10 out of 60 for the 2012 apps is 1/6 or 17%. not sure if that qualifies as a bunch.
the primary theoretical advantage is that if the data being processed can be kept on the apu die vs sent out to pcie bus or to memory, latency and power use can be reduced.
Here is the catch. If you look at the schemantics and busses. There aint any difference. They still sit on the same bus as before. Its just in 1 package instead of 2.
They simply offer nothing in technical terms besides lower manufactoring costs.
Intel is the one with the most integrated IGP since SB. Since it got access to the L3. But again....both are lightyears away from what people think it is when they talk about it as a unified and fully integrated structure.
Here is a good example. While marketing can always call it something else. Just like AMDs Alink Express and Intels DMI thats just renamed PCIe. Its still the same thing we always had.
The green arrows are...well...illusional at best since it actually cross the crossbar (Thats also why the CPU cores are green). The Trinity GPU sits on a PCIe or HT link(Most likely PCIe). Just like before with IGP(HT link) or discrete cards(PCIe).
If you go into windows device manager for Intels HD graphics. It even says the location is on PCI 0. I am sure it says the same for Llano/Trinity.
People too quickly fall for marketing