2x4GB isn't more expensive than 1x8GB.
Also, A6-5200 will operate at 100% performance with a single 8GB 1600MHz memory stick. Both Celeron 1007U and BayTrail ATOM will need dual stick memory in order to have 100% performance, raising the total cost of the platform.
But you wont use CB in ST configuration , isnt it , so the ST
number is totaly irrelevant and should not even been used
for caracterisation of the CPUs perfs , at least for this application ,
so the final result is that Kabini is better in CB.
Cinebench is heavily Intel Optimized, if the single thread in that benchmark is so close between the two, then it will be higher for the Kabini in the rest of the apps. But even if single thread performance of A6-5200 is within 10% of the Celeron 1007U the MT performance is in another dimension.
Clearly for low power (less than 20W), A6-5200 is the clear winner both in CPU and iGPU performance. A lot of people will be able to do light work with those SoCs at a very low power usage. Intel doesnt have any competitive solution at that price/performance/power as of now.
Also, A6-5200 will operate at 100% performance with a single 8GB 1600MHz memory stick. Both Celeron 1007U and BayTrail ATOM will need dual stick memory in order to have 100% performance, raising the total cost of the platform.
Kabini is already so memory starved it hurts its performance and impacts the platform in a very negative way. Specially when IGP is used it turns into a slugfest.
I dont get why you try to oversell it so much on one of its most weak points.
I was not aware cinebench was so intel-leaning. I was merely giving one of the few benchmarks I ran when I had an A6-5200 in my possession. As for wattage, I thought A6-5200 was 25W rated?
I do wish they had gone with dual channel memory on Jaguar- if ARM processors selling at 20 dollars have it seems like it shouldn't be prohibitively exppensive.
With all that said I still think kabini is an amazing value- That Kabini laptop damn near matched my Alienware m11x which was an ~$1000 laptop 3-4 years ago. The age of cheapass okay gaming are upon us.
Core i3 3217U HD4000 with single memory.
http://techreport.com/review/24856/amd-a4-5000-kabini-apu-reviewed/7
And that is only A4-5000 with 128GCN cores at 500MHz, A6-5200 has 128GCN cores at 600MHz.
Well if HD4000 performs that horrible, just imagine how pathetic HD2500 on Celeron 1007U will look like with single memory.
Core i3 3217U HD4000 with single memory.
http://techreport.com/review/24856/amd-a4-5000-kabini-apu-reviewed/7
And that is only A4-5000 with 128GCN cores at 500MHz, A6-5200 has 128GCN cores at 600MHz.
Well if HD4000 performs that horrible, just imagine how pathetic HD2500 on Celeron 1007U will look like with single memory.
So you're basically showing that Kabini is still far too slow for any 1366x768 gaming (lowest laptop res) with modern titles. Let me help you posting a few more benchies:
1024x600 @ Low & no AA. Damn, that would hurt my eyes in a desktop monitor
Didn't know office work use FP intensive workloads. Well, things you learn on AT forums
...didn't say that.
Core i3 3217U HD4000 with single memory.
According to notebook check, it is 25 watts
And ofcource they are having a high price. Who wouldnt. Nobody in this world price their products by production cost. By "gorging",- didnt you mean profit maximising? - Is Intel "gorging" on 80% of their portfolio - i havnt heard you say that?
According to the review that was made with a 850W PSU
wich is hardly a good set up but still we can see that
the 25W Kabini consume the same as a 18W E-350 when
running 3DMark and 5W less when iddle.
http://adrenaline.uol.com.br/biblioteca/analise/784/amd-a6-5200-kabini.html?pg=02
The old zacate plateform consume 31-34W when fully running
and 11W when iddle , this say that Kabini is grossly binned
and that whatever is not on spec at 1.5ghz will be deemed
2ghz with a very conservative TDP to fit them all...
power usage from mitx MB to MB can vary a lot, even using the same CPU... that was the case with the E-350 boards.
There will be far less variation with kabini since it s a full SoC....
What I'm saying is that Kabini seems much more expensive than Brazos, so either TSMC 28nm isn't as cheap as 40nm was when Brazos arrived, demand is really high for the chip or they want a smaller but more profitable market share for the chip. In any case, something is different. This is not Brazos 2.
my point was, comparing a Kabini board with a E-350 board can be misleading if you have one of the extremes...
and E-350 was also always combined with the same "southbridge",
I'm not saying it's wrong. AMD is on the market to make money and they have a duty to their shareholders to charge the highest price they can to maximize their returns. What I'm saying is that Kabini seems much more expensive than Brazos, so either TSMC 28nm isn't as cheap as 40nm was when Brazos arrived, demand is really high for the chip or they want a smaller but more profitable market share for the chip. In any case, something is different. This is not Brazos 2.
There is no reason they will not flood the maket wit this as they did with zakate.
So you're basically showing that Kabini is still far too slow for any 1366x768 gaming (lowest laptop res) with modern titles. Let me help you posting a few more benchies.
And we already seen how it goes downhill outside the CPU weak prescripted benchmarks.
http://forums.anandtech.com/showpost.php?p=35371206&postcount=7
40% difference in prescripted benchmark. 400% difference ingame between 800Mhz and 4900Mhz.