Validation of a full block currently takes more than 1 second on most commodity hardware, and that's with the 1MB blocksize limit. There's a reason Satoshi choose 10 minute interblock times.

But it's a fair question to ask, and we should be concerned about performance. The number of hash operations required to update the UTXO index is approximately:

2 * (I + O) log (N)
I: # inputs in new block
O: # outputs in new block
N: # outputs in UTXO set

So it scales linearly with the size of the block, and logarithmically with the size of the UTXO set. The constant 2 is because there will be two indices maintained, not the 1 index currently used for block validation, although one should expect the constant factor to be different anyway.

Besides the constant-factor difference, the added cost is that log(N) factor. That's not a super-significant change to scalability, and in my own opinion is definitely a worthwhile cost for secure lightweight clients and quicker sync times.

However when it comes to performance and scalability, the elephant in the room is ECDSA, not hashing. A modern consumer PC can hash about 100 MB/s, but only do about 100 ECDSA signature verifications per second. The number of signature operations is related to the number of inputs, and therefore the size of the block. That's a speed difference of 3 to 5 orders of magnitude, depending on natural variation on the size of outputs and the number of signature operations per input.

EDIT: And thank you for your contribution!