What is the requirement for full crawls? I commonly never do full crawls unless
I make managed property changes or some other maintenance activity that requires
it. If these are in a single Web application, then I'd find the root of the
problem and not necessarily carve up into multiple content sources. That being
said, you need to get to the root of the bottleneck, so:

How many documents total out of the 4 TB?
Have you excluded the index directories from system anti-virus scanning? That
can take 50% or more of available resources.
How many processors are in your indexer? How much memory?
Does it perform other application duties, such as Excel? Those might be
impacting performance.
Have you noticed long disk queue lengths on either the Indexer's disks or
content database disks?
What Web server is the entry point for the crawler? Is it dedicated, using the
indexer, or using default DNS along with users?

If you could help us with a little more background, I'm sure we can steer you in
the right direction.