I did consider this direction when I was designing the device, but there are a few disadvantages of going this way. The 5V rail can need a lot more than 2A quite often. The CPC is rated at 1.7A for the bare machine. Actually in reality it only pulls about 1.2A for a bare machine sitting idle, however, as you know, people now strap on extra RAM, ROM, RS232, USB Mice, WiFi, LAN, Spy Satellite Control Systems (maybe that's just me

) and other great stuff. 2A isn't enough for a modern CPC setup. A robust DC/DC converter that can supply 3A is a big chunky piece of hardware at least three or four times the size (and cost) of my device.
It's easier, smaller and cheaper to create a 12V 400mA (4.8W) supply from 5V than trying to create a 5V 3A (15W) supply from 12V.
Now let's take a look at the Chinese device, and a problem that people often get caught out with. The Chinese advert states the MAX figures for all values and these are the values for the IC, not for their circuit. You can not combine these: Can it supply 2A - Yes probably, Can it supply 5V - Most likely. Can it continuously supply 5V at 2A - Definitely not. You only have to look at the size of the components they've used to know that something's not right. 5V at 2A is 10W of power. Does that circuit look like it can handle 10W of power continuously? According to the datasheet, the chip can dissipate an absolute max of 1.39W. They claim that at 5V 1A it runs at 92% efficiency, so let's assume it runs at 88% efficiency at 2A (being very generous here as that's the value for 3.3V, no value for 5V in the datasheet and this assumes an ideal circuit). This means that at 5V 2A running at 88% efficiency, it would need to dissipate >1.2W, that's going to be one hot little chip with a relatively short lifespan.
External Components:
According to the datasheet, the minimum inductor current value can be calculated using the formula:
Ilp = Iload + (Vout / 2 x fs x L) x (1 - Vout / Vin) .... Let's grab the calculator.... = 2.428A
(I've chosen a 10µH Inductor, as it's the one used in the datasheet and most likely what they used)
Now check the internet for what a 2.5A 10µH Inductor looks like. It's 5 times the size of what they've used. They also chose a non-shielded inductor, so you have a very hot 340kHz transmitter inside your CPC. Now we've got and underdimensioned inductor to deal with too. This will completely invalidate their efficiency claim and the circuit will already have reached its limit somewhere slightly above 1A.
You'll also notice that they didn't even bother including the required electrolytic capacitor, just some cheap ceramic caps, so the ripple on the 12V side could and will be massive (and they conveniently don't state values for this). They also seem to have left out the bootstrap diode. So really they've just slapped together the absolute minimum required so that a voltage will come out at the other end.
There is a very good reason why this device costs $1.84... It's because that's all it's worth. I know the Chinese have drastically brought down the cost of electronics, but expecting to get a reliable 5V 2A supply from a tiny slapped together $2 PCB just isn't realistic.
Buy some for yourself, connect them up to a load and push them up to the claimed limits and do some measurements. Let me know how it goes and be careful not to burn your fingers.
Bryce.