Comparing 1 bit or $n$ bits is faster?
1 bit rt? So, with a valid bit, CPU can detect a cache miss faster and fetch the required block from main memory to cache.
For cache hit, we always require a tag comparison. But valid bit can make cache misses slightly faster.
And with multiple cores sharing data, valid bit has more significance - to invalidate cache entries like when a different core updates its copy of cached data.
PS: Have to add more.