rp2: (rp2350) adding definitions to pin.csv makes the board go faster/slower

[nspsck]

Port, board and/or hardware

rp2 WeAct rp2350a-v10 and rp2350a-v20

MicroPython version

MicroPython v1.26.0-preview.127.g7a55cb6b3 on 2025-05-18; WeAct Studio RP2350A-V10 with RP2350

Reproduction

Default pin.csv

GP0,GPIO0
GP1,GPIO1
GP2,GPIO2
GP3,GPIO3
GP4,GPIO4
GP5,GPIO5
GP6,GPIO6
GP7,GPIO7
GP8,GPIO8
GP9,GPIO9
GP10,GPIO10
GP11,GPIO11
GP12,GPIO12
GP13,GPIO13
GP14,GPIO14
GP15,GPIO15
GP16,GPIO16
GP17,GPIO17
GP18,GPIO18
GP19,GPIO19
GP20,GPIO20
GP21,GPIO21
GP22,GPIO22
GP25,GPIO25
GP26,GPIO26
GP27,GPIO27
GP28,GPIO28
LED,GPIO25

Added GPIO 23 and 29:

GP0,GPIO0
GP1,GPIO1
GP2,GPIO2
GP3,GPIO3
GP4,GPIO4
GP5,GPIO5
GP6,GPIO6
GP7,GPIO7
GP8,GPIO8
GP9,GPIO9
GP10,GPIO10
GP11,GPIO11
GP12,GPIO12
GP13,GPIO13
GP14,GPIO14
GP15,GPIO15
GP16,GPIO16
GP17,GPIO17
GP18,GPIO18
GP19,GPIO19
GP20,GPIO20
GP21,GPIO21
GP22,GPIO22
GP23,GPIO23
GP24,GPIO24
GP25,GPIO25
GP26,GPIO26
GP27,GPIO27
GP28,GPIO28
GP29,GPIO29
LED,GPIO25

Building using the following command:

make BOARD=BOARD_NAME -j

Expected behaviour

Both pin.psv should not affect performance of the exact code.

def stresstest_vanilla(random):
    for i in range(0,100000):
        random.random() * random.random()
        
        
def stresstest_led(ledEnd, ledStart, random):
    for x in range(0,100000):
        random.random() * random.random()
        #toggle LED to verify runtime
        ledEnd = time.ticks_ms()
        if (ledEnd - ledStart >= 500):
            led.toggle()
            ledStart = time.ticks_ms()
    return ledEnd, ledStart

Observed behaviour

Added GPIO 23 and 29

Default

This shouldn't be a margin of error since across mutiple runs, the difference is consistent. Even bigger gap when overclocked further. Even more interesting, if i add other definitions such as:

GP0,GPIO0
GP1,GPIO1
GP2,GPIO2
GP3,GPIO3
GP4,GPIO4
GP5,GPIO5
GP6,GPIO6
GP7,GPIO7
GP8,GPIO8
GP9,GPIO9
GP10,GPIO10
GP11,GPIO11
GP12,GPIO12
GP13,GPIO13
GP14,GPIO14
GP15,GPIO15
GP16,GPIO16
GP17,GPIO17
GP18,GPIO18
GP19,GPIO19
GP20,GPIO20
GP21,GPIO21
GP22,GPIO22
GP23,GPIO23
GP24,GPIO24
GP25,GPIO25
GP26,GPIO26
GP27,GPIO27
GP28,GPIO28
GP29,GPIO29
LED,GPIO25
LED2,GPIO24
KEY,GPIO23

LED2,GPIO24 and KEY,GPIO23, the performance drops below default by a definitively noticeable margin.

Additional Information

I am well aware losing 60 ms in 100 k cycles is probably not a big deal at all. Tho this is a quite interesting observation to make. Feel free to move this to discussion or mark this as closed if you think this does not need a fix. :D In that case, sorry for bothering! My sincerely apologies!

Code of Conduct

Yes, I agree

[robert-hh]

I would see that as statistical variations and a topic for discussions. The data from pins.csv creates static table, not directly affecting executed code. It may shift items in flash, having eventually impact on caching, but that's it. I do not know whether calling random() itself has a constant execution time. But it creates a float and will therefore allocate RAM from the heap, causing eventually automatic gc, which in turn has a big effect on code execution.