Number of bytes stored in checksum

This kind of check code depends on the usage/purpose/acceptable amount of calculation.Common check codes are

·Parity
·Checksum
- Easy to calculate and fast
- Low transmission error detection capability
- Not useful for tampering detection
(Limited to cases where the highest priority is given to the 8-bit microcomputer, etc., using the check code.)

·CRC (Circular redundancy check)
- It's a little hard to calculate
-
High transmission error detection capability - It is impossible to detect tampering
(The balance between performance and calculation volume is well balanced and practical.) It is also used in rar zip etc.)

·Unidirectional hash function (SHA256, etc.)
- Computational and time consuming
- Transmission error detection is high (too expensive to use for transmission error detection)
- Ability to detect tampering
(It's hard to use a microcomputer with one chip due to the large amount of calculation, but the ability to check is the best.)

If you want me to select a check code of 4 bytes (32 bits) for the original 2MiB data,
The CRC32 is well balanced, so if you're an oiler, you'll adopt it.

Differentiating CRC32 from MD5

If the value 0-255 is up to 2048K, it is not possible to underflow the signed 32-bit.You can prove it on your desk.

In overflow/underflow cases, it is common to compare only the low n bits of the checksum.