Alex Atkin UK

Ryudo, on 04 Apr 2014 - 10:13 PM, said:
Funny thing how wrong they are This is a MK Arcade a sequel to the first.



This is the 2013 version. This is more likely what it would have looked like on PS3/360

Just look how much better MK8 looks compared with the first video's. That's what I call optimalisation. Correct me if I'm wrong but the lighting of the scenes is just soooo incredibly well done. It's what sets it apart from the rest. It looks like the scene's have some kind of global illumination applied to them. Some models and circuits have such nice shadows applied to the characters and models themselves. But in some screens it looks like there are multiple kinds of shadows applied. Objects on the actual track seem to use some lighter lighting engine, just the backgrounds and character models have so many shades. I can't get my ...around it, but I'm actually thinking this looks much more sophisticated to 3DWorld...but in that project the video's where nicer by every new release. Good to see they used that experience to optimize MK8 in its final stage.

1 player on screen plus 1 player on gamepad shouldn't take more resources than 4-player split-screen.

Wii u gpu is actually around 3-4x slower than the one in ps4 and xbone. Not only does it only have a small amount of the shader units ps4/bone have, its also a generation behind in shader cores, utilizing vliw5 instead of gcn. (360 is a generation behind wii u shaders for reference)

Bandwidth is pretty important, but vauge and useless unless you know what KIND of bandwidth you are dealing with.

There is:

instruction/operational bandwidth, which is what you need to actually perform add multiply subtract divide etc.

And:

Data bandwidth, which is used to move assets to and fro.

When product marketers lumps them together it (purposfully) obsfucates the meaning of the bandwidth, to simply make a big attractive looking number.

Operational bandwidth is for the data crunching of the shader cores. The more operational bandwidth you have the more your gpu can crunch until you reach the point where yor shader core throughput is maxed out, somewhere beloe peak theoretical performance (thanks to inevitable thins like cache misses, and waiting on dependencies keeping all cores from being utilized). Below this point you will be bottlenecking your performance, approaching it you will find severe diminishing returns, after it the excess bandwidth is wasted money.

Data bandwidth moves assets like textures to the processors that will, well process them. The more bandwidth you have, the more you can move at once to be processed. Below the amount your gpu can process will be bottlenecking performance as the gpu waits for data to arrive. More than what your gpu can process is wasted money.

I think the Nintendo Enthusiast piece is kind of a joke - it’s not even a real rebuttal.

IMO, the thesis of the original article is that Nintendo doesn’t give third parties the type of support they need, not that the “WiiU is harder to develop for." I think the intent there is to kind of poke Nintendo and say “hey, you should be doing this better."

It seems like every time an article like the Eurogamer one appears, sites like these immediately retaliate with one like this, apparently without even having read the source article closely.

the guy dev they are talking to though says he hasn't worked on PS4/X1 so he cannot compare the WiiU to them, so he goes on about how the WiiU is better & easier than the X360 & PS3 

i don't see the point in everyone going about how much better than the WiiU is than the X360 & PS3, the WiiU SHOULD be better than two eight year old consoles.

look guys deny all you like you all rubbed your privates talking shjt about power 7 cpus 
 
whilst WIIBOY101 TOLD ALL OF YOU ""HOW LONG AGO"" EXACTLY WHAT WIIU CPU WAS GIVE CREDIT TO A GUY WHO DESERVES IT 
 
ACTING OBTUSE WHEN THE GUY WAS RIGHT YEARS AGO SAYS MORE ABOUT YOU THAN HIM YOU HAD A BRAIN BOX OF A GUY GIVING YOU THE INFO BUT ALAS NO ONLY MONGS ARE ALLOWED TO EXPRESS THERE TOUGHTS LOL
 
the cpu is what wiiboy said it was YET YOUR STILL SOING IT PRTENDING WIIBOY NEVER TOLD YA WEN CLEARLY HE DID 
 
GROW UP SERIOUSLY GROW THE F**K UP
 
its 3 inproved broadways together @ 45nm with cutting edge IBM EDRAM as catch EXACTLY AS DISCRIBED BY WIIBOY101 ADMIT THIS...
 
each core vs a jaguar utterly owns look at ARM there 4 stage cores only got as far as a cpl of hundreb mhz expresso is hitting almost 1.3 ghz 
 
 
that is by far by a huge amount the fastest out of order 4 stage cpu EVER MADE 


basic math and commonsense prove this FANTASIST WRONG 
 
3MB LEVEL 2 CATCH IS 12 X THE BROADWAY NOT 6 
 
EDRAM LEVEL 2 CATCH PER CORE SO CLOCK SPEED VS BROADWAY X HOW MANY CORES X 2 FOR EDRAM AS IBM CLEARLY STATED AS CATCH ITS 2X BANDWIDTH OF SRAM 
 
THATS 10 X BANDWIDTH OF BROADWAY 
 
12X THE CATCH 10 X THE BROADWAY BANDWIDTH (clearly not 6) 
 
we have tri core vs single core 
 
we have ibm SMP tech 
 
we have custom MCM AND CUSTOM BUSES AND DIRECT LINK TO BOTH EDRAM OF GPU AND MAIN RAM 
 
we have a die shrink to 45nm from 90nm 
 
clearly 10 to 12 x the broadway cpu 
 
BASIC COUNTING ABILITYS LIKE A 5 YEAR OLD CHILD GO ALONG WAY 
 
6 X BROADWAY NO DUDE NOT AT ALL 
 
BY THE WAY A BROADWAY @ 729MHZ UTTERLY DESTROYS A ATOM AT 1.6MHZ 
 
SO IMAGINE EXPRESSOs real world performance and the up to 50x boost in power wen coding at assemly level