System Faults Didn't Stop THQ From Bringing Metro: Last Light To Wii U

[MegaDave]

Lafiel said:

miz1q2w3e said:

Lafiel said:

miz1q2w3e said:

zarx said:

actually most people would be right in that case http://www.anandtech.com/bench/Product/47?vs=677&i=505.506

I don't have to tell you why that's wrong, but ok.

what kind of relevance have CPU only benchmarks for a game system? gaming performance probably is more important, no? (not to mention, that the i7 in the bench is 2.66GHz, so it's a lot faster than the 1.5GHz i7 Nem originally proclaimed to be faster than a 3GHz i3)

I just wanted to counter his post/ spreading of misinformation. Anyone with knowledge should know that clock frequency is not the only measure of a CPU's performance.

Yes, the lately often quoted architecture does play the biggest role in CPU performance, yet Nem's example was flawed from the beginning as i3 and i7 are _same gen tech_. The i7 has better per clock performance mainly due to larger caches and so on, but the underlying architecture is largely the same.

I expect the WiiU's main chip to have a better per clock performance than the Xenos for example, but I don't know how much that can be if it operates with less transistors as some reports point to. (same process -> same transistor size -> but smaller die size = lower transistor count?) Normally CPU performance does scale pretty linear with transistor count, but a considerably better (or better suited for it's functions) architecture can upset that aswell.

I think this comparison makes the point quite well. http://www.anandtech.com/bench/Product/93?vs=70

3.5ghz dual core,4 thread 'netburst' pentium (like xenon with one core chopped off) roughly equivalent to 1.6ghz dual core, dual thread based on 'core'(like the wiiU cpu with one less core).

Die sizes are 162mm^2 compared to 77mm^2 on the same process(65nm).

The clocks arent a perfect match but it shows whats possible at least.