1.2GHz * 4 = 4.8GHz
1.5GHz * 2 = 3.0GHz
quadcore was faster than dualcore at here
This is likely a typical way of thinking about it, but it is just plain wrong.
To answer your question, if the 630 had 1GB of RAM it would probably smoke the L920 in my opinion.
You are mistaken about the 630. In practice the 920 will be faster at almost everything, with the rendering of complicated web pages possibly being the only exception. You are making the typical mistake of overestimating the importance of the number of cores vs the performance of each core, and underestimating the importance of software! That last one is the biggest mistake people make, because it is the most important factor when determining how a CPU will affect perceived performance in everyday use. The fact that the WP app store contains essentially zero apps that make use of four cores, means the third and fourth cores in the Lumia 630 won't be doing much most of the time.
On the other hand, I completely agree with the main point you are making. Comparing different smartphone CPUs is very difficult. If the number of cores and clock frequency is all you've got to go on, then those specs are only useful for comparing CPU's that are very similar. Otherwise core count and clock frequency are just good cover-ups for what is essentially guesswork.
from that one, we can see that the 630 runs OS apps better, while the 1020 runs 3rd party apps better
I don't see that. In fact, I don't see the benchmarks in that article revealing anything at all really.
Above, Catholic Tech Geek mentioned how important it is to differentiate between single- and multi-core performance. He couldn't be more correct. What are these benchmarks showing us however? The article doesn't say, but they appear to be multi-threaded tests (as are most benchmarks). But think... what exactly does that tell us, considering that almost no smartphone apps are heavily multi-threaded? Answer: Nothing....
That is the problem with the benchmarks used in the WPC article you found. Worse yet, that benchmarking app WPC used just provides some unit-less numbers. The benchmarking app doesn't even tell us what those numbers represent. If you don't know what is being measured, then the measurements themselves are meaningless, and that is pretty much the situation we have here. Well, those benchmarks aren't completely meaningless. They do tell us which device is best at running that particular benchmarking app. However, unless you intend to use your phone primarily to run that benchmarking app, that information isn't very helpful (that is one of the main reasons PC graphics cards are predominantly tested using actual games, not GPU benchmarks). Any benchmark must tell us what is being measured (just stating 'system' or 'web' is rather ridiculous, it means nothing). Without that knowledge, we can't say how, or even if, the differences between devices translates into meaningful differences during everyday use. That however, is the only thing that really matters!
I can easily write a benchmark "proving" the 630 is four times faster than the 920, but I can also write a benchmark showing that the 630 is only half as fast as the 920. Neither of those benchmarks means anything, unless I also tell you exactly what is being measured! I hope I'm driving that point home.
Here are examples of more useful benchmarks:
Motorola Moto G (
uses the exact same SoC as in the Lumia 630)
Motorole Droid RAZR HD (
uses the exact same SoC as in the Lumia 920)
When it comes to determining how well a device performs in everyday use, the single-core measurements are the most important by far. As you can see, the Lumia 920's single core performance is 43% higher than the Lumia 630's. You can think of this as meaning that each core in the Lumia 920 is (on average) 43% more powerful than each core in a Lumia 630. Because the circumstances under which more than two cores are simultaneously active are very rare, the Lumia 920 will usually perform better. Roughly 43% better, but here too, it depends on the app. That 43% is just the average of all single threaded tests, but if the CPU workloads generated by those benchmarks aren't similar to the ones generated by the apps you run, then they aren't meaningful. That 43% number should just be considered a rough approximation.
If the app uses a second thread then CPU related performance on the Lumia 920 will be roughly 86% better. Only once an app utilizes more than two threads simultaneously will the advantage start swinging back in the Lumia 630's favour. Like I said, such exceptions are rare, but they do exist. The most important exception is likely IE. Web browsers often can make use of more than two cores, but to what degree depends on the web pages that are being loaded.
I realize this may not be what people want to hear. Some people find this fascinating, but for other's it's nothing but frustrating because they want simple answers. In reality there are no simple answers when it comes to CPU technology. CPUs belong to the most complex machines humans have ever built. It's good to keep things simple, but reducing them to two numbers, core count and clock frequency, takes that a bit far.
RAM is a different issue entirely, but that would be a topic for a different post. I believe the GPU hasn't been mentioned at all so far, but it too is a very important component when it comes to perceived performance on a smartphone.