I don't agree with the term "theoretical" here and its always not about money. Power is a precious thing and its not about saving ones electricity bills.
Its about thinking for the entire planet than thinking about yourselves. Its part of a very big discussion that isn't ideal for the situation here.
Peak power draw happens when cpu is at 100% load. Most multithreaded apps like handbrake stress all the cpu cores to 100% and i've seen it coz i use it. I guess people will buy 8350 taking its performance into account in those scenarios. Even if we forget power consumption for a moment, what about outright performance?
From what i see, overclocked performance isn't that good too. Let's say even if it reaches, 4.6ghz with a good cooler, how much performance increment we can see?
*i.imgur.com/WkO0V.png
The chart speaks everything. But if people are willing to save 4k, then can go for the cheaper platform. But what about those who are seeking outright performance?
Power consumption will be an icing on the cake for them.
For intensive visualization again overclocking or no overclocking,fx8350 is a better choice simply because of more cores and IOMMU support.
In case of 3D rendering it is quite close between fx8350 and i5 3570k.
In final rendering of models in applications like 3DSMAX,Maya,Revit fx8350 generally fx8350 should be better because these are heavily threaded but it is not the case always.
With overclocking i5 3570k would be quite fast.Again in ray-tracing fx8350 has the advantage.
With 5k saved one can also get a better graphic card which in some cases can translate into moving from hd7870 to hd7950.In such case not only there would be huge improvement in gaming but also a good level of improvement would be seen in compute like in case of OpenCL based 3D rendering.
In situations like this rendering through hd7950 alone would be faster than the combinations of fx8350/i5 3570k+hd7950.
Viewport renderings would be lot faster with a GPU having better compute.
Another thing to consider is,though in most of the 3D modelling applications final render and test render would use full cores upto maximum utilisation there are lot of other functions which are lightly-threaded like previewing the model in viewport and considerable amount of time is spent there are well.Here the single-threaded performance comes into play.
Now applications like Euler3d which are used for CFD analysis the 8 core fx is considerably slower because this application is memory bandwidth limited which is significantly better in case of intel 2nd gen and 3rd gen processors.Now if we move to an application like Ansys which is also used for numerical analysis and fluid dynamics simulation the fx8350 should be lot faster.
An fx8150 is around 10% or more faster than i7 2600k in ansys.Applications like Abaqus which are used for elemental analysis gains a lot from GPU acceleration which means a combination with better GPU would seal the case.
As far as power consumption is concerned at stock the difference would be around 100Watts at load.But that won't matter much.Ivy is lot more power efficient but still it won't be a factor for selection in many cases.
So while selecting the platform the things that would matter are one's requirements,how much he/she can spend,overclocking,applications used etc.
Can you please give some links to support your claims? I agree with the virtualization part though and i5 k processors lacking iommu.
Talking about the fluidic dynamic simulations and benches involving complex calculations, i don't see amd being any match with intel.
Several linux benchmarks like dolyfyn and lammps ( fluid and molecular dynamics) perform brilliantly on intel architecture while poorly in amd. Why is that? Could you give some similar
windows benchmark to confirm what you said? The key here is MPI ( message passing interface) which is a code in fortran or c which is written to assist parallel computing. Amd made 8 cores but failed to harness them in a software point of view. You said rightly about encryption but what about performance at similar clocks?
And can you also give links to support that rendering claim because i believe intel still has the edge here and the gap widens when overclocked.
Its not about utilizing more cores but to see how capable those cores really are.
Anyways, i need to see the performance of 8350 over 3570k in 3d rendering (maya in particular and not synthetic like 3dsmax), fluid dynamics and all.
Btw check the following:
*www.tomshardware.com/reviews/fx-8350-vishera-review,3328-8.html
*www.tomshardware.com/reviews/fx-8350-vishera-review,3328-10.html
*www.tomshardware.com/reviews/fx-8350-vishera-review,3328-12.html
Now the clock differences at stock are big between 3570k and 8350. What will happen when both are clocked similarly lets say at 4.5ghz?
I doubt the fx can hold up and tomshardware meant the same in their article.
Performance gains of i5 in overclocking are significant than the fx overall and i guess even in those multithreaded scenarios. IPC again comes to play here and its not about scheduling.
.neither they know about bulldozer's failure nor about bulldozer.
