Linux – How to get an ARM CPU clock speed in Linux

armcpu-speedlinux

I have an ARM-based embedded machine based on S3C2416 board. According to the specifications I have available there should be a 533 MHz ARM9 (ARM926EJ-S according to /proc/cpuinfo), however the software running on it "feels" slow, compared to the same software on my Android phone with a 528MHz ARM CPU.

/proc/cpuinfo tells me that BogoMIPS is 266.24. I know that I should not trust BogoMIPS regarding performance ("Bogo" = bogus), however I would like to get a measurement on the actual CPU speed. On x86, I could use the rdtsc instruction to get the time stamp counter, wait a second (sleep(1)), read the counter again to get an approximation on the CPU speed, and according to my experience, this value was close enough to the real CPU speed.

How can I find the actual CPU speed of given ARM processor?

Update

I found this simple Pi calculator, which I compiled both for my Android phone and the ARM board. The results are as follows:

S3C2416

# cat /proc/cpuinfo
Processor   : ARM926EJ-S rev 5 (v5l)
BogoMIPS    : 266.24
Features    : swp half fastmult edsp java 
...
#./pi_arm 10000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
...
8.50 sec. (real time)

Android

# cat /proc/cpuinfo
Processor   : ARMv6-compatible processor rev 2 (v6l)
BogoMIPS    : 527.56
Features    : swp half thumb fastmult edsp java 
# ./pi_android 10000
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
...
5.95 sec. (real time)

So it seems that the ARM926EJ-S is slower than my Android phone, but not twice slower as I would expect by the BogoMIPS figures. I am still unsure about the clock speed of the ARM9 CPU.

Best Answer

AFAICT the clock of a S3C2416 looks just like that of a S3C2443, or similar processors of its family. The Linux source code for it suggests that there are a number of closely-related clocks.

Choice snippet:

    pll = get_mpll(mpllcon, xtal);
    clk_msysclk.clk.rate = pll;

    fclk = pll / get_fdiv(clkdiv0);
    hclk = s3c2443_prediv_getrate(&clk_prediv);
    hclk /= s3c2443_get_hdiv(clkdiv0);
    pclk = hclk / ((clkdiv0 & S3C2443_CLKDIV0_HALF_PCLK) ? 2 : 1);

    s3c24xx_setup_clocks(fclk, hclk, pclk);

    printk("CPU: MPLL %s %ld.%03ld MHz, cpu %ld.%03ld MHz, mem %ld.%03ld MHz, pclk %ld.%03l MHz\n",
           (mpllcon & S3C2443_PLLCON_OFF) ? "off":"on",
           print_mhz(pll), print_mhz(fclk),
           print_mhz(hclk), print_mhz(pclk));

Update from OP

I have been looking for this output from dmesg, but I could not find anything - the dmesg output was littered with debug messages, and the beginning was missing. Clearly the kernel message buffer was too short to hold all messages until I connect by telnet. By putting /bin/dmesg > /tmp/dmesg.log early in the startup process, I was able to get this output, confirming what I wanted to know:

Linux version 2.6.21 (gcc version 4.2.2)
CPU: ARM926EJ-S revision 5 (ARMv5TEJ)
Machine: SMDK2416
...
CPU S3C2416 EVT3
S3C24XX Clocks, (c) 2004 Simtec Electronics
S3C2416: mpll on 534.000 MHz, cpu 534.000 MHz, mem 133.500 MHz, pclk 66.750 MHz

Related Solutions

Understanding Clock Speed and Other CPU Factors

Here are some of the things that can affect overall CPU speed:

The quoted CPU speed is the speed at which the internal clock runs. This is faster than the external clock (the FSB clock) which determines how fast the CPU can access the resources on the motherboard. A fast internal clock means the CPU can process faster, but a slower FSB means it takes longer to access memory and such.

This can be partially offset by the L1 cache (very fast memory inside the CPU itself) which can pre-fetch data from memory while the CPU is doing other things, so a larger L1 cache can yield a faster performance in certain circumstances.

The number of cores is important as well. A 1.6GHz dual core processor has approximately twice the processing speed (in optimal circumstances) of a single core 1.6GHz processor. This never realistically gets achieved in real life, but it is still much faster to have 2 cores than one. It's almost like having 2 processors.

The efficiency of the CPU itself has an effect too. Each instruction that the CPU performs takes a certain number of clock cycles to complete. A more efficient CPU will use less clock cycles to perform complex operations (floating point, for example) which makes the overall processing quicker.

The chipset on the motherboard can have an effect as well when it comes to the CPU accessing resources on the motherboard.

Pretty much everything in the computer has a bearing to a greater or lesser extent on the overall speed of the system.

How to Interpret Output of cat /proc/cpuinfo in Linux

The entries are logical processors; objects used by the kernel internally, not necessarily tied to physical devices. (Note that they all have the same physical id.)

In other words, they represent the four cores of a single Intel i3 CPU.

Also, as noted on this post (which is not exactly a duplicate but closely related), those cores are logical as well – the CPU has two cores physically but supports hyperthreading.

Note that some Intel processors (the i5 included) use hyperthreading, a system where a single processor has (for example) 2 pyhsical cores, but will provide 4 logical cores - allowing the operating system to treat the processor as having more cores than it really does.

If your system has the lscpu tool (part of recent util-linux), it would output a more human-readable summary of the CPUs installed, for example:

Thread(s) per core:    2
Core(s) per socket:    2
Socket(s):             1

Best Answer

Related Solutions

Understanding Clock Speed and Other CPU Factors

How to Interpret Output of cat /proc/cpuinfo in Linux

Related Question