谁跑过D1a的core_mark分数吗,我这里有T113的跑分,想来对比下

smiletiger LV 6

2K performance run parameters for coremark.
CoreMark Size    : 666
Total ticks      : 12751
Total time (secs): 12
Iterations/Sec   : 4166
Iterations       : 50000
Compiler version : GCC6.3.1 20170215 (release) [ARM/embedded-6-branch revision 245512]
Compiler flags   :
Memory location  : STACK
seedcrc          : 0xe9f5
[0]crclist       : 0xe714
[0]crcmatrix     : 0x1fd7
[0]crcstate      : 0x8e3a
[0]crcfinal      : 0xa14c
Correct operation validated. See README.md for run and reporting rules.
CoreMark 1.0 : 4166 / GCC6.3.1 20170215 (release) [ARM/embedded-6-branch revision 245512]  / STACK

另外我想问一下,我使用smp和不使用,跑分是一样的,这个smp到底是怎么用的啊,是不是没啥用,在rtt里面建立任务的时候会自动分配cpu还是需要人为去指定
或者能不能一个任务两个核一起来用,
另外在附上memory_perf的分数

MemoryPerf version 1.0.0
Copyright (c) 2022 SummerGift (summergift2019@gmail.com)
Licensed under the MIT License version.

Memory performance testing start...
address: 0x40100000, length: 0x100000, iterations: 200
Data length : 209 MB.
--------------------------------------
8-bit write speed test begin.

[2022-04-18 15:41:26.787]# RECV ASCII>
Spend time : 0.223000 s.
8-bit write speed: 940.426880 M/s.
8-bit read speed test begin.

[2022-04-18 15:41:27.346]# RECV ASCII>
Spend time : 0.559000 s.
8-bit Read speed: 375.161346 M/s.
--------------------------------------
16-bit write speed test begin.

[2022-04-18 15:41:27.472]# RECV ASCII>
Spend time : 0.111000 s.
16-bit write speed: 1889.326050 M/s.
16-bit read speed test begin.

[2022-04-18 15:41:27.802]# RECV ASCII>
Spend time : 0.328000 s.
16-bit Read speed: 639.375610 M/s.
--------------------------------------
32-bit write speed test begin.

[2022-04-18 15:41:27.896]# RECV ASCII>
Spend time : 0.091000 s.
32-bit Write speed: 2304.562744 M/s.
32-bit read speed test begin.

[2022-04-18 15:41:28.067]# RECV ASCII>
Spend time : 0.160000 s.
32-bit Read speed: 1310.719971 M/s.
Memory performance completed.

有没有大神拿其他片子来对比一下,这个a7到底什么水平啊

caozilong LV 5

A7 大概1.6个DMIPS/MHZ,平均下来每个clk执行1.6条指令，和谷歌标称的A7 1.9 DMIPS/Mhz差别不大，中间可能有编译器优化，运行环境不同引入的差别。

aozima LV 7

rtt里面建立任务的时候会自动分配cpu还是需要人为去指定
或者能不能一个任务两个核一起来用,

默认会自动分配吧，也可以绑定。你这跑分，可以改下代码，同时跑2份core_mark看看部分。

smiletiger LV 6

@aozima 怎么同时跑两份啊,我用的是你移植的rtt,smp部分不太懂,启动第二个核心的时候那个时钟tick要不要开启啊,你那边的代码里面注释掉了定时器,

aozima LV 7

@smiletiger 你要测性能的话要开启的，我那个代码是从qemu那边复制过来的，要重新适配T113才行，你可以试试加上心跳。

aozima LV 7

@smiletiger 注意下，那个移植里面，不光第2核的心跳没启动，0核的心跳也没启动呢。

smiletiger LV 6

@aozima 我已经开启了cpu1的心跳,
cpu2的心跳不开启的话cpu2不能参与任务调度吗,但是我用rt_thread_control(tid, RT_THREAD_CTRL_BIND_CPU, (void *)1);这个绑定cpu是可以正常调度的呀

柚木鉉 LV 9

Linux TinaLinux 5.4.61 #265 PREEMPT Mon Apr 18 15:41:13 UTC 2022 riscv64 GNU/Linux

2K performance run parameters for coremark.
CoreMark Size    : 666
Total ticks      : 13456
Total time (secs): 13.456000
Iterations/Sec   : 2972.651605
Iterations       : 40000
Compiler version : GCC8.1.0
Compiler flags   : -O2   -lrt
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 : 2972.651605 / GCC8.1.0 -O2   -lrt / Heap

MemoryPerf 不知道能不能在RV上跑，没测试

aozima LV 7

@smiletiger 没心跳只是与时间相关的不正常，主动让出不受影响。
建议把cpu1的心跳也开起来吧，期待！

smiletiger LV 6

@yuzukitsuru 请问你这个是f133的分数吗或者是d1或者d1s吗,这个core-Mark在linux下和rtthread下面跑分是不是不一样啊,我优化了一下代码下载可以跑到4170了
2K performance run parameters for coremark.
CoreMark Size : 666
Total ticks : 11990
Total time (secs): 11.990000
Iterations/Sec : 4170.141785
Iterations : 50000
Compiler version : GCC6.3.1 20170215 (release) [ARM/embedded-6-branch revision 245512]
Compiler flags :
Memory location : STACK
seedcrc : 0xe9f5
[0]crclist : 0xe714
[0]crcmatrix : 0x1fd7
[0]crcstate : 0x8e3a
[0]crcfinal : 0xa14c
Correct operation validated. See README.md for run and reporting rules.
CoreMark 1.0 : 4170.141785 / GCC6.3.1 20170215 (release) [ARM/embedded-6-branch revision 245512] / STACK
msh />

smiletiger LV 6

@yuzukitsuru 又随手跑了一下 Dhrystones
dhrystone_test

Dhrystone Benchmark, Version 2.1 (Language: C)

Program compiled without 'register' attribute

Execution starts, 32000000 runs through Dhrystone

[2022-04-19 11:07:53.418]# RECV ASCII>
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]: 32000010
should be: Number_Of_Runs + 10
Ptr_Glob->
Ptr_Comp: 1073951700
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: 1073951700
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
[2022-04-19 11:07:53.538]# RECV ASCII>
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
Dhrystones per Second: 4484933
VAX MIPS rating: 2552

msh />

因为一直在纠结用f133还是t113,在两者之间徘徊,所以才想测试一下性能,以便决定以后用哪个

柚木鉉 LV 9

@smiletiger 这个是F133-B的分数，我估计跟平台应该有关

Randolph LV 6

麻雀D1s跑的，优化等级o3

smiletiger LV 6

@randolph 谢谢大家提供的数据,我实际测试下来发现t113s3优势非常大,我来跑rtt+awtk t113s3开启smp模式可以丝滑跑120帧,但是f133只能跑60帧,而且60帧还不稳定,不知道是riscv64的开发环境刚出来没优化好还是其他问题,

smiletiger LV 6

@aozima 大神能讲解一下smp方面的rtt问题吗,我现在只开启了一个定时器,然后rtconfig里面定义了RT_USING_SMP,然后跑gui发现非常丝滑,
我把RT_USING_SMP这个定义关闭了在跑gui发现非常卡了

然后我想着不是还有一个定时器没有开启吗,我就在void secondary_cpu_c_start(void)
{
int timer_irq_number;

timer_irq_number = 92;

rt_hw_vector_init();

rt_hw_spin_lock(&_cpus_lock);

arm_gic_cpu_init(0, platform_get_gic_cpu_base());

// arm_gic_set_cpu(0, timer_irq_number, 0x2); // timer1

// timer1_init();
//rt_hw_interrupt_install(timer_irq_number, rt_hw_timer1_isr, RT_NULL, "tick1");
//rt_hw_interrupt_umask(timer_irq_number);

rt_system_scheduler_start();

}
这里把定时器2也开启了,现在发现之前开启的那个定时器1中断不进去了,进去的都是定时器2的中断,gui的任务夜跑不起来了,也就是说我在second cpu里面在开启一个定时器就不正常了,找了资料说smp需要开启2个定时器,为什么我这里开启2个定时器之后,前面那个定时器中断进不去了呢,然后我把second cpu的定时器关闭,性能上也达到了双核的提升了呀,因为我如果禁止smp,那么性能比较差的明显的,我现在迷糊了,定时器到底要不要开启2个,能否讲解一下这方面的疑惑,非常感谢