Block non-convergence soon?

[mimarob]

hello!

am I rightly informed that the nonce is only 32 bits?

Does this in turn mean that there are "only" 2^32 or about 4 billion (4E9) combinations to try for each block?

If so, then when network hashing capacity reaches 4 000 000 kHash/sec the block would be solved in one second on average?

This contradicts the onlïne hash-rate calculator, which states about 4 hours, at present difficulty.

Sorry if this has been up before, but I'm really puzzled..

[davout]

The nonce overflows very often, but the timestamp changes often too , and even if it didn't the merkle root can

[mimarob]

aha, so in general any given block with the nonce as the only variable is not likely to have a solution at all?

Are there some statistics on how often the non-nonce parts of the current block changes?

[Mike Hearn]

There is an "extraNonce" part of the block. It's where the script would be in the generation transaction, called the coinbase data. When nonce overflows extraNonce is incremented.

[jgarzik]

aha, so in general any given block with the nonce as the only variable is not likely to have a solution at all?

It is always possible that there is no solution for the entire range of nonce (0 - 0xffffffff).

However, the timestamp and extranonce are changed every few seconds, as is merkle tree root.  This helps avoid a situation where we get stuck with no solutions.