Acceleon in action!

I've played Doom on a Sharp Zaurus. I've also played ported PC adventure games such as Space Quest and Monkey Island. Let me say that those LCD screens are far more usable for gaming than you seem to imagine Nick.

There are some software 3D engines out there for games on the Z, check out http://www.eongames.com. They are quite rough, simple, and slow. They do get the job done for these simple games though.

About CPUs in PDAs. There is no PDA CPU that has a FPU. They all have bad latencies to RAM, and no L2 caches, and very limited L1 caching. They are not exactly overly capable for software rendering. The 400MHz XScale has some serious performance issues if you read up on some boards. The 200MHz Strongarm usually can outperform it. (Though this is almost certainly due to inefficient code, and that programmers are not taking advantage of its instruction set)

Nick, I can't see why you are so strongly against 3D hardware in PDAs. There really are no disadvantages to adding more capable hardware. It's not like these super high end PDAs have battery life right now anyway :)

With increasingly better manufacturing technologies I don't see why we can't have DX6 level graphics in a PDA with little battery-level side effects.

Ah well, I'm always interested in seeing what the engineers can pull off. I wish BitBoys luck, and hope they've finally found a market that will let them play around. Same for the PowerVR MBX guys.
 
swaaye said:
Nick, I can't see why you are so strongly against 3D hardware in PDAs. There really are no disadvantages to adding more capable hardware. It's not like these super high end PDAs have battery life right now anyway :)
I'm not -against- 3D hardware in PDAs, I just don't see it happen. I've got a Nokia 3310, and I know it's screen isn't comparable to those new color screens, but the Space impact game that comes with it is hardly playable because you get a cramp in your fingers. The current generation of fast 3D games on the BGA look to me like nearly the best thing doable with such screen and controls, so a CPU upgrade seems more logical to me.

But I do wish BitBoys all the luck with convincing manufacturers to buy their products. I'm afraid though that Acceleon technology will hardly influence the customer's decision. Like I said, time will tell to see whether I'm wrong, but you guys are soo heavily biased towards hardware acceleration from the PC platform that you think it can't go wrong.

For the console market it took a while for the developers to give up their freedom because of increasing CPU performance. With it's inherently lower resolution and quality it's not unthinkable that new consoles will feature software rendering again. Even on the PC we see games like Unreal Tournament 2004 get their software rendering support back. Now with handhelds the complete step to hardware rendering might totally not happen...
 
Dont forget these cores are relatively small and inexpensive compared to the price of the machine ... the high end ones will usually get a 2D core anyway, which drives the premium down even further.

To top it of when you are playing 3D games on them the ASIC will always give the equivalent quality of a software solution at a lower power consumption. It all depends on how much importance 3D gaming has to consumers ... the lower power consumption and higher quality can obviously offset the increased cost when it does become an important enough factor. For game devices this is already the case at this time IMO.

I could be wrong, because the high tech industry as always thinks it is in their best interest to keep price hidden and only release such information under NDA ... but area wise the cost of adding these things to a SOC are minimal.

I think manufacturers which make devices for which gaming is an important application and which ignore this, like the N-Gage/Zodiac/Taplay, will start to downplay 3D gaming and putting more emphasis on other games/applications when comptetitors start adding 3D cores and totally outclassing them in every way the moment you play a 3D game.

The PSP is a monster, and it will set a standard ...
 
Like I said, time will tell to see whether I'm wrong, but you guys are soo heavily biased towards hardware acceleration from the PC platform that you think it can't go wrong.

Not really. There´s a difference between saying that you need a healthy combination of CPU and graphics power, than insist that JUST either/or will be adequate.

Even on the PC we see games like Unreal Tournament 2004 get their software rendering support back.

You can´t be serious about that one, using it as an argument for the specific point. The only other reason why Epic chose to implement it is that they´re trying to catch a much wider userbase and not because it would represent in any way optimal rendering for the game, rather the abysmal opposite.

For the console market it took a while for the developers to give up their freedom because of increasing CPU performance. With it's inherently lower resolution and quality it's not unthinkable that new consoles will feature software rendering again.

Next generation consoles are goint to sport most likely PS/VS3.0 shaders, where the architectures come much closer to CPU logic. CPU´s are still going to be needed for AI and physics, but consoles reverting to software rendering is more of a joke.

I bet I won´t be able to crank out a higher than 20-25fps average in 1024 on the current 2400+ XP in the original UT, whereby if I switch to 3D rendering w/o any AA/AF in 1024*768 the minimum framerate will be >60fps, the average framerate is even too embarassing to mention. UT2k3 is out of question anyway, since it´s heavily T&L optimized.

I haven´t had time to check your Q3a renderer any further, but is that what you mean?

Q3.JPG


:rolleyes:
 
Ailuros said:
I haven´t had time to check your Q3a renderer any further, but is that what you mean?
Is that on an Athlon XP 2400+? I get exactly the same framerate on my Celeron 1200. :p What were you running in the background?

Besides, I already said it's far from fully optimized. The renderer is close to optimal but the TLC is in plain C. And the most important reason why it could be much faster is because Quake III totally isn't targeted at software rendering! I just used the same PVS because it was a fast way of reducing polygon count. For a software renderer a portal clipper like in Unreal is much more efficient. I can make it at least as fast as Picomatic if I had the time, but I'm just a student and right now I'm having exams...
 
I was online, which reduces framerate by a couple of frames, but hey q3a is more a multiplayer than anything. Besides that's a completely empty scenery. Even if one would assume that you'd optimize it further and get into the >40fps department, as soon as a couple of bots jump in and the heavy firefights kick in, it's going to be single digit glory. I understand it's just an experiment, but the sky (or what's supposed to be the sky) looks quite ugly.

I'm really too bored to reinstall the original UT and switch to software rendering, but I doubt anyway that my estimate was too low.

Oh and since I forgot to mention it, I really and honestly wish Bitboys to be successful with their Acceleon. This conversation is not directly related to their hardware at all.
 
i just played some of the original UT witht he software renderer.

At 640x480x32bit, i got from 37 to 47 FPS, avg for a good botmatch with 4 bots was 43 FPS.
At 800x600, it became IMO, unplayable, with drops down to 25fps and an average of about 30 FPS
i didnt bother to try 1024x768.

And yes, it was ugly even at 800x600.
 
Ailuros said:
I understand it's just an experiment, but the sky (or what's supposed to be the sky) looks quite ugly.
Uhm, do you even have any idea what you are talking about? That's the way it's stored in the file format, and a shader file describes how it appears. I just didn't take the time to implement it, but a simple projection shouldn't slow it down at all. And you should be glad I got the Bezier curves, BSP traversal, visibility, lightmaps, gamma correction, scene models, etc. working or it all looked a lot worse. But you probably didn't notice that or thought it's all done automagically? Sigh...
Oh and since I forgot to mention it, I really and honestly wish Bitboys to be successful with their Acceleon. This conversation is not directly related to their hardware at all.
Like I said, I wish them all the luck as well! I'm just afraid they won't be very succesful and maybe only a few geeky handhelds will use their technology. In my opinion there's just no use for 3D hardware on tiny screens with cumbersome controls. Ok graphics will look stunning but is it just me who cares about playability and gameplay?

But me nor you can predict its future, so Kristof was right... time will tell.
 
I have to say, I'm on record here as being pretty much on the same side as Nick on this.

The absolute screen real estate size just makes me question the whole need for dedicated 3D on PDAs...let alone cell phones. I've said this ever since 3dfx acquired gigapixel and started evangelizing the idea.

3D on a cell phone? I just keep completely scratching my head.

Dedicated gaming hand-helds....I can certainly see some 3D being useful, but still not convinced that the limited screen real estate will really require it.

3D on a PDA...questionable...the vast majority of people that I know that own and use a PDA, still don't use anything more than calendar and phone book. (yet, that doesn't keep them from upgrading their PDA every year or so....it's almost more like a status symbol than useful...)

The only use I can see on a PDA is for the set of individuals that want a PDA and a dedicated gaming console.

Every time I hear one of the graphics companies mention "cell phones", I cringe a little bit. I mean, I don't even recall those cell-phone / PDA combos being met with any particular success....

I am much more bullish on 3D chip designers pushing into set-top cable / satellite receivers and now built-directly-in to TV digitial receivers...

But...time will tell....
 
From dealing with even simple 2 color 132x64 LCDs commonly found in portable devices like MP3 players, etc. Let me guarantee you that CPU power alone quickly becomes untenable as the solution. Even simple scrolling and outputting the LCD frame starts to take about 10% of our CPU budget when we get "fancy". Managing any sort of interesting content (higher bit depth, faster frame rate, larger display) is right out.

While I'm not sure "3d" hardware is necessary, some sort of bitblt, multilayer, blending, etc is a must. You just can't do that efficiently in software.
 
another vote against 3d pda .... :) , but I also have an idea how&when 3d can become useful ....
- when one will be able to use glasses as display and gloves as "joystick" :)
ahhh, and what an idea - imagine PDA smart enough to autostart your latest 3d FPS, the moment it senses danger for your life .... LOL, "Hitchhikers guide to the Galaxy" rulz!
 
RussSchultz said:
While I'm not sure "3d" hardware is necessary, some sort of bitblt, multilayer, blending, etc is a must. You just can't do that efficiently in software.

Agreed. I can certainly see the relevance of "hardware" 2D operations such as those you mentioned, and moving all the way up to AAd fonts or even full screen AA.

I am personally only leery of pushing 3D hardware down into these small devices.
 
Look its simple... millions of people are buying gameboys with a screenres of 240x160 and they play games on it for hours and hours... playing a 2D or 3D game should not matter. We are now talking about devices with displays of, say, 320x240, and this would be insufficient for playing games ? Phones today come with displays of 176 x 208 pixels not all that much different from the current gameboy so why would 3D or gaming be impossible ? Button trouble can and will be solved by some ingenious design. Display too small... well if the resolution is there it all will just look like higher dpi prints... aka better. Small displays are not a problem, the number of pixels is what counts, high dpi's just looks better and smoother on their own.

How people you know use PDAs is really not relevant, the key driving force for this market is : gameboy sells, people play games (sucky things like snake) on their phones for quite a bit of time (not permamnent but lots for small amounts of time)... why not sell them something with high quality 2D and 3D graphics ? And maybe make some money of it...

Screensize or pixel res IMHO seems like no problem, it works for millions on the gameboy so it should work for you and should work with 3D. Buttons is something people will find a solution for. PDA usage or Phone usage... games are being played today and if games get better more and more people will play.

Settop box and satellite has been pushed for many years now (ATI, PowerVR, NEC, STM) and with what success ? We'll have to see what the push into PDA and Mobile will do...

As said before time will tell...
 
I would'nt say that i'm against 3D on PDA's, smartphones and such. But i'm definitely questioning the need for it. Though there are some games on my GBA that i think would work in "real" 3D even though the screen is small. Mario Kart Advance f.e.

Only problem is that i prefer Super Mario Kart in front of the N64/Gamecube versions :)
 
Kristof said:
Display too small... well if the resolution is there it all will just look like higher dpi prints... aka better.

This is our fundamental difference.

To appreciate my point of view: Imagine a 1600x1200 screen the size of a pin head, viewed at the typical distance of a handheld, (1-2 ft.)

Wow...really high DPI. Does that mean can play Quake on it, better than a 15 inch screen at 320x200? DPI isn't the only measure of the amount of useful information that can be garnmered from any give display. It's also the "I" part of DPI, that is absolute size (relative to our total field of vision.)

Somwhere between a typical PC display size, and a pin-head, we start getting too small in terms of an absolute sese for screens with "lots" of information to be communicated clearly. It is merely my belief that "cell phone screen size" is too small, and PDA size is right around the limit.

Small displays are not a problem, the number of pixels is what counts, high dpi's just looks better and smoother on their own.

Again, I disagree, and I don't see how you can't at least admit that screen size is a problem. It's just a question of at what size the problem is generally realized.

How people you know use PDAs is really not relevant,

Sure it is. It's not scientific by any means, but demand is ceratainly relevant to the success of a product.

the key driving force for this market is : gameboy sells...

Right, because it presumably has fun games. Not necessarily 3D games.

people play games (sucky things like snake) on their phones for quite a bit of time (not permamnent but lots for small amounts of time)... why not sell them something with high quality 2D and 3D graphics ? And maybe make some money of it...

I'm not knocking giving it a go at all. I'm saying that "sucky things like snake" are played because they are addictive and "fun", despite not require 3D. And a "good, clean" display of graphics doesn't require 3D. 3D is used (generally) to convey more information in a given amount of real estate, than you could in 2D. This allows you to make a wider variety of games.

But therin lies the rub...if the screen real estate is small enough such that 3D can't really convey thaty additional information effectively...then it's not going to help.

Settop box and satellite has been pushed for many years now (ATI, PowerVR, NEC, STM) and with what success ?

Quite limited to date, because set-top boxes don't require such 3D features either. That's the point. And it is questionable if standard TV and broadcast services will every really benefit from 3D displays either. Though the reason why is clearly the opposite...it's not so much display issues, it's utility.

Longhorn is going to be interesting. A first real crack at a 3D user interface. If longhorn (or a successor) is successful in this, we might see the same concepts being carried over to TV displays, on-line guides and interactive content, etc.

However, We'll have to see what the push into PDA and Mobile will do...

Yes, and I seriously do look forward to watching it unfold. I suspect we'll see decent success in the hand-held dedicated game market, "limited" success in the PDA market, and almost nil in the phone market. Just my opinions.

As said before time will tell...

Indeed.
 
Joe: as a notice, Acceleon provides also 2D acceleration. so it is not all about 3D. one thing that should be telling this too is that G10's target market is anything from your car stereo / positioning system OSD to cellular phones.
 
Nappe1 said:
Joe: as a notice, Acceleon provides also 2D acceleration. so it is not all about 3D. one thing that should be telling this too is that G10's target market is anything from your car stereo / positioning system OSD to cellular phones.

Of course, the point is, that adding 3D support will add to product cost....and will that cost ultimately be justifiable in the cell-phone market where, AFAIK, ability to render 3D images is far down on the list of priorities of what cell-phone customers demand.
 
Bjorn said:
I would'nt say that i'm against 3D on PDA's, smartphones and such. But i'm definitely questioning the need for it. Though there are some games on my GBA that i think would work in "real" 3D even though the screen is small. Mario Kart Advance f.e.
I'd like to say that 3D as we know it on the PC will never work on a hand-held (cell phone, PDA, gameboy, or whatever). But that doesn't mean there is no reason to go for 3D at all. After all, 3D as we know it on the PC doesn't even work well for game consoles that connect to the TV, though the comparison is more similar.

I would tend to expect that 3D for consoles would be more simplistic in nature. Racing games and perhaps some sort of flight sims would only work well with proper 3D rendering (I would expect arcade-like controls for both). But any type of game will look better when rendered in 3D, including side-scrollers.
 
The difference between 2D and 3D is quite significant

1) on average, in a 3D title, every frame, every pixel receives a new value (gets written)

2) in a 2D title, much of the time, only fraction of the screen pixels need to be altered. This is done via sprite techniques, and scrolling.

For example, if you take a typical side-scrolling platformer, when you reach the "end level boss" typically, the screen as a whole no longer moves, and the only pixels that continually change are those related to your character, the boss character, and bullets/explosions, etc


Your argument for a more powerful CPU will fall on deaf ears in the mobile industry. You're lucky if you get more than a 100Mhz CPU (the MS SMartphones are getting 250Mhz XScale). Because of cost, size, heat, and power issues, mobile manufacturers are looking for products that are power efficient, cheap, and easy to manufacture. (Nokia for example, routinely rejects RAM increases for their phones because of slightly less battery performance, or an extra $1 or two of cost)
 
Uhm, do you even have any idea what you are talking about? That's the way it's stored in the file format, and a shader file describes how it appears. I just didn't take the time to implement it, but a simple projection shouldn't slow it down at all. And you should be glad I got the Bezier curves, BSP traversal, visibility, lightmaps, gamma correction, scene models, etc. working or it all looked a lot worse. But you probably didn't notice that or thought it's all done automagically? Sigh...

I did state that I realize it´s just a quick experiment; either way it´s still unusable even if you optimize it to crank out double the speed it already is. Interactivity is going to bring the CPU to it´s knees non period.

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

Joe,

I have personally no single interest in gaming on handheld or small devices, let alone a mobile phone.

My current mobile phone though, as of course the display on my car cd/audio system have trouble to even display simple text and the limitations are more than just obvious.

Get a healthy combination of CPU and graphics power in analogy to upper thresholds of power consumptions and most limitations are gone and it makes little difference what you use it for. Neither just the graphics chip alone can replace the CPU, nor the opposite will.

Ever seen how dogslow car navigation systems or GPS displays are these days?

Somwhere between a typical PC display size, and a pin-head, we start getting too small in terms of an absolute sese for screens with "lots" of information to be communicated clearly. It is merely my belief that "cell phone screen size" is too small, and PDA size is right around the limit.

And that´s the reason why 3D doesn´t make sense on small handheld devices? Keep it all on the CPU, let the MHz there scale and you´ll be searching for a power plug everytime you have an incoming call.

As said before time will tell...

Half of the world´s largest semiconductors must be completely out of their minds. ;)
 
Back
Top