The trouble is that while amd and intel both make cpu's they have very different approaches. You can't directly compare core count or clock speed between them because it will never make sense in terms of the actual performance. As others have mentioned intel currently has the upper hand on ipc (instructions per clock). Cpu speed is measured in clock speed, how many times the processing clocks cycle in a given time frame. Intel's architecture allows it to process more data per clock cycle. So now it's like asking which gets the job done better, carrying a gallon of water from point A to point B in 3min or carrying 2 gallons of water from point A to point B in 5min?
Benchmarks help because it removes the confusion. For those who say but amd has more cores or amd has higher clock speeds, how is this possible? Rather than try to dissect the workings under the hood unless it's a hobby or interesting for the sake of it, it's much easier to look at benchmarks for applications someone may actually use. Video editing or gaming or whatever the case may be. Then it doesn't matter how many cores, clock speeds etc, someone can say ok amd xyz vs intel xyz, these are the price points and here is how well each did in programs I'll use.
JimmyEatWord got it a bit backwards. Intel consistently tops the gaming performance charts. Intel cpu's in similar price ranges tend to perform equal or slightly better, to say amd is cheaper no, it's really not. An 8 core amd is cheaper than an 8 core intel but that's because of amd's horrid performance. They have no choice, if their 8 cores matched intel's, they could sell for the same price. This goes back to not paying attention to core count or clock speed but what performance the various cpu's actually bring to the table. That's what they're priced at. As for intel being good at zipping/unzipping, who cares etc, that would be amd. The one benchmark where amd exceeds intel is in zipping/unzipping.
As Barty1884 pointed out, zen should be here soon. It's a chance for amd to play catch-up, they've understood their weaker ipc is an issue. They've realized that the module design consisting of 2 processing cores with a lot of shared front end resources is a poor design that doesn't work in most real world scenarios. Compared to intel's architecture where each processing core has its own resources. That's what zen has done, they've gone back to a linear approach where each core gets its own resources. They've cut back clock speeds and focused instead on ipc. It's also a chance for them to catch up in terms of using ddr4, usb-c, m.2 and a variety of other tech their outdated am3+ platform didn't have.
If looking at gaming performance, most games heavily rely on 1 or 2 threads. Even if they're multithreaded and an 8c/8t or 4c/8t or 8c/16t cpu shows a little activity across all cores/threads, often times most of the activity is on just a couple cores. That means much of the 8 threads the amd fx 8xxx offers do very little work and much of the heavy lifting falls on a couple of weaker ipc cores. The i3 on the other hand has 2 high ipc cores to work with and competes quite well. In some cases there will be a bit of stuttering on an i3 because it lacks 4 true cores. It depends on the game since not all games are the same.
The i3 6300 represents rather poor value at around $150 since the i3 6100 is around $35-40 less and only 100mhz slower. It doesn't mean all intel cpu's are overpriced or poor value, it just means there are better options for the money. Just like the fx 9590 is extremely poor value at $200 when the fx 8350 is $145. The 9590 will require decent water cooling like an h100i, the 8350 won't. The 9590 will require a more expensive motherboard with the power delivery, vrm and cooling for the vrm in order to handle it, it takes a stronger power supply etc. After investing all that, performance gains over an 8350 or minimal and you could have put together an i7 system (for which amd has no equivalent) and still saved money.
Long story short zen is amd's attempt to correct their mistakes and get back up to speed. Intel has done this in the past as well when their p4 prescott chips used a 31 stage pipeline that caused a lot of headaches. P4 northwoods and willamettes only used 20 stages, if the cpu had to reprocess data shorter pipelines made for a faster recovery. Both have made poor decisions, intel has more resources and is capable of making corrections sooner.