Benchmarking the new APUs

I have seen a few benchmarks on the new Intel Hazwell APUs against the old AMD Trinity APUs. However, I cannot find any that compare apples to apples. Can you guys give us a real world benchmark of the new Hazwell as an APU against a comparable AMD Richland series APU? I think that I read that the new A10-6800K uses DDR3 2133 without over clocking.

Also, I am curious to see how these AMD processors pair up with a discrete graphics card; and if you can use a series higher than what is recommended on the box. For example use a 7000 series card instead of a 6670 card for the A10? Furthermore, will the overall system take a performance hit if a higher end DDR5 card was used to pair with it since the APU is stuck with DDR3?

If someone can post a link of a site that has already done benchmarks like this, please post it.

 

Intel's Haswell isnt an APU. Here is a link to a new review thats just come out that may be of some interest to you. Comparing between the two in terms of computational grunt its a ferrari vs a datsun, Haswell and even an old i3 wipes the floor with them, but with in relation to integrated graphics the apu's really shine.

http://www.anandtech.com/show/7032/amds-richland-vs-intels-haswell-gpu-on-the-desktop-radeon-hd-8670d-hd-4600

Not sure if you understand correctly how ddr3 speeds are determined. 2133 is a speed that the memory modules support and can run at and also need to be supported by the motherboard. Its saved into the rom as an XMP from which you load in the bios, defaults speeds vary. High frequency ddr3 does improve gaming performance with apu's significanty as it did with the last batch of trinitys. AMD though officially is only supporting 1 dimm per channel when running at 2133. 

You can use whatever gpu you like with these chips, and no you system wont take a performance hit with the gpu using ddr5 and the apu utilizing ddr3. One of these with a 7850/70 will be a gaming beast - except for say arma, civ5, shogun 2 etc. (cpu intensive games).

Remember these chips are focused at very low end computing, their computational power is really poor. The whole point of an integration between a gpu and cpu becoming an apu is that with modern software that utilizes such things as openCL the cpu can offload those computaional requests to the imbedded gpu, which has many, many more cores than that of the cpu side. Thus completing those operations many times faster.

Hope this is of some help and if i missed something out sorry.