        should be:   1
  Int_Comp:          18
        should be:   18
  Str_Comp:          DHRYSTONE PROGRAM, SOME STRING
        should be:   DHRYSTONE PROGRAM, SOME STRING
Int_1_Loc:           5
        should be:   5
Int_2_Loc:           13
        should be:   13
Int_3_Loc:           7
        should be:   7
Enum_Loc:            1
        should be:   1
Str_1_Loc:           DHRYSTONE PROGRAM, 1'ST STRING
        should be:   DHRYSTONE PROGRAM, 1'ST STRING
Str_2_Loc:           DHRYSTONE PROGRAM, 2'ND STRING
        should be:   DHRYSTONE PROGRAM, 2'ND STRING

Microseconds for one run through Dhrystone:    0.8 
Dhrystones per Second:                      1243935.8 
DMIPS:                                      707.99 

Rolled Double Precision Linpack Benchmark - PC Version in 'C/C++'

Compiler     xtensa hifi4 dsp xtensa-elf-gcc 10.2
Optimisation -O2

norm resid      resid           machep         x[0]-1          x[n-1]-1
   1.7    7.41628980e-14   2.22044605e-16  -1.49880108e-14  -1.89848137e-14

Times are reported for matrices of order          100
1 pass times for array with leading dimension of  201

      dgefa      dgesl      total     Mflops       unit      ratio
    0.07600    0.00300    0.07900       8.69     0.2301     1.4107

Calculating matgen overhead
       100 times   0.43 seconds
       200 times   0.85 seconds
       400 times   1.70 seconds
       800 times   3.41 seconds
      1600 times   6.81 seconds
Overhead for 1 matgen      0.00426 seconds

Calculating matgen/dgefa passes for 5 seconds
       100 times   8.05 seconds
Passes used         62 

Times for array with leading dimension of 201

      dgefa      dgesl      total     Mflops       unit      ratio
    0.07621    0.00231    0.07851       8.75     0.2287     1.4021
    0.07621    0.00231    0.07851       8.75     0.2287     1.4021
    0.07621    0.00231    0.07851       8.75     0.2287     1.4021
    0.07621    0.00231    0.07851       8.75     0.2287     1.4021
    0.07621    0.00231    0.07851       8.75     0.2287     1.4021
Average                                 8.75

Calculating matgen2 overhead
Overhead for 1 matgen      0.00426 seconds

Times for array with leading dimension of 200

      dgefa      dgesl      total     Mflops       unit      ratio
    0.07629    0.00231    0.07860       8.74     0.2289     1.4035
    0.07629    0.00231    0.07860       8.74     0.2289     1.4035
    0.07629    0.00231    0.07860       8.74     0.2289     1.4035
    0.07627    0.00232    0.07860       8.74     0.2289     1.4035
    0.07629    0.00231    0.07860       8.74     0.2289     1.4035
Average                                 8.74

Rolled Double  Precision        8.74 Mflops 

2K performance run parameters for coremark.
CoreMark Size    : 666
Total ticks      : 16258
Total time (secs): 16.258000
Iterations/Sec   : 1414.688154
Iterations       : 23000
Compiler version : GCC10.2.0
Compiler flags   : -O2
Memory location  : STACK
seedcrc          : 0xe9f5
[0]crclist       : 0xe714
[0]crcmatrix     : 0x1fd7
[0]crcstate      : 0x8e3a
[0]crcfinal      : 0xd340
Correct operation validated. See README.md for run and reporting rules.
CoreMark 1.0 : 1414.688154 / GCC10.2.0 -O2 / STACK

目前hifi dsp采用的自编译的gcc编译器，采用R528 SDK提供的指令集，对linpack的跑分存疑，谁有candence explore能否跑个分比对下，谢谢

yao0718LV 3

@yao0718

Best posts made by yao0718

Latest posts made by yao0718