Blockchain miners calculate the hash of (block data + nonce value). They actually guess (are looking for) nonce in order to get hash staring with a defined number zeroes. (The number of required starting zeros in hash value define the difficulty of finding a nonce that fits - proof of work).
My question is do all miners guess/try nonces from the same pool or range (this way they actually non-efficiently test the same guesses many times) or they somehow split the pool of possible nonces among themselves in order not to try the same nonces multiple times?