>> Если использовать uint8_t как bool, то производительность упадёт. Лучше использовать uint32_t
>> - производительность выше.
> Для intel 32bit - да. А других архитектурах может быть быстрее.

Из относительно современных нашёл такое только для Atom (не путать с Silvermont и более новыми - 45 nm Intel Atom processors introduced Intel Atom microarchitecture. The same microarchitecture also used in 32 nm Intel Atom processor)

D.3.3.5
Parameter Passing
Due to the limited situations of load-to-store forwarding support in Intel Atom microarchitecture, param-
eter passing via the stack places restrictions on optimal usage by the callee function. For example, “bool”
and “char” data usually are pushed onto the stack as 32-bit data, a callee function that reads “bool” or
“char” data off the stack will face store-forwarding delay and causing the memory operation to be re-
issued.
Compiler should recognize this limitation and generate prolog for callee function to read 32-bit data
instead of smaller sizes.

Assembly/Compiler Coding Rule 18. (MH impact, M generality) For Intel Atom processors, “bool”
and “char” value should be passed onto and read off the stack as 32-bit data.