Did you read the article? That reviewer tested those CPUs, he didn't make stuff up as he went.
Yes, but its a really sh***y review. Lets see metro 2033 at 1440p 33 fps average (nice do you think many gamers are really going to play a shooter at 33 fps average? No they are going to lower settings and aim for at least 45+. Plus metro 2033 is very fps variable). So the metro 2033 benchmark (which is very gpu limited too which makes 0 sense for a review aimed at gaming cpus) according to the other crossfire settings is ~50 fps average which probably isn't sufficient for someone who has the money to game at 1440p with a 7970 (seriously spending $100 more on their $1500 setup will gain them significantly better fps).
Dirt 3 is pretty much playable at 60 fps on anything (again poor choice since it doesn't show significant cpu scaling).
Civ 5 is easily playable but with more units....???
Sleeping dogs is again easily playable and again the one 7970 setting isn't really playable. Very gpu limited game.
Testing with games such as BF3 (MP), SC2, WOW, Gw2, Crysis 3, etc would paint a whole different picture. In a games such as crysis 3 the a8 will easily only get about half the fps of an i5 (and probably be unplayable on ultra 1080p settings).
Poor review overall, only a few games tested (4 is not nearly enough to make a conclusion) and the games tend to be very gpu limited which is pointless for a gaming cpu review. Sure the review may have showed certain numbers and proven that an a8 gets playable fps but you can prove anything if you pick only a certain few tests.
And minimum fps.... the review is of subpar quality for this site.
Why silly? If there's evidence just accept it. I was surprised too but would never label it stupid.
The evidence is poor.
Why is it stupid? If you have a budget that doesn't allow more CPU this is precisely the trade-off one should make. The whole article demonstrates that you don't NEED a killer CPU to get good performance in most games... but video cards are often worth every penny in terms of usable performance.
An A-series / 7970 absolutely is the right decision compared to an i5 / 7850 or 7870. It absolutely will give you better overall gaming performance.
Is an i5 / 7970 better? Of course it is, but it also costs more and gives very small benefit in MOST (but not all) games. Is an i3/7970 better? Most of the benchmarks show that the quad A-series performs better than an i3... so the correct recommendation absolutely is the A-series.
This is what I don't get about people hanging out in the CPU forum. They seem incapable of really understanding that real people make trade-offs. Not everyone is going to drop all the cash necessary to be on the bleeding edge. If you're willing to take just a small step back from ultimate performance, you can actually do fairly well on a more moderate budget.
A comparison between A8 and 7970 and i5 and 7850/7870 is really not that great. The price difference between a8 and i5 is not nearly the price difference between 7850/7870 and 7970. As a percent cost for the entire system the i5 is better as well (if it costs $100 more but raises the fps on a $1000 build by 30% then you are getting more bang for your buck).
Do you need an unlocked i5? No. Do you need a higher quality mobo? No. Get a lower model i5 and a cheaper mobo and the price difference drops significantly (given how poorly haswell overclocks you arn't really missing anything. Ivy can do 4 bins above max turbo and that gets you close enough to k series). That alone will perform much better than an a8.
This review doesn't show minimum fps as well or any stuttering/ frame variance as well.
And in my opinion given the choice between a slightly stronger cpu and weaker gpu and a slightly stronger gpu and weaker cpu I'll take the cpu every time. I can always lower my settings to play the game but if I can't get sufficient fps then my experience is ruined.