TEV a keeper or a loser?

[Shifty Geezer]

The trend in Japan is smaller & quieter= is better.
So the Wii, and Gamecube were designed to please that market.
It just so happens that its also popular in other markets.

2 points. 1) slightly larger and more power consumptive isn't going to make a difference. Or do Japanese consumers look up the dimensions and power-draw stats and base purchasing decisions on those? 2) It wouldn't have to be any larger and louder. Again, we're not talking about putting a monster GPU into Wii, but a small, efficient, yet modern design that could achieve better results than Hollywood under similar conditions.

 

[MDX]

2 points. 1) slightly larger and more power consumptive isn't going to make a difference. Or do Japanese consumers look up the dimensions and power-draw stats and base purchasing decisions on those? 2) It wouldn't have to be any larger and louder. Again, we're not talking about putting a monster GPU into Wii, but a small, efficient, yet modern design that could achieve better results than Hollywood under similar conditions.

Here is a question... why do you think that ATI/AMD didnt advise Nintendo on the same thing? They got their name on the console, you would think they had a lot of input in the whole thing. I just wonder, if Nintendo plans to go back to ATI/AMD, what they would advise for the future console. Just recently:

"Iwata told Bloomberg that once developers begin to say that improvements to games can no longer be made for Wii, then the Big N will be ready to bring out its next console. The man also added “None of our developers is making that kind of complaint.”"

 

[obonicus]

"Iwata told Bloomberg that once developers begin to say that improvements to games can no longer be made for Wii, then the Big N will be ready to bring out its next console. The man also added “None of our developers is making that kind of complaint.”"

But isn't that a meaningless statement? Aren't people still cranking out amazing (relatively, of course) stuff on C64s and Amigas?

 

[Mobius1aic]

Well for the next generation I envision a system with a new more modern GPU with the Hollywood kind of bolted on either in the same die to save space or close to it. As for the CPU I'm sure Nintendo will go with another PowerPC or POWER based CPU, probably dual core, and possibly derived from Power6 architecture. Assuming Nintendo will want to stay a generation behind to save costs, they should go with a GPU like the RV730 (ATi 4650/70), with 1 GB of GDDR5 as it's unified RAM pool. Though I can see them going with something as low end as the RV710 (ATi 4350/4550) since it would be a drastic improvement and a decent set up for HD rendering. Who knows how far they are willing to go, but I see the ATi RV790 being the absolute extent for the Wii's sucessor.

 

[Shifty Geezer]

Here is a question... why do you think that ATI/AMD didnt advise Nintendo on the same thing?

They may well have done, or they may have just kept quiet and had the people paying them tell them what they were paying them for. Nintendo wanted BC with GC and their existing code. They got what they asked for.

 

[Entropy]

Why?
Why?
Why?

Seriously Shifty, those points you question are perfectly clear and valid, and are further explained in the "Iwata asks" series. You may not share the values that are expressed, but the reasons are perfectly clear and surprisingly frank - constraining development budgets and cutting time to market by direct transfer of assets isn't sexy for instance. But it makes perfect sense. You just don't like it.

How many of Wii's sales can be attributed to the low energy consumption? How many due to the small size?

People have different values. A lot of people DO NOT want a game console to be imposing and eye catching in their living rooms. I'm solidly middle-aged, and in my socioeconomic circles it just isn't acceptable even when it can be blamed on the kids. A game console is an embarrassment. Hide it. Something small and cool and quiet is easy to tuck away, and less intrusive if you can't. (Besides, if the console kit can be put in a small bag, people can take it to their friends place to play together - don't underestimate the value of physical social gaming.)

This is not unique for consoles, it can't have escaped you how portable computers have grown to dominate private sales, even when not actually carried around and in spite of higher cost and lower performance, for the simple reason that they can be put away when not used. Size matters. A lot.

(Anecdotal example: only one person in my circle of contacts have a PS3, in spite of a wide-spread interest in HD film. And he has exchanged it twice due to excessive mechanical noise. All others consider it a bulky, noisy monstrosity unsuitable for their living rooms.)

Your post boils down to this valid question:

Would it have been possible and of reasonable cost for Nintendo to create a Wii with a more sophisticated GPU than Hollywood?

20/20 hindsight is a beautiful thing.
Nobody in the industry could predict the success of the Wii, even Nintendo were taken by surprise by the response of the market. At the time of conception going with a design that ensured portability of assets, a short time to market and profitability even at low volumes was a good idea. And it has proven itself.

Nintendo could choose to go for further improved portability for the next generation rather than increasing computational capabilities. A fanless design would be appreciated. It would be neat if they could get rid of moving mechanical parts entirely. It might be possible to fuse portable and stationary gaming, and if they choose not to, they might instead choose to increase the graphical capabilities. Or lower prices. Or use improved (costlier) controls. Or.... We simply don't know what the successor to the Wii will look like but there is little reason to believe that it will be built to satisfy the desires of GPGPU happy posters at the Beyond3D console forums.

Why is it so difficult to accept that the Wii is built according to a set of priorities that you don't share? Be happy that it introduced (and improved with the MotionPlus) a new means of game interaction, variations of which will be adopted by the rest of the industry.

 

[Shifty Geezer]

Seriously Shifty, those points you question are perfectly clear and valid, and are further explained in the "Iwata asks" series. You may not share the values that are expressed, but the reasons are perfectly clear and surprisingly frank - constraining development budgets and cutting time to market by direct transfer of assets isn't sexy for instance. But it makes perfect sense. You just don't like it.

Not entirely true. I recognise the aesthetic appeal of a small, slim box, and would rather PS3 was that way. I see the advantages. I just don't think these factors would be that compromised by an alternative GPU such that there'd be a negative impact on Wii buyers, such that current buyers of Wii would be turned off because the case is 10% larger.

20/20 hindsight is a beautiful thing.

It doesn't need hindsight though. At the time Nintendo had options, like a Flipper refresh or an off the shelf GPU or a newly commissioned GPU. One of those options would have been an R300 derivative, 40 million transistors, 90 nm, fitting inside a slimline case, more powerful and more flexible than a Flipper refresh and easier for third-parties without extensive Fkipper experience to develop on. The reasons against that are added cost to get the chip and Nintendo needing to learn some new programming models.

Nobody in the industry could predict the success of the Wii, even Nintendo were taken by surprise by the response of the market. At the time of conception going with a design that ensured portability of assets, a short time to market and profitability even at low volumes was a good idea. And it has proven itself.

Yes, there's no denying it worked. I just don't agree with the decisions made to bring it out, and certainly don't see the reasoning extending into the next generation of Nintendo hardware which is the thread topic.

Why is it so difficult to accept that the Wii is built according to a set of priorities that you don't share?

I do accept that it was built around priorities I don't share. I just don't agree with those priorities and am arguing my reasons why I believe those priorities are wrong, and why TEV is no good for next-gen hardware. There is not a single direction Nintendo could go (apart from backwards compatibility up the wazoo) where an existing solution isn't better suited than TEV. There are mobile GPUs if they want the smallest, most power efficient device. There are off the shelf options. There are monster processors and versatile CPU/GPU hybrids and scalable solutions aplenty.

 

[Entropy]

I do accept that it was built around priorities I don't share. I just don't agree with those priorities and am arguing my reasons why I believe those priorities are wrong, and why TEV is no good for next-gen hardware. There is not a single direction Nintendo could go (apart from backwards compatibility up the wazoo) where an existing solution isn't better suited than TEV. There are mobile GPUs if they want the smallest, most power efficient device. There are off the shelf options. There are monster processors and versatile CPU/GPU hybrids and scalable solutions aplenty.

I'll meet you and admit that the Wii, although I like the design ethos, is a bit less than I would prefer from a GPU standpoint. I just don't think it matters a whole lot in the greater scheme of things.

I don't agree that the alternatives open to Nintendo at the time of decision making was all that appealing given the design constraints they wanted to work within. At this point in time however, there are more options, and Nintendo is perhaps the most interesting company to try to predict because they really have a multitude of paths open to them. If you want to have any chance of being correct when making predictions, it makes sense to see what they have prioritized before and how they reason.
* Their policy is to make game devices, not multimedial media hubs, or computing platforms, they have no other line of business they desire to tie into or strengthen or market to manipulate. That is not to say additional functionality won't be there, just that the purpose is strengthen the overall appeal and position of the games console, not to tow any other line.
* Nintendo prioritizes usability. Battery life, size, ruggedness and so on are considered important. They put a handle on the GameCube!
* They seem to have a preference for in house or custom designed electronics. Don't know why really.
* They don't care about specs for their own sake.
* They don't design hardware that is immature, but will work well two lithographic generations down the line. Their devices are, in that sense, mature from the get go.
* They like novelty, Virtual boy, Dual Screen, Wiimote - they like to bring something new to the table.

IMHO, when speculating, the more of these properties that fit the prediction the more credibility it has. (For the record, I believe they will upgrade the graphical capabilities for their next generation offering, but I emphatically don't believe that it will be the focus of a new device.)

 

[Shifty Geezer]

I don't agree that the alternatives open to Nintendo at the time of decision making was all that appealing given the design constraints they wanted to work within. At this point in time however, there are more options, and Nintendo is perhaps the most interesting company to try to predict because they really have a multitude of paths open to them. If you want to have any chance of being correct when making predictions, it makes sense to see what they have prioritized before and how they reason.

I've got no idea what Nintendo are going ot do! I just think, asked "Is TEV a keeper or a loser", it's a loser. Nintendo may go with it anyway despite my protestations!

 

[Entropy]

I think fearsomepirate has a point in that Nintendo probably don't care that much about where PC graphics technology is at. Their current devices, the DSi and the Wii, have capabilities that made sense for Nintendos purposes, they are not extensions of off the shelf PC parts. What this means for a future device is difficult to say though, but if using fixed function or limited programmability GPUs saves gates/unit performance, and thus saves power and money, Nintendo is unlikely to care much what shenanigans nVidia is up to to convince investors that they has a business model beyond games accessories.

Or put in other words - if the next Nintendo console supports IEEE compliant FP64, I'll eat my hat.

 

[fearsomepirate]

I guess one difference between me and you is that I don't think we can know what would have happened otherwise. Reality has too many variables. What we do know is what actually happened, and that is that Wii is a very successful product without Nintendo having done what everyone on this forum thinks they should have done.

Let's put it the whole another way with a few simple questions -

1. Would it have been possible and of reasonable cost for Nintendo to create a Wii with a more sophisticated GPU than Hollywood?
2. Would making that choice lead to an insanely expensive machine, one with an insane power draw, a stupidly large console, or require many times the development budgets compared to current budgets?
3. If Nintendo had made that choice, would they be struggling to make money now, or would they still be making multu-billion dollar profits?

Answers:
1. I don't know. Nintendo defines "reasonable cost" from the context of its own corporate goals, shareholder expectations, and risk analysis. The market goals of Wii made it a very risky machine.

2. I will answer with another question: If you're angling for new, non-gaming customers, how do you make these kinds of decisions without hindsight or past data? I think Nintendo's decisions were smart: They don't know how big a machine has to be to turn people off, so they make it as small as possible. They don't know the ideal price point, so they give themselves lots of room to cut price if necessary. They don't know how the "always on" thing will work out, so they minimize the power draw. They want to release with Twilight Princess, so they derive the hardware from Gamecube.

3. It's hard to say. If the additional cost made them too risk-averse to make software like Wii Sports and Wii Fit, then they wouldn't have sold 50 million Wiis. Customers can also be very price sensitive as well; would they have 50 million Wiis out there with a $300 machine? Maybe they couldn't have gotten Zelda ported to the new machine in time; would they have sold 50 million Wiis without a major IP at launch?

The market response says the Wii is a good deal. In business terms, it's simply impossible to say Nintendo made the wrong decision with its GPU--they are raking in money and have sold 50m of the things. Like I said, reality is too variable--maybe Nintendo would have made 10m more customers with a fancier system. Or, maybe they would have lost 20m with all the other things that changed. And from a business standpoint, releasing a DSi-like successor to the Wii makes a lot of sense.

 

[Shifty Geezer]

All sorts of stuff

Much as I disagree with your view, and won't be budged, you make some fair arguments.

 

[AlNom]

1. I don't know. Nintendo defines "reasonable cost" from the context of its own corporate goals, shareholder expectations, and risk analysis. The market goals of Wii made it a very risky machine.

Cost will be related to mm^2. Answer is clearly, yes, they could have gone with a more flexible hardware that is at least more conducive to current programming models. All else follows for power consumption because the chip would be the same size and with an appropriate clock speed (or set of clock speeds for multiple domains).

 

[MDX]

So as far as I've read from this thread there is practically no TEV defenders for Nintendo's next generation console?

None that at least say that it can be improved upon to compete against GPUs that
the competing consoles might use in the next generation.
Has it no future in Nintendo keeping a future HD console small, quiet, inexpensive?

By the way, do any of you know if any Nintendo portables ever used TEV?
There was once that rumor that Nintendo's scrapped portable was going to be a portable GameCube. Im wondering if this is where Nintendo is (was) trying to go with it.

 

[swaaye]

So as far as I've read from this thread there is practically no TEV defenders for Nintendo's next generation console?

None that at least say that it can be improved upon to compete against GPUs that
the competing consoles might use in the next generation.
Has it no future in Nintendo keeping a future HD console small, quiet, inexpensive?

It's old tech. Technology moves on. TEV and the rest of the GPU was a graphics solution developed with knowledge and limitations from a decade ago. Just like we don't use DirectX 7/8 level GPUs anymore, we shouldn't be seeing more Flipper tech. Today's tech is superior in every way, even cost. It's evolution and we've progressed in many ways beyond every aspect of the concepts in Flipper and those other old GPUs.

 

[Entropy]

It's old tech. Technology moves on. TEV and the rest of the GPU was a graphics solution developed with knowledge and limitations from a decade ago. Just like we don't use DirectX 7/8 level GPUs anymore, we shouldn't be seeing more Flipper tech. Today's tech is superior in every way, even cost. It's evolution and we've progressed in many ways beyond every aspect of the concepts in Flipper and those other old GPUs.

I'd argue that the old way of doing things is superior in delivered performance per gate (and thus power draw and cost) than the newer stuff. Provided, of course, that you what you want to achieve is doable within the confines of what the older tech allows.

Going forward, Nintendo have a number of options open to them, some of them allowing the TEVs a future.
One of them is to talk to their developers, ask them where they would prefer to have the current architecture extended and talk to AMD about what can actually be done in terms of extending the current archtecture within reason. They then scale this to the point where it can comfortably handle 1920x1080, see what this would cost in die size and power draw, and if it's OK, they go with it, otherwise they go once more through the chain to see where they can compromise with minimal impact.

This would produce a graphics solution that would be backwards compatible, adresses the most acute developer concerns, gives the new console sufficient performance to look OK on the displays in use during its lifetime, and which doesn't waste gates on features that has little relevance for their games console.
It probably wouldn't fit a DX7/8/9/10/11 description, it would be a custom solution unique to Nintendo.

If they decide that complete backwards compatibility is a priority, I'm sure they will at least explore the path outlined above.

 

[tongue_of_colicab]

Is it realistic to think that there is enough stretch in flipper to really make it something ''next gen''? Wouldnt it need a big rebuild? Isnt it the most logical to just go with a standard cpu? I understand why the went with a souped up flipper for wii but it wouldnt make sense to do it again unless it would be really easy to upgrade flipper would it? AMD can probably give them something easily can do what a HD flipper could at the same price/power usage and other than nintendo themselves developers would probably be off cheaper with a ''normal'' gpu right? Because Wii is being the odd one amongst xbox/pc and PS.

Also would they really need flipper that much for BC? There is a fairly decent wii emu for pc already. If ''amateurs'' can do that, than surely nintendo must be capable of building something that works good enough for the most part. Also, how important is BC really? x360 has crappy bc, ps3 doesnt have perfect bc either (apart from the old ones) but how many people really didnt get a console because of not having BC? I dare to say close to none. Probably even more so for the Wii casual crowd. How much of the would still care about their old wii fit/brain training as soon as they have the new and improved one?

I think BC is overrated for consoles.

 

[Squilliam]

Also would they really need flipper that much for BC? There is a fairly decent wii emu for pc already. If ''amateurs'' can do that, than surely nintendo must be capable of building something that works good enough for the most part. Also, how important is BC really? x360 has crappy bc, ps3 doesnt have perfect bc either (apart from the old ones) but how many people really didnt get a console because of not having BC? I dare to say close to none. Probably even more so for the Wii casual crowd. How much of the would still care about their old wii fit/brain training as soon as they have the new and improved one?

I think BC is overrated for consoles.

Considering that a typical Wii owner may have over $200 invested in accessories by the end of the generation, some backwards compatibility at least for the physical interfaces may be desirable. A fully loaded Wiimote costs $80 and theres usually two. Add Wii Fit and an extra M+ module for the original control and it quickly adds up.

 

[Entropy]

Also would they really need flipper that much for BC? There is a fairly decent wii emu for pc already. If ''amateurs'' can do that, than surely nintendo must be capable of building something that works good enough for the most part.

Of course Nintendo can also opt to go with emulated backwards compatibility, or for that matter, drop the ability to play Wii titles entirely.
And for each of these options (hardware BC, emulated BC, and dropped BC), there are a multitude of scenarios. Nintendo really are the most difficult company by far to predict.

Gates (transistors) cost money and power. If they choose to go the emulation path, I still don't believe they would choose an off-the-shelf PC part. It isn't Nintendos modus operandi, and the PC designs are pretty far from a good fit to Nintendo stated design goals and ethos. Furthermore, Nintendo volumes are really large compared to most GPUs. I remember seeing claims that production of the RV770 chip had passed the 2 million mark, and that has been a hugely successful product in the marketplace. Contrast that with Nintendos 50 million and growing Wiis to date, and it seems a custom design for Nintendo is quite reasonable.

I don't know. If they go with emultion, 24 bit FP would ensure better precision than 8 bit fixed point. A feature set that allows trivial mapping of old GPU features to the new architecture would be a given, they could afford to design away most or all corner cases. Conserving gates would be a priority, so features that don't bring solid benefits/gate for this particular console should be excluded. Certainly no high precision FP and no features that target GPGPU for instance. But I'm incapable of predicting much beyond that.

And if they decide to drop BC, all bets are off.

 

[Tahir2]

I don't gamble but Wii accessories will not be compatible with Wii2.

Call it a hunch...