By using this site, you agree to our Privacy Policy and our Terms of Use. Close
Pemalite said:

If you had cared to read. I did state the Gflops is not an accurate way to determine the capabilities of two different processors.
You can have a graphics processor that is 2 Teraflops and it can be slower than one that is 1 Teraflop. That is the entire point.

And searching around, I still haven't found reliable professional benchmark numbers for the Wii U CPU.
When it comes to SIMD though, I fully expect the Xbox 360's CPU to demolish the Wii U's CPU, but that isn't always going to be the case.
Point me to information that says otherwise please.
The WiiU also has that big fat piece of edram, which can act like an L4 cache, that kinda' reduces the cache miss hit penalty substantually, along with a shorter pipeline (Aka. Pentium 4 vs Core Duo), Out-of-order execution (Intel Atom Silvermont vs Bonnel)

The WiiU GPU can also do some tasks that would have been performed on the Xbox 360 and Playstation 3 graphics processors, such as frame buffer effects or Morphological Anti-Aliasing, which does even things up a bit in the Wii U's favor.

As for the GPU we *know* it's terascale derived. That means VLIW5. That means Tessellation. That means Radeon 4000/5000/low-edd 6000 that means modern techniques and functionality. It's not GCN 1.0, 1.1 or 1.2 though, which is unfortunate.

And 45nm isn't power hungry, it's a mature process with very low amounts of leakage, granted it's old today...  But it's cheap and mature, perfect for a cheap  SoC which isn't going to be 5~ Billion transisters in size.

If Marcan is been honest, the CPU is practically 3 overcloked Gecko's for all intents and purposes ... 

I'm scratching my head wondering if it's supposed to perform better than the turds in the sub-HD twins ...

What do you mean by "frame-buffer effects" ? Do you mean post-processing ? That can be done with pixel shaders on GPUs with a DX9 feature set. MLAA is another post-process AA technique that never gained much traction. Personally I would've advertised support for custom sample locations like how Nvidia recently added it with 2nd gen Maxwell for MFAA. That feature is a boon for temporal anti aliasing ...

Tessellation performance in Evergreen is pitiable for the most part ...