I don't think nintendo will ever again advertise performance as a selling point. Their customers don't care about it, so why should nintendo bring it up? They may release a very capable piece of hw and you will never hear anything about its prowess, bar developer comments & armchair analyses.
NV got DS2 contract according to BSN
- Thread starter TEXAN*
- Start date
This is what came to mind the first time the rumors sprung up. Does the philosopy borrowed from Gunpei Yokoi match nVidia's proposal? I would hardly call Tegra "off the shelf". But if there's any credence to the whole "high resolution graphics" statement Iwata claims, then it seems Nintendo would not be shy of using higher end hardware when needed (also take into account Miyamoto's argument that Ninty's next console would most probably have HD given the prevalence).
Pushing everything through the same logic would result in a pretty slow game engine imo, I've heard from multiple developers that they handle translucency through a different path to opaque as is logical for something that needs special case handling.
I wasn't suggesting that alpha test can replace translucency, however foliage is a great example where atempting to sort by depth would be both practically difficult and slow.
Quite possibly, as you say a loose sort of some sort on at least some of the translucent data is inevitable.
You generally seem to be saying the opposite to what most of the devs I speak to say, state multipliciy does not generally fix the batching problem (at least not without introducing other problems) and translucency *is* normally handled within a different path to opaque/alpha test data. Given that it's opaque data that dictates your overdraw I go back to my original point.
To be honest I'm surprised you haven't raised deferred lighting as a justification for early Z machines multiplying by such a high an overdraw factor, although that just trades overdraw reduction against the huge bandwidth penalty paid for additional depth buffer acesses and for sending input geometry multiple times.
John.
If their customers don't care about it then why spend money on it? Nintendo only stopped advertising performance because their performance stopped being competitive.
Panajev2001a
Veteran
You are answering your own question and your next line just states an obvious side-effect of it
.Nintendo are a game developer and publisher. They release their own hw that satisfies their own primary business. They sell hit software titles, not 'gigaflops' or 'terabytes' - nintendo don't sell anything to developers. Or at least they don't sell anything to developers on performance merits (they sell it on userbase merits). Last but not least, the performance of their hardware appears to be quite competitive for what they sell on it. Yes, it would suck for HPC uses, but i doubt nintendo lose much sleep over that fact.
Unless translucency is not subject to state batching, latter is just another sort from the POV of the drawing end of the pipleine. If you're going to sort by state, you can just as well sort by state + spatial order - that would not be any slower.
Right, but how is state sorting absolved from that? Under such conditions the pipeline should not attempt *any* sorting, or if draw order was not an issue but state is, cut the spatial key from the sort; basically, switch the sorting keyset from one including spatial info to one not including such (as we assume you still want state batching). Neither difficulty nor speed complexity would change.
Most pipelines I've worked with do not make deliberate attempts to impose any state multiplicity discipline/methodology whatsoever. Even for such heavy hitting state elements like texture residency such pipelines would leave everything to the final stage - default state batching at draw. And I realize that you're speaking from the position of deferred rendering, where your hw offloads (part of) spatial sorting work from the client, but from where I stand I may not have the luxury to delegate that to the underlying hw. IMRs are still common, and opaque overdraw is not something to be taken lightly, right?
The hardware I predominantly work with does not have spare bandwidth to burn. While input geometry multiplicity can be addressed through transform feedback buffers (subject to availability), the algorithm's texturing bandwidth appetite is just unsurmountable - there's only so much you can drop your g-buffer's vector fidelity before it starts to show in the shading. That's why i'm sticking with ubersaders for the time being. That also saves me the need to explain to the artists what eats their precious bandwidth for 1:1 HD-res textures : )
Last edited by a moderator:
You don't sell gigaflops and terabytes to developers, you sell them to customers. If you have powerful 3D hardware that blows away the competition then you're foolish not to say something to this effect.
For this generation the performance of their hardware sucks compared to the performance of what the other guys put out. If it didn't suck then they'd probably be advertising that fact.
As a means to an end for their own game goals I don't think Nintendo cares much about high performance either.
Panajev2001a said:
I'm not asking why they're not spending money marketing it, I'm asking why they'd spend money on developing and manufacturing performance they don't care about. darkblu is making the claim that Nintendo won't advertise high performance that they have because their userbase doesn't care - I'm claiming that if their userbase doesn't care they won't make high performance handhelds to begin with.
MfA said:
And Nintendo is so far not marketing this.
Do they really need SGX+ level for stereoscopic 3D at 30FPS? DS is frame locked at 60Hz, so it already has the rough capability to perform stereo quality of what it's outputting at 30Hz instead if some additional buffers were added to accommodate this. Process bumps over the 2004 DS 3D hardware plus a gross increase in the internal geometry buffers (say, from dozens of KB to hundreds of KB) would go a pretty long way to improve the platform without actually changing the technology.
People are also looking at the "high resolution" claims as an indicator of performance but I think what's driving that is not even the 3D acceleration but the stereoscopic displays, which are probably not even available in anything under "high."
I'm sure that's what sony were thinking when the notorious 'customers will get two jobs to afford a ps3' came out.
How much advertising of technical prowess did the cube receive last gen? And it was a heck of a good game console design - arguably the best one across the board.
Of course they wouldn't. Because it does not sell their machines. Because their customer base does not comprise of coders and/or teens or teen-minded DINK males.
My post was all about marketing, though ; )
This is irrelevant to what I'm saying, although it actually strengthens my argument that Nintendo isn't intending on selling cutting edge hardware tech.
It received plenty. You might not remember very well because it was released 9 years ago. You do have an interesting concept of "across the board", though; I think it was pretty well behind XBox. Or are we talking about something outside of technical capabilities when we mention "good design"?
I feel like I'm talking around you.
You design to your audience. You advertise to what you've designed. Nintendo's no exception to this rule of common sense.
He was responding to me. I do hope it's now clear to everyone what your stance is, though.
Good design for a game console includes a sustainable product (for a full console generation, for instance). A dumbed-down pc hardly matches the bill. An in fact it did not match the bill at all. The reason we had it as a console is because ms had some billions to burn, and a blitzkrieg to fight. In contrast, the cube is still sold in a facsimile form one generation later, making a dime or two for its maker. As re the amount of specs advertisements the cube received - i remember some before the launch, mainly in the form of tech demos, but that's about all i recall. The darn console was designed in the shape of a lunch box, for pete's sake - what appeal did that have with the 'omg, gigaflop' crowd?
No. You design to your audience and you advertise to your audience. What you've designed warrants advertisement only in those aspects that communicate with your audience. How many red bull commercials have you seen stressing the chemical composition of the product (i.e. taurine & caffeine)? Or do you believe gillette razors are actually 'advanced shaving systems'?
Once again, nintendo's audience does not buy computational power, they buy something else. That in itself does not imply nintendo consoles have zero computational power.
Sorry, I felt like i had to clarify as i was not sure you caught that aspect of my original post to you (your reply to which Panajev2001a was referring to).
I strongly contend that Nintendo is successful because of their games and controller concept, not because they're reusing Gamecube hardware. If Microsoft had the first/second party IP, brand appeal, launched with a Wiimote, and had the actual rights to the design then I see no reason why they couldn't have reproduced Nintendo's success while only modestly improving on the XBox design. And they probably would have been able to sell it for less than Wii's go for. Even XBox 360 was sold for less than Wii for much of their competing lifespans thus far.
You may think that XBox was not an elegant design but that says nothing for its ability to get the job done, a characteristic it showed no substantial technical handicap in (unlike say, N64, a product whose technical capability was lauded by Nintendo long before release but had some crippling limitations). XBox was held back by company stigma and third party interest (and snafus like initial controller design), not because of what was inside the box. Well, technically speaking anyway; it got a lot of bad press for being "just a PC", but if the audience didn't know this then they would have been none the poorer off for it.
Sorry, let me clarify; you advertise the positive points of your product. You seem to be under the strange illusion that advertising technical capability would be damaging for Nintendo, that is, if they were actually competitive. Your "zero computational power" comment is obscuring the issue, this is clearly about relative capability.
Nintendo has even promoted Wii's hardware. They just kept it vague enough so people didn't realize it sucked.
I've not devled into the causes behind nintendo's success. i've been merely discussing the viability of the cube as a console design, and why it was better than another design you brought up against the cube.
We have two designs: one of them was carried over to the next product generation, the other was abandoned early, never to be returned to again. By any measure of design success, one them wins over the other.
If ms considered the xbox a successful design in any shape or form, they'd have carried over something from that design into their next console. Ironically, xbox360 has much more in common with the cube architecturally, than with the first xbox.
I suspect we may have different understanding what 'the job' for that device was. Here's mine: from the POV of ms it was the proverbial trojan horse - the device to let them a footstep onto a new and highly competitive market. From the POV of the established market rules, though, the original xbox was anything but a technically viable product - any other vendor would have choked on it; ms choked on it, but it was acceptable for them as they played by their own rules. Once ms found themselve in the position of an established market player they shied away from the 'pc-on-a-stick' paradigm, unsurprisingly.
Oh, xbox had its own 'WTF were they thinking?' handicaps - the location of the active framebuffer in UMA was such one. It brought to notorious RAM access contention between GPU and CPU. That design decision was never repeated again on a console (the 2A crossbar controller found its way into a local controler in subsequent geforce designs, but that's an entirely different environment).
I'm not even discussing the audience's response. I'm talking from the POV of technical merits - what good was the product for its maker, if the maker actually cared about their bottomline. If ms had a few more billions to burn on that bonfire they could've brought to the game a top-of-the-crop pc confiuration - do you belive that would've been a good console design?
Not damaging. Just plain useless. I think we will be able to see that very soon - let's see how 3DS' ad campain will proceed past its initial unveiling (where some specs might be announced).
Sorry, you lost me here - how did they promote the hw?
By other measures of design success Gamecube has wins over XBox, but "how much was the design recycled for the next generation" is a very poor measure. I think your any measures fails to include the one of how good the games looked, and there XBox had the advantage. MS couldn't have carried the design if they wanted, since they didn't have nVidia's onboard anymore.
What exactly was not viable about XBox?
I said substantial - I'm well aware of the RAM contention, but I simply don't think it was on the level of N64's issues which were much more serious handicaps to either latency or capacity wherever you looked (not to mention the same common framebuffer issue).
Of course, all of the handheld SoCs we're looking at have that same issue too. Is nVidia making a fatal mistake with Tegra in having unified memory?
If MS dumped billions in giving away high end PCs that'd be one thing, but I think you're exaggerating how much they were drilling through. I think the "repurposed PC" argument is misleading, as several hardware features were actually developed for it and therefore had some of the same basic advantages any custom PC hardware has. Is your beef really with x86 being in a home console? Or a GPU design that shares a familial relation to a PC one showing up in a console (yeah, not like Dreamcast did that move)..
Your experiment is invalid because it lacks control - it only works on your presupposition that Nintendo is withholding impressive information on their console, not that they're withholding mediocre information. If they DO release specs then that'll put it to rest, but that'll make the entire argument moot, since it's speculation as to why they won't advertise them.
Of course, I still think your assertion that no user demographic cares about the graphical capabilities of a console anymore is just silly.
Do you know that Wii contains the chips "Broadway" and "Hollywood?" That's how. If you think telling us the capabilities of the hardware is a waste of time then how about telling us the names? But apparently that's supposed to impress someone.
Panajev2001a
Veteran
Moving into PR and away from coding
?The HW specs for both Wii and Xbox 360 were dependent on the kind of product Nintendo and Microsoft wanted their respective new console to be.
Because of what Wii's purpose and goals were, Nintendo was able to just enhance GCN's tech removing some bottlenecks here and there as well. MS wanted a different console, re-using any old-gen design and enhancing it was not an option.
The advantage in graphics fidelity the xbox had over the cube (as minor as it was) was irreleveant given the design was not sustainable for the market it was trying to compete in. Again, a top-of-the-crop pc would have had even bigger advantage in that aspec - would that have been the better console? Sorry for repeating myself, but you did not quite answer that question the first time.
Also, we will have to agree to disagree on this one: a design's reuse versus a design's early abandoning is not exactly a conundrum. Please note I'm talking from the POV of the design owners, not the userbase, the devs, the press, or anybody else.
NV was responsible only for the GPU/north bridge, from what i remember. MS could have easilly continued in the same lane of pc centrism, if they deemed that suitable. Instead they went with a highly-custom, bottleneck-ironed design. Why?
Its BOM and price/performance for the market it was trying to compete in.
N64 was nintendo's first (and among the world's first) 3D consoles - of course you'd have more fundamental handicaps there than in something of the same breed a few generations down the road. Why even bring that up?
Cannot speak of tegra just yet, but none of the other SoCs I'm aware of repeats xbox's framebuffer-in-UMA design. SGXs are TBDRs, and yamato are tilers too. Both offer some 'direct shader memexport' functionality, but none of that is meant to be the primary modus operandy of the parts.
Ms did dump billions in giving away upper-end pc's. And my beef is with designs poorly suited for their target markets. As for the dreamcast - it was a too advanced design for its time and/or vendor (read: sega could not afford it at that stage). And yet MS could wish the xbox was as streamlined a design as the dreamcast was.
We will know the specs of the new handheld one way or another, so we will have all the information necessary to draw sound conclusions from the experiment.
Come on now, do you really believe nintendo hoped to gain popularity among the specs-minded audience by giving out such information as the codenames of their silicon? 'Hollywood at 90nm SOI' - that blurb has got to totally win over the gigaflop crowd, yes? For all we know, it could be just a japanese custom, same way that blood type of characters in japanese RPG/brawlers is usually quoted - the point of that puzzles me to this day.
Nope, just the codebase getting fatter and fatter and the compile turnaround getting longer and longer ; )Panajev2001a said:Moving into PR and away from coding
Outside of errata fixings, TEV stages increasing and the addition of a couple of extra TEV inputs, I can't think of any architectural changes that would point out at an 'Aha! Here we had this major bottleneck' moment.
I'm not disputing that. I'm arguing about why ms went with a fundamentally different console design the second time around.
What does that even mean? Is the 3D hardware double buffered with a forced buffer switch on every frame whether the engine is ready or not? I kinda doubt it, but regardless there are 3D DS games with framerate dips already however it works exactly ... and they look like crap even then.
Panajev2001a
Veteran
RAM was one of them (Wii has much more RAM than GCN), CPU clockspeed too (Gekko was used quite a bit to help the GPU, some games like Luigi's Mansion do some of the character lighting on the CPU).
Even if they had some bigger bottleneck than that, they did not really need to solve it, not for the console they wanted to deliver
.On DS there is no framebuffer. It renders the screen per-scanline, every 60Hz as the display is redrawn. There is a fixed amount of RAM for buffering primitives output by the geometry engine. If the rasterizer falls behind this fixed number it starts dropping scanlines, but it's relatively difficult to exhaust fillrate. If the geometry engine runs out of RAM to output to then it stops accepting input until buffers are swapped.
It is possible to render the screen in multiple frames by using a capture mechanism to get the display contents and display it in a 2D layer, but it takes a lot of RAM and will cut your framerate exactly in half (or thirds, etc). Since there's no way to capture the depth buffer you have to partition the depth space and for most types of scenes this isn't easy to balance.
Framerate doesn't dip in DS games, or at least not due to the rasterization and overloading the geometry is quite difficult too. If framerate appears to drop then it's the CPU that's not keeping up with something.
DS games looking ugly is irrelevant to framerate. Or to put it another way, they look ugly because they're architecturally incapable of having scenes with more than a couple thousand triangles (and because of other limitations)
Similar threads
- Locked