2.01 std::cout I/O Overhead vs Microcontroller UART & ARM Cortex-M ITM Trace

1. Standard Output Streams (std::cout)

std::cout is an instance of std::ostream that buffers characters before flushing to standard output.

2. std::endl vs '\n'

std::endl writes a newline character AND forces an explicit buffer flush (stream.flush()). In high-frequency logging loops, this ruins I/O performance. Using '\n' avoids unnecessary flushing.

1. The Flash ROM Bloat of <iostream>

Including <iostream> pulls in heavy locale formatting machinery, dynamic stream buffers, and static initializers, instantly consuming 20KB to 50KB of Flash ROM—often exceeding total available ROM on small microcontrollers!

2. Instrumentation Trace Macrocell (ITM / SWO)

ARM Cortex-M3/M4/M7 cores feature a dedicated hardware ITM (Instrumentation Trace Macrocell) peripheral. Writing a byte to ITM->PORT[0] outputs debug characters over the 1-pin Serial Wire Output (SWO) at 2+ MBaud with zero CPU latency and 0 Flash bloat.

Zero-overhead hardware ITM debug logging:

#include <cstdint>

// Hardware ITM Stimulus Port 0 write (0 ROM bloat!)
void itm_putc(char ch) noexcept {
    volatile uint32_t* const ITM_STIM0 = reinterpret_cast<volatile uint32_t*>(0xE0000000);
    volatile uint32_t* const ITM_TER   = reinterpret_cast<volatile uint32_t*>(0xE0000E00);
    
    if (*ITM_TER & 1UL) { // If ITM Port 0 is enabled by debugger
        while (*ITM_STIM0 == 0); // Wait until FIFO ready
        *reinterpret_cast<volatile uint8_t*>(ITM_STIM0) = static_cast<uint8_t>(ch);
    }
}