I'd missed Tunafish's post - he makes some interesting points there.

I guess I'm expecting something similar to you from the WiiU. Better than parity from the most cost effective hardware they can put together, and in a family living-room friendly box. A bit more than that would be nice though, if it comes.

Maybe so as I've read something very similar to that (nothing about gaming benefits though, but wouldn't that help physics if used?). I also read that the switch to DX10 started to cause poor utilization of their shaders in VLIW5 which in turn leads to the transistor reduction in VLIW4 due to trying to improve utilization by "trimming the fat" so to speak. I've also read that AMD had more plans for VLIW4, but because the fab was still at 40nm they passed to avoid an even bigger die than what it was.

I can definitely agree with your view though about it ending up as a modern 360 (though our views on that might differ). I'm not saying the other direction as fact, just one that I believe is very plausible.

I imagine that this is the sort of news that we have been waiting for. Does it sound like TSMC's new 28nm manufacturing process will be available to Nintendo considering an expected february availability:

http://www.dailytech.com/Nextgenerat...ticle23158.htm

They're gonna need it to control heat and power in that chassis.

I'll advise you to be more careful when trying to put false statements into other people's mouths.

I dismissed (and still do) your speculation that the case size stays the same because Nintendo itself has stated that the current form isn't the final one.
Never have I "speculated" that the case will be "way bigger". In fact, I've made no speculation whatsoever to the case's size.




Why?

To add to what Tottentranz is saying about the case size change, when Iwata first showed the Revolution he said that they planned to make the case even smaller to about the thickness of 3 stacked DVD cases. Just because they showed the concept at E3 doesn't mean it will stay that way. If heat shows itself to be an issue, they will change it before mass production.

They would use NEC, not TSMC. NEC is a part of the 28nm alliance.

http://www.xbitlabs.com/news/other/d...velopment.html

Where should I begin... you know what... I won't even bother and just reiterate what I said before. There's little reasons for us to speculate that the design will change more than just a bit. Not just reason, there's no indication that it will. Iwata would say what he said even if the box stayed exactly the same. This is PR speak, as always. Just changing one screw will qualify is "changed, thus not final". And in most cases, the design shown a year before usually is quite final. PS3 lost HDMI ports and whatnot and got a different controller, but the looks and size stayed the same.

Well that depends how powerful the console is now doesn't it :)

Because they probably want to have launch titles. Unless by 'major hardware overhaul' you really meant something like, clocked slightly higher. Also they are probably going to need to be in production of physical units several months before launch, time is very short for anything not cosmetic.

<edit> as for the speculation about them using 28nm, there's just no reason to believe that Nintendo would gamble on it. They are launching 1 year or more ahead of their competitors, they know they won't be competitive in performance when Sony/MS launch their next gen. There's just no reason to push the envelope to be slightly closer behind their competition at the cost of profitability.

Actually considering the circumstances now surrounding the Wii and with the 3DS being sold at a slight loss, I would be surprised if they did slightly push the tech ahead in order to win the appeasment of more 3rd party developers and publishers.

As to launch titles, I thought most liekly they will have the system out late November?

Because that's working out so well for their bottom line?

And if they want to have titles then, it'd probably be best to have some idea what the hardware is going to be before this point.

Yes, results would have been much worse if they'd been left with millions of 3DS units gathering dust in a warehouse somewhere, not to mention lower game sales.

so you feel their experience with the 3ds tells them the wii:u will be a failure and they should start bailing in advance of the launch?

I don't know the context you are using "this point" in, but the final Wii dev kits didn't go out till as early as six months before launch so that would mean devs may not have Wii U final kits till January at the soonest if that's any indication.

Side note: I get the feeling the launch gap between Nintendo and Sony/MS won't be as big as some believe. This is why I think they should launch no later that September. At the same time using 28nm wouldn't be much of a gamble if they aren't using TSMC to make it. After all I've always thought it was strange IBM said what process the CPU would use, but AMD didn't for the GPU. Nintendo likes their secrecy and they might very well be targeting 28nm, but if it doesn't work out then they haven't put themselves in a compromising position to the public expectations.

What are you talking about?

I just said that lowering the price of 3DS wasn't the reason for their bad financial results, a simple point which should be easily understood..

You've mentioned NEC before but are you sure they even have the appropriate 28nm tech ready (high performance, not low power) or even the supply capacity that Nintendo requires? You're making a number of assumptions here I think.

Well, hardware power is one thing, but I thought part of Nintendo's problem is getting the controller to work as intended. Wasn't there a rumour about them having issues there? *shrug*

All those 01.net rumors seemed to have been based on old info. E3 showed the controller were working just fine. And when they released the specs on the dev kit, apparently devs had already had them for a few months.

LOL. Remember the whole idea started as (and still is) a hypothesis.

http://forum.beyond3d.com/showpost.p...&postcount=263

So really everything I've said is an assumption so to speak, just with as much supporting info as I can find. Of course I don't know if they do or not. But considering they are a part of the alliance that developed 28nm, it would be tough to assume they wouldn't have the capability. The point of the alliance was to accomplish that process in an affordable manner for all who would use it. At the same time we don't know when they would go into full production so who knows if it would be an issue at that time.

Right, well, we don't even know if they are sampling or even developing the right 28nm process that Nintendo requires for a medium to high end part (not mobile/ low power), and with less than a year away, it's a tough assumption to hope for at the moment without any news since 2009 (as far as I know).

I think a 1TFLOPS GPU from AMD using VLIW5 or VLIW4 in mid-2012 can hardly be considered medium-end, much less high-end..

Furthermore, isn't 28nm LP going to be ready before 28nm HP?
I remember the rumours of Southern Islands having relatively low clocks and therefore being able to use LP transistors and gaining time with that.
Is there anything stopping Nintendo from using a 28LP GPU, considering a Q3 2012 launch?

The point is the type of transistor and what sort of GPU you Nintendo folks are asking for. This isn't some low power mobile part you're asking for.

Indeed.
The gracious demands for a decently clocked or sufficiently powered part from fans. Didn't you folks start saying it was no problem to put a 4890 or 5770 or 69xx class part in there? This is not something you achieve on LP without billions of transistors.

You folks? Do you think all Nintendo fans hang around in a massive room somewhere and make joint posts. :lol:

Y'all look alike. So sue me. :3

Not offended or anything, just thought it was hilarious.

The way you brought up an idea you got from some other Nintendo fans as if it was somehow attributable to him as well because he likes Nintendo :lol:

:p All jokes, of course. But anyways, LP is for like... low power parts that don't make sense for Nintendo's console here, maybe for 3DS or something, but not this.

Why not? Frequencies won't be higher than I say around 700 Mhz (729 Mhz if the use 3x multiplier for Wii backwards compability) and low power is always a good thing in such a relatively small case.

Other topic: What are the changes that WiiU will have a 3 core CPU as rumored but also Broadway as a companion core for Wii backwards compability (+they could use it for the OS)

Yes.

which sounds like just another good reason to not be making other changes

Yes, we are Legion. :wink:

The talks are of a 1TFLOP GPU. Legion never claimed the GPU to be clocked at 900MHz like the 4890, or having 2.6B transistors like the HD69xx. ~600MHz at the size of a RV770 was the projection from the start, which isn't that ridiculous even for 40nm in that case.


As a comparison, the HD4850 does 1TFLOP with 800 ALUs / 10 clusters @ 625MHz, through 956M transistors (256bit memory controller included).

625MHz on 28LP in a ~1B transistor chip isn't achievable? Besides, the GPU would even be rather small, wouldn't you say?






I think having a Broadway would be rather redundant. Using an OoOE 3-core PowerPC would probably grant full Wii BC.

Making a "Starlet" comeback wouldn't be that much far-fetched though, as it really made a difference preventing Wii's CPU overhead for the (then complex) I/O. Maybe a souped-up version using a Cortex A or Cortex R core this time, for the added complexity of dealing with the controller video/audio streams.

Aren't some believing the GPU is going to be like 10W and have 100 ALUs? Sounds like the right one to use based on what they are saying. :wink:

But referring back to one of your other posts, 28nm is not something I hope for. A Wii to Wii U jump in power is comparable to what we saw in PS360 from their predecessors. So I'm good whether it happens or not.

But yeah 2009 seems to be when all this started and at the same time info was kept in check about certain components. But I would believe considering when Cayman was released that the designing of VLIW4 would have started around the time Nintendo started planning their next GPU. And when looking at the fact that it would be considered the low-end of the next series of cards they should be getting it at a decent cost.

I'm definitely expecting an OoOe cpu and I get the feeling it will be similar to what MS wanted Xenon to be. And I think the new Starlet is going to be a multi-core ARM this time.

http://www.thinq.co.uk/2010/10/1/mar...ii-2-chip-win/

Doesn't have to be multi-core, the Armada XP is also available as single core:

http://www.marvell.com/embedded-processors/armada-xp/

Could be. But with the needs of that controller I wouldn't be shocked if it does use a multi-core version.

Final devkits are out, and rumor has it they differ quite a bit from earlier kits. Supposedly more powerful than anticipated, too.

More powerful as in, the WiiU will keep up with NextBox and PS4?

No, as in: It exceeds previous target specs in some areas. Supposedly. No greater context was given.

Depends if you're expecting WiiU to launch as a $500 loss leader.

good to hear

Hmm, given lherre had talked about earlier revisions of Wii U kits basically being broken and difficult to get anything running on (supposedly one of the reason it was difficult to gauge it's performance), interesting to hear if true.

I tend to be a little bit skeptical of positive news about Wii U (aka final dev kits), but if this is true real performance/spec leaks cant be too far away you would think.

I still reckon it's a 4770 GPU in there...

And that wouldn't be half bad, but I heavily doubt it. Whatever is in there wasn't even taped out until late 2011 - it's most certainly not a 2009 GPU.

4770 die shrink to reduce power even more and a few custom parts on top...

Nintendo like energy efficient chips, And a die shrink would reduce power, Heat and meen they can get more per waffer.

It's also in line with the comments that Wii U is 50% faster 360 and PS3.

Or a 4000 and a 5000 series hybrid.... Incorporating a 5000 series tessellation unit as lets the face it, The tessellation unit the 4000 series have is useless.

RV740 level tech would need to be toned down quite a bit to be only 50% range. Which i fully expect to happen because Nintendo wants to keep the console so small

Not really, An 8800GTX is roughly twice as fast as the console GPU's and a 4770 is pretty much level with said 8800GTX.

That comment wasn't based on final info/hardware though. Why else would we hear of there being surprise at the final kit? Because they tried to guesstimate too early.

The HD4770 is about 4 times as powerful as the 360 or PS3 GPU's. For instance Xenos is a SM3.0 GPU with the equivalent of 240 stream processors at 500Mhz. While the HD4770 is a SM4.1 GPU with 640 stream processors at 750Mhz.

Not that the WiiU will use that GPU, similar performance maybe, but the GPU itself will be more modern.

RSX is basically a downclocked 7900GT with some rops disabled. A 8800GTX should be a lot more than twice as fast.

I'm still keeping my fingers crossed for an RV770 as the baseline for the custom Wii U GPU.

the 8800gtx was released some 14 months later than 360..and it was a massive jump compared to 7800gtx...it was like night and day.

So someone brought up the GF116 in another thread, and boy does it have a crazy memory setup (1GB on 192-bit ):

http://www.anandtech.com/show/4221/n...short-at-150/2

It got me to thinking about the 768MB, >1GB, and 1.5GB rumours. Naturally, to hit 1.5GB, you need 12Gbit worth of chips. The easiest path whilst maintaining a single density is 6x2Gbit or 3x4Gbit. Given the limited motherboard space, I'm inclined to think no more than 3 chips per side (on top and underneath).

For 768, you just halve the density or # of chips. 3x2Gbit or 6x1Gbit

And I thought I recall seeing some coy remark about >1GB but not confirming 1.5GB... so what if...

>_>
<_<
ಠ_ಠ

Bus width would be another question entirely (96-bit or 192-bit) as well as the RAM type. Both GDDR5 and DDR3 have up to 2Gbit densities available whilst DDR3 alone has 4Gbit and up (from what I can google).

------------------

A hypothetical discussion at Nintendo might have been...

"768MB?"
"Maybe too little... not much of a leap over 360"
"But dev kits traditionally have double the RAM"
"Yes, but we painted ourselves into a corner with the chassis size"
"Multiple Densities!"

http://uk.ign.com/articles/2012/01/2...as-current-gen


IGN says X720 will have the equivalent of a HD6670, which is 20% faster than the Wii U.
If X720 has the equivalent of a Turks (Northern Islands 480sp, 24 TMU, 8 ROP), then the Wii U could have something along the lines of a Redwood (Evergreen 400sp, 20 TMU, 8ROP).


They also claim this HD6670 translates into 6x the graphics capabilities of Xenos. I honestly don't know how a 718M transistor chip at 750MHz can be considered be 6x faster than a ~350M one at 500MHz. Even with the generational gap considered, there's an overhead related to DX11 compliance.
Plus, this 6x performance advantage doesn't make any sense while looking at the fillrate and flops from each GPU.



Of course, this is IGN talking about GPUs, which is about the same as a monkey talking about quantum physics... so there's this chance that although their sources are right, this guy might've heard "sixty-eight hundred" and at the time of righting thought "was it sixty-eight or sixty-six? bleh I don't care, it's probably the same thing".


For example, a Barts-based GPU (Northern Islands, 960/1120 sp, 48/56 TMU, 32 ROP) would fit that performance description quite nicely, with the Wii U getting something along the lines of a Juniper (Evergreen 800 sp, 40 TMUs, 16 ROP).



Wrapping things up, either the "6x" claim is false or the "HD6670 performance" is false, or both are false.

Develop is reporting that Wii U is twice as powerful as Xbox 360

more here: http://www.develop-online.net/news/3...ul-as-Xbox-360

I get the feeling that MS want to keep thermal specs similar to 360. Putting something with power usage similar to Cayman into a set top box might be problematic, especially considering how they had engineering problems with 360 already.

6x is definately a stretch. 4-5x isn't unreasonable if you're looking at pure shader performance though. Its 3.3x faster on paper but efficienct gains between R500 and R9xx would add a lot of performance too. Just look at how much faster the 3870 was compared to the 2900. And the 2900 itself should be more efficient than Xenos.

Sounds about right but only if running pretty slow, say 500Mhz.

It's only a tech demo but it looks very promising!

http://youtu.be/I9bhS_mMjPU?hd=1

that's ancient

But it still looks amazing, The first 10-15 seconds of the video shows perhaps some of the most natural looking lighting I've ever seen.

I mentioned it was ancient because been linked and discussed on this forum already.

And I think our definitions of amazing are quite different.

They must be as I game a monster PC and that demo impressed the hell out of me in places.

Still looks pretty good to this day. Not cutting-edge or anything, but good.

Off screen footage does that.

It does look amazing but it's hard to tell if that's simply the off screen effect.

While it does add something, we do know it looks far better than the official footage we saw, and there's no direct feed available of the build on the show floor

Yeah, for whatever reason, the direct feed release doesn't have as many effects as the full thing shown on the show floor. One of the most noticeable difference is the self shadowing that we don't see in the direct feed.

Don't know if it has been posted before but it seems Nintendo hired Mat Atkinson, ex Crytek Director of Technology, as a consultant software engineer. A new Zelda running on the Cryengine 3 would be sweet :grin:.

www.linkedin.com/in/markatkinson99

Self shadowing is old technology...

Nobody said self shadowing is new technology..

Considering the math and basics of programming self shadows and most now real time capable effects have has been around since the 1970s, I'd say it's old technology :razz: It's quite amazing what people at the University of Utah were doing back then.

-----------

Had a couple thoughts about the rv740/4830 rumour for the early WiiU dev kits. The mismatched rumours can be reconciled if you consider the mobility radeon 4830, which is actually based on rv740, and is thus a 40nm part. I'll note that the 4830 mobility was clocked at 400MHZ at the lowest and was paired up with either DDR3 or GDDR3 on a 128-bit bus.

That said, if they were already having heat issues, the only logical path would be to hope for 28nm, lest they gimp the GPU even further. I'm still not convinced that Nintendo is going to suddenly absorb the high costs of securing 28nm capacity this year (i.e. 1 million GPUs for Q4 2012). But one can always hope for it since that's the natural course to take, and I have no doubt that was the intent, but being beholden to TSMC's schedule (or Global Foundries for that matter) was always going to be precarious, which makes this 2012 launch that much sillier in the grand scheme as far as technology goes.

There's still a problem regarding the size of the chip in that instance. rv740 is a 137mm^2 chip and moving to 28nm would probably make that too small to use a 128-bit bus, let alone a 192-bit one. Of course, that's precluding the event that AMD beefs up the chip even further so that it's just big enough.


--------------

Of course, if what was seen was a "big" chip, then that doesn't jive with the 140mm^2 chip, which is for all intents and purposes, pretty small in its own right. The reasoning put forth thus far is too vague for anything to be concluded ("seeing a big chip, therefore process node is certain"). A big chip would be rv770, the original 256mm^2 die, where the 4830 has two disabled SIMDs and is thus equatable to the rv740, albeit with differing process nodes. Given the vagueness so far of just what that size is, there's nothing definitive about the progression from here. If the original kits stemmed from the 55nm line, then 40nm would be the logical progression.

As it stands, we have nothing definitive about the particular technology used in the kits or if all the kits were the same or how the kits were updated.

The reports of the WiiU being underpowered doesn't really work considering just how much power a 4830 mobility has over the current gen, even if it were clocked as low. There's little to reason to suggest that it's low power, unless the developer who made the comment was only expecting more (for whatever reason).

-------------

If we also consider that the thing is supposed to have eDRAM strapped onto it, we can reason that the I/O bus is going to be fairly wide on top of a 128-bit bus, so that already places a minimum die size that's possible (north of 100mm^2 if we consider a similar I/O footprint as 360's GDDR3+eDRAM buses). So in that sense, a straight die-shrink of an rv740 would never make sense, it'd have to be bigger.

From what I understand it had two SIMDs disabled, so that would eliminate the RV740.

However since you mention the bus impact from the GPU size of a 28nm RV740, what about a 96-bit bus? Hypothetically asking due to your post.

Is there a reason it cant be the RV730? That's always been my first guess.

Because that's not what was in the kit. Though I bet if it were underclocked (as we found out) enough, the performance would be close to the RV730.



EDIT: Using Wikipedia numbers a 4830 is clocked at 575. From there let's use Alstrong's 400Mhz for the underclock.

The RV730XT is 480 GFLOPs with a pixel fillrate of 6GP/s and a texture fillrate of 24GT/s.

The 400Mhz 4830 is 512 GFLOPs with a pixel fillrate of 6.4GP/s and a texture fillrate of 12.8GT/s.

So the biggest difference I see is that the underclocked 4830 has almost half the texture fillrate of the RV730XT.

How do you know what was in the kit, and what is it?

If you dont know, then how can you say it's not an RV730?

Because I was told it was essentially a 4830 (an RV770 with two disabled SIMDs). And as you can see being underclocked to a certain point puts it on par with the RV730. So performance-wise you would be pretty much correct in your assessment of what was in the kit. But that's not the GPU that was in the kit.

It just always seemed odd to me why Nintendo would go with a bigger GPU (IE RV740 or 770) and underclock it or anything like that. Why not go with the small GPU in the first place (I guess thermals, but...)?

I mean, going with an 800m+ transistor GPU (on the order of 3X Xenos/RSX) and not having it perform much better than Xenos/RSX just doesnt seem really possible, or moreso desirable. It's like throwing away money. I find it odd if it really ends up being the case.

I suppose a lot of things are odd though, heck if it's so low performance why not do an SOC, base the GPU on something more modern, on and on...perhaps they have other considerations but if certain things pan out Nintendo's rep as the worst engineers ever is cemented in my mind, this will be a new low.

All ASSUMING Wii U ends up only PS360 class and uses an crippled RV740 GPU coupled with low performance RAM, or something weird and seemingly terribly inefficient like that.

Maybe using an crippled RV740 just ends up cheaper than any other low performance option or something, because yields are good and it's old, despite the large amount of silicon. But if you're using something with up to 800 SP's, you ought to dominate PS360. Shoot I'm using an HD4890 (800 SP's) and it's still quite the capable PC card, runs BF3 1080P mostly Ultra at ~30 FPS (campaign anyway, hear MP is more demanding). What an 800 SP part could do in a closed box would be quite amazing.

Basically this sums up my feelings perfectly:

The dev even claimed they would have to cut back PS360 games to run on Wii U. So it's definitely not your latter concern (expecting more).

I would think, if we throw out BG's info, it'd be more like an RV730, clocked low. Then you start to get into PS360 performance.

The other option is maybe the GPU isn't bad, but the slow RAM just cripples it?

I think there's plenty of evidence this thing is pretty weak, and not much beside GAF fanboy thread hoping and innuendo, that it's strong.

Weak evidence: No uprising of devs saying it's powerful, really (if you correct me, please provide a link). Killer Freaks at E3, looks terrible, the other Wii U E3 demos, definitely not impressive, on PS360 level, 720P no AA.

Considering what Wii ended up being, I wouldn't be surprised to see the final box have a GPU with 160 SPs. I do expect Nintendo to maintain their usual secretive nature regarding the actual specs.

Yeah my other thought is something custom based off of R700 series, if it's PS360 level only.

All this constrained of course by the 01.net rumors that it's R700 based.

Just out of curiosity, were you told simply 4830 or that it was rv770 with two disabled SIMDs? The point I'm trying to make is that there's a major difference between a 4830 and a mobility 4830 as far as manufacturing goes. The latter is based on rv740, which although identical in hardware spec to the rv770 with two disabled SIMDs, is built on 40nm and is a 137mm^2 chip. The desktop part is basically a binned version of the 55nm flagship chip, which is 256mm^2.
A certain die size would be needed for particular memory bus sizes, and we have to make room for the eDRAM I/O, but that's probably not the main concern for the early devkit. Radeon 4670 is probably fine, but that's 55nm. On 40nm, the 5670 was shrunk to ~104mm^2. It's getting tight for a 128-bit bus.

So that does indeed leave TDP cooling issues. Having more of the physical hardware there may just give them some breathing room for playing with the clocks whilst attempting to simulate what they expect for the size of the final customized chip. Who knows? We don't know what they want specifically other than catching up technologically (unified shader part).

Banking on a future die reduction for a new process is extremely risky, especially given that they have no control whatsoever on when and how much the process is going to cost them. Microsoft may have done that with the extremely accelerated schedule for 2005, but Nintendo doesn't strike me as having similar thought processes or even being so desperate.

Well, it really isn't hard to come up with something faster than Xenos, bandwidth notwithstanding (in a PC environment). A Radeon 4650 or 5550 ought to beat it just by virtue of having more shading power, everything else being the same on paper (also clocked slightly higher than Xenos), and of course there's support for more recent developments in the Direct3D specs. But, these two would be in the same league as current gen.

As for rv740, even if it were <500MHz, it's still got a huge advantage in ALUs (>2.5x raw hardware). ROP and TMU-wise, it has double the units, so in the end it's close to twice the paper specs in those two regards. Overall, that's probably the closest you'll get to taking an off-the-shelf PC chip and ending up with "2x" 360.


As small as a tricore PPC may be on 45nm, it'd be a not so insignificant piece of real-estate for an SoC for such a small console, if you know what I mean. Moving to 32nm (SOI or bulk) would also sort of limit manufacturing options. TSMC doesn't have 32nm SOI for instance. Global Foundries seems to be having some sort of issue, but I'm not entirely clear on the situation. I'll have to check when I have the time, but my impression wasn't so favourable in light of how AMD is progressing with Fusion.

I already discuss it above, but rv740 is a 640 ALU part with 4RBEs (16 ROPs) and can filter 32 texels per clock. And we can see how easily that can beat Xenos on paper, even with lower clocks.


Well, we do have to consider that perhaps the devkit in question doesn't have eDRAM. The bloke doesn't even mention that part of the spec. If that's the case, then the "slow ass DDR3" is responsible for all the framebuffer stuff too - even if it's DDR3-1800, a 128-bit bus would only allow for 28.8GB/s for the entire system (note that he doesn't imply a NUMA configuration either).

It'd be easy to see why one might scale back PS360 games.

-----------

Anyways, it's more fuel for fires. Just waiting for something actually concrete. :/

Oh so you have no problem considering throwing out my info despite it actually coming from someone in the know, and at the same time you have no problem latching onto other things especially from what sounds like someone not even verified. Yeah, ok. You're more biased than you think I am. Just look at this post. Nintendo has the worst engineering rep? That's one of the most asinine things I've ever seen typed by a person. I think it's clear who the biased one is just from that comment alone.

Anyway it was most likely clocked that way for stability. Lherre said himself that every time they tried to push the kit it would freeze and that the GPU was the problem. After all that is a 55nm part in a small case.

The Killer Freaks demo was the PS360 build. If you go back and watch the demo the game still has the Xbox 360 context button.