Fair enough. The best I can offer you is an actual test from a DSi I ran today over the course of my business day, on a freshly-charged unit.
10:30 -
Pop Island DSiWare launched in demo/attraction mode (
game video).
17:30 - Noiticed the power LED was red.
18:09 - Power LED went blinking, at which stage I shut down the system.
Total run time from fully-charged to 'shut-down-now!' indication - 7h 39min. Let's go see what Nintendo's fact sheet has to say about that (my DSi's brigness level bolded):
Code:
Brightness Approximated Continuous Play
1st level 9-14 hours
2nd level 8-12 hours
[B]3rd level 6-9 hours[/B]
4th level 4-6 hours
5th level 3-4 hours
The problem I have with this is that it's giving Nintendo benefit of the doubt that the screen brightness really eats away so much, this is not an exhaustive verification. It's pretty important because the figure you got is in the range of "good, not amazing", unless at the low level (and I wonder what conditions are necessary for that level). Anyway, don't you think those levels are quite a lot lower for something that only lost 18% of battery life? The ranges went down way more than that, and I fully expect DSi to be made on a smaller and therefore more energy efficient process. I doubt this is incorporating anything that uses the higher clock speeds, at least not for the high end figures. So something doesn't add up - unless Nintendo is revising their original claims.
Maybe I'll test on my DS Lite sometime...
Because Nintendo's battery-times lead in front of the competition is so huge that it makes not sense for them to push even further, just for the sake of competing with themselves, I'd guess. Also, how is Nintendo's 840mAh Li-Ion battey (in the DSi) so much cheaper than PSP Go's 930 mAh (given the battery size is comparable), let alone 'the cheapest battery they can get'?
PSP Go was just dumb in a lot of ways, everyone knows this, probably including Sony.. I hope you won't mind me avoiding a comparison.. I'd rather look at the original PSP-1000 at 1800mAh or PSP-2000 at 1200mAh. They also sell higher capacity ones, but I digress. Bear in mind that battery capacity DOES improve per unit cost, just slowly. 1500mAh is pretty standard for small form factor phones.
Anyway, I'm finding this a little uncomfortable because I feel like I'm basically fending off a circular argument that's coming from two different people at the opposite ends and I'm kind of getting crushed. You ask "why improve battery capacity if it doesn't need it", DeadlyNinja asks "why improve performance if it hurts battery life"; obviously both can be done at the expense of price. I'm not saying that Nintendo needs to extend its bottom line, but I am making the point that that's where the decisions are lying.
Your point stands. What I don't understand is why you think that perf/Watt is Nintendo's leading decision factor for their handhelds? Why are you not factoring absolute watt too?
For the performance ranges we're talking about they're the same thing. Unless the device needs to run at < 100MHz or > 1GHz (rough numbers, lots more margin than what we're talking about, don't hold me to them exactly please) it'll get better battery life out of better perf/Watt by clocking the faster architecture at the lower clock.
I'd venture to guess nintendo's combined criteria scale is something along:
- right consumer pricerange (by far the most important factor for such a product)
.
.
- absolute power draw (determining the battery life)
- perf/Watt (determining the processing power of the device, namely the best for the intended battery life)
Yes, price is king. And as far as I'm concerned Nintendo doesn't price aggressively, not in the slightest. I give MS and Sony far more credit for that. They sell hardware at huge margins and they don't drop the price until they really have to. This is on top of their extremely successful software sales. Obviously they'll be defended as doing nothing wrong, or even praised for maximizing their profits; I'm not going to criticize them, instead I'm just going to yearn for more competition to force them into chasing lower margins. It's not that I'm cheap, I just want better hardware.
But I'm not really representative of what Nintendo wants - hardware that's good for the games they want to do - or what their market wants - said games. I'm pretty interested in what the machine can do outside of commercial games, and it's here where the weak CPU is most limiting. Fortunately, Nintendo probably has little intentions of making it very available to common folks for such purposes in the first place, leaving us to inevitably hack around on it like usual.
Aside from the fact I think we should take that 'ARM11' rumor with a grain of salt (people tend to call everything v6 by the 'ARM11' moniker; it could be one of Marvel's v6 hybrids just as well, which, apropos, have SIMD),
I've never, ever seen someone call something an ARM11 for being ARMv6. I've heard about a million people call Scorpion a Cortex-A8 though. I'm sure whoever dropped the rumor, IF credible, also knows better.
I don't think anybody seriously has made any '2x ARM11 @ 266MHz = Gecko @ 485MHz' claims in the context of sheer CPU performance. Unless I missed something earlier.
"The physics and AI should be no worse then GameCube was capable of."
I imagine you don't think physics and AI to be non-CPU problems, at least not for Gamecube or 3DS. Regarding his later response, GameCube did have fixed function T&L and I'm sure games used it. So it's pretty apples to apples.