【FireBeetle 2 ESP32 C6】Comparison test of several esp32 performances

dcexpert

【FireBeetle 2 ESP32 C6】Comparison test of several esp32 performances [Copy link]

 

In order to understand the performance of esp32c6, a simple performance comparison was made on several development boards. A short test program was written to calculate fractals, pi, etc. The test is not too rigorous and is only used as a reference.

From left to right: rpi PICO, esp32s2, esp32s3, esp32c3, esp32c6. For hardware comparison, please refer to the post [FireBeetle 2 ESP32 C6] esp32 series chip comparison .

The test procedure is as follows, including addition, multiplication (division), fractal, pi and other calculation procedures. The time used for each test is calculated separately. The shorter the time, the better the performance. In order to ensure that the test is carried out in the same environment, the firmware of the development board is upgraded to the latest test version before the test.

from time import ticks_ms, ticks_diff
from machine import freq
from platform import platform
import gc

def run(func, *param):
    gc.collect()
    t1 = ticks_ms()
    if param == None:
        func()
    else:
        func(*param)
    t2 = ticks_ms()
    print('calc time:', ticks_diff(t2, t1), 'ms\n')

def mandelbrot(iter=80):
    def in_set(c):
        z = 0
        for i in range(iter):
            z = z * z + c
            if abs(z) > 4:
                return False
        return True
    
    for v in range(31):
        for u in range(81):
            if in_set((u / 30 - 2) + (v / 15 - 1) * 1j):
                print(' ', end='')
            else:
                print('#', end='')
        print()

def pi(places=100):
    # 3 + 3*(1/24) + 3*(1/24)*(9/80) + 3*(1/24)*(9/80)*(25/168)
    # The numerators 1, 9, 25, ... are given by (2x + 1) ^ 2
    # The denominators 24, 80, 168 are given by (16x^2 -24x + 8)
    extra = 8
    one = 10 ** (places+extra)
    t, c, n, na, d, da = 3*one, 3*one, 1, 0, 0, 24
 
    while t > 1: 
        n, na, d, da = n+na, na+8, d+da, da+32
        t = t * n // d
        c += t
    return c // (10 ** extra)


def add_test(N=100000, A=1.1, B=2.2):
    for i in range(N):
        A + B

def mul_test(N=100000, A=1.1, B=2.2):
    for i in range(N):
        A * B


print('\nSystem:', platform())
print('Frequency:', freq())
print('Memory:', gc.mem_free()+gc.mem_alloc())
print('\nBegin test\n')

print('Calculate 1000000 additions')
run(add_test, 1000000)

print('Calculate 1000000 multiplications')
run(mul_test, 1000000)

print('Calculate 1000000 divisions')
run(mul_test, 1000000, 12345, 1/13)

print('Calculate mandelbrot with iterations 200')
run(mandelbrot, 200)

print('Calculate 1000 digits of pi')
run(pi, 1000)

print('Calculate 10000 digits of pi')
run(pi, 10000)

Test results (time unit is ms)

	esp32c3	esp32c6	esp32s2	es32s3	rpi pico
frequency	160MHz	160MHz	160MHz	160MHz	125MHz
ram	203264	342528	7797248	252416	233024
Addition 1,000,000 times	5798	5642	7693	6403	10392
Multiplication 1,000,000 times	5955	5555	7670	6334	10257
Division 1,000,000 times	6263	5937	7307	5884	10313
mandelbrot (iterations 200)	4526	3862	8154	4122	7331
Calculate pi to 1000 digits	407	420	675	463	562
Calculate pi to 10,000 digits	32997	30260	60229	36467	66017

From the test results, we can see that the performance of esp32c6 is slightly better than that of esp32c3, which is about 5-10% in total. The performance differences of several esp32 chips are not too big, but the esp32cx series with risc-v core still has a slight advantage. It seems that the subsequent H series (low-cost and low-power wireless communication) and P series (high-performance and high-security general-purpose) will also be worth looking forward to.

chejm

The ESP32 performance comparison summarized by the host is very practical and of great reference value for selecting boards. Thank you for sharing.

秦天qintian0303

From esp32c3 to esp32c6, why is the gap so big?

dcexpert

Qintianqintian0303 posted on 2024-4-28 09:14 From esp32c3 to esp32c6, why is the gap so big?

The manufacturer is arbitrary.