This system will be Wii silent.

Realistic expectatition is 320-400 SPUs @ about 500Mhz. It will operate at minimum 5x performance deficit to competition.

Like Sonic said it doesn't sound dumb at all. Plus ERP posed something similar so I don't think he would like to be considered as being dumb. :lol:

Right. To be honest as I think about it, the context of that info seemed to relate more to the visuals the GPU would put out. Which is why I believe I assumed lighting would be a part of how that's address. Li Mu Bai suggested the same thing.

I'm pretty sure it will be faster than the competition, as the competition when it launches is 360/ps3.

If it really should find use as GPGPU my guess would be at least 384/400 shader units, you have to keep in mind that it already needs 1,5x the power of XB360 in worst case because of the second screen. Estimated +100GFLOPS for GPGPU purposes.
So my guess 384/400 - 512/480 shader units @~600 MHz.

Someone on GAF found teh spec of the wiiu power brick, " 230V, 50 Hz, 0.9 A and the output is 15V, 5.0A. Obviously :P"

http://www.neogaf.com/forum/showpost...postcount=4531

Wii used 12V, 3.7A.

75W max total system power draw, including optical drives, wireless radios and so on. That's some seriously weak-ass hardware then! Oh well, not as if it's really any news I guess.

Rated DC output of a power supply isn't necessarily the power consumption under load.

e.g. Wii PSU is rated for 12V@3.7A (44.4W) yet the measured consumption is around 20W.

Of course that's the MAX rated draw. The actual draw will be less than that.

It worse than that. Like other have posted. Its likely 45-50 watts used while running.

Wii for example used less than half rated power.

Xbox360 uses 2/3 rated power

Whats newest 360 brick's rating? (guess I should waddle over and look, lulz, will do in a sec)

And where did this GAF dude manage to hold of a Wii U PSU spec?

The first batch of Slims (with HANA still separate) had a 135W DC output rating. The latest slims have a 115W DC output rating (HANA combined with the southbridge). Measured load consumption is around 90W.

I think some event in EU.

50W doesn't leave much room for much performance once you take out the optical drive. That's what? 30-35W split between the GPU and CPU after memory and accessories and optical drive are considered? It doesn't look like it'd be much more powerful than a current generation console outside of perhaps more RAM.

Personally I'm waiting for a major 'gotcha' like the Wii U OS and tablet using up 512MB of RAM which leaves only 1GB for games.

Oh, I think that's pretty much a given. Hell the OS alone may take 512MB. Too be fair, maybe it can be reduced later.

Yeah, that's what I meant by "max" power. Sorry if I was being unclear (I guess I was in shock from the news :)), I just wanted to point out that there wasn't much room to move around in with such a constraint on power.

I doubt that drive's going to consume very much. Likely, it's a 2x BR drive like in PS3 (for cost reasons), so the disc rotation speed should be quite low. As the spindle motor is normally the biggest consumer, that leaves the electronics, but BR hardware should be quite mature by now and I doubt the chipset will gulp down very much either...

The console has up to 75W at its disposal.
The fact that no console (or any electronic applicance that I know of) constantly uses all the power they have available, it doesn't mean the sum of the TDP from each component can't reach a value close to that.

So yes, typical consumption of the Wii U may be around 50W, but the peak power consumption for the GPU+CPU could be well up to 60W.




Why? You don't want the console to succeed?

If it did reach 75 Watts, it would be a badly designed piece of hardware. You usually have a security margin of 1.5 to 2.0. And since consoles usually run "optimized" software, I can't see the peak being very much different from the average case, either. Some noise is there, sure, but most of the time, it won't be.

Also keeping in mind that the PSU most likely isn't 100% efficient, either.

Console do run close to max consumption. Of course they will not be using 75w. The power supply couldnt handle that first off and it wouldnt be reliable at anywhere near what you are talking about....


60watts would push it to 80%. That is the very upper limit and I dont see it even that high. I know when building a pc you try to keep in with 40-70%, that is the rule i have always seen.

And where is this "security margin of 1.5 to 2.0" coming from?


PSU efficiency has nothing to do with rated output.


What do you mean "it can't handle"? It can handle up to 75W because its internal components are rated as such.
It's obviously not prepared to do an "average" of 75W because that would mean the peak values are going above that number, but the peak is 75W and not any other number.

I see, so you're comparing it to bad PC PSUs.

But no, 80% isn't the "upper limit". Even in PCs, only "bad" PSUs won't be able to handle the rated wattage.

Some PSUs (Corsair, Seasonic, BeQuiet) can even handle sustained wattages well above their rated output, and all the good ones will handle some 5W above its rated specification just fine.


Yes, many uninformed people use the "up to 70%" rule for buying PSUs, without even knowing why.. That's mainly because there's practically no quality regulation for PSUs and their rated maximum output and they'd risk buying a "800W PSU" from "LCPower" or something that won't even do 600W and damage some components due to unstable voltages.

So said my prof in constructional engineering as well as electrical engineering and "experience"... Say, look at PS3, with its 380 Watt PSU, but only using about 200 Watts under full load.

Your typical slot load BD-rom drive eats about 5w of power.
Your typical configuration of Power 476fp takes up 1.6w per core.

Here's what I posted on GAF:
Slot load BD rom: 5w (1A at 5v)
Flash memory: 0.25w
Wifi modules: approx 2w per chip
2GB DDR3: approx 5-10w, depending on configuration

The "Security Margin" you speak of is applied in the components themselves (capacitors, rectifiers, resistors, etc.). It's not in an end-product like a console PSU.
A well-built PSU will output its rated wattage perfectly well for as long as you want, because its less-capable component is able to drive that voltage/current with a security margin of 1.5/2.0.

I'm pretty sure your prof will say the same.

That's the phat model, right?
Why won't you check the power usage in an AAA game from 2012 and see if it's still 200W?

Probably quite a bit less than that.

Are you really going to argue this? It will lead to nothing but dissapointment. All CE devices run closer to 50% than 100% on maximum usage

Funny.
So when you turn on your 600W microwave at maximum power, you think the output is actually 300W?

ToTTenTranz has issued an easily verifiable challenge. Someone measure the power draw running a known AAA title (Uncharted 2 or 3 should be accepted as taxing the system) and it can be proven one way or the other, and someone fed crow.

I'm more confident on the microwave test, lol.

http://www.anandtech.com/show/3774/w...box-360-slim/3

The slim used around 90W to run games on a 135W supply. (And that's 90W on the outlet side. The brick is at best 94% efficient, but that's highly doubtful, so it's probably supplying a good deal less than 90W to the console)

Those power supplies have to be rated assuming you've got every single peripheral under the sun plugged in. If you're playing a demanding Kinect game (is there one?) while charging multiple controllers over USB, using wireless on a weak signal with the antennae at maximum strength downloading a game at full-rate to an HD in the background, or whatever other power consuming tasks you can think of. Even for that absolute worst case, there has to be some headroom (10% at the minimum, just to make sure you don't exceed the brick).

There are also the power spikes from spinning up the optical drive & random accesses. The rate at which the drive reaches nominal speed will have an effect, of course (usually modest for notebook solutions to keep the power down).

Dunno, I switched classes a while ago and graduated in a different course...

Though... but using your logic, if all parts have this margin, then the rated output should effectively mirror this margin too, no?

Also... running your PSU at full load for a longer period of time WILL lead to it dying pretty fast. Just look at PC PSUs. There's a very good reason why 150 Watt GPUs recommend 24A at the 12V rail (i.e. 288 Watt). I know I know, PCs are a different case, but still basic "engineering" logic should still apply.

Dunno, I sold mine... but I highly doubt that current games use more power than the tested ones... like Final Fantasy XIII and GTA4. Current games surely aren't "furmark" to our systems, compared to the older ones... at least not by a larger than negligible difference.

Yes, I imagine the optical drives can eat a good amount of power as they spin up.

Assuming the Xbox 360 S transformer is able to output 135W, it would have to draw 145W if you assumed 94% efficiency, which is probably high. It's probably between 80 and 90%. In any case, your 360S is not going to draw 145W from the wall, at any point, period.

Like Al said, the Wii uses about 20W during gameplay, and the output on the supply is rated at 40W or something like that.

Edit:
Another thing to consider is that they're not designing these supplies. They're just rebranded off-the-shelf supplies. They're likely to look at quite a few supplies with different ratings depending on what's available. They may end up with one that has a ton of headroom, if it is abundant, cheap and of good quality.

Side-note: 15V DC seems a bit unusual. :p

NO one designing a mass market computer produce product would do anything you say. I don't even you believe this non sense. Do you believe they will use more than 60watts?

We are all just uninformed. Even if every other console ever made follows these rules. Lol I mean really if you needed more than 50 w get a bigger power brick. Its not built into the system just get a bigger one....

While in theory you made be correct that they could do this. It is so unlikely it not worth talking about.

as the other has posted. This is not true at all.

Xbox 360
Slim System .06W
Off Idle 70.4W
Halo 3 87.0W
Rockband 2 82.7W
Gears of War 2 88.0W
Red Dead Redemption 90.4W



so this microwave will be using 1100 watts to make 600 watts. Not sure how is backs up what you posted.

It's been shown that current consoles use almost as much power at idle as they do at full load, and therefore the power draw in a "single-A" game should therefore be pretty much the same as a triple-A game...

This is immaterial. Hitachi Deskstars at 7.5k RPM using five metal platters and a phat metal hub spindle spin up to full speed using only 15W, when standby power is in the 8-10W range. A single, low-RPM carbonate disc isn't going to need a whole lot to spin up. The stepper motor controlling the optical block way, way, WAY less. Seeking is likely in the milliwatt range, IE rounding error pretty much.

No? There are some spec lists for BD drives on google listing up to 22W for said cases while nominal operation was much lower.

http://www.pioneer.eu/eur/products/a...202/index.html
http://www.pioneer.eu/eur/products/a...3BK/index.html

*shrug* Guess I misinterpreted then. :p

Sure, there could be BR drives using a crapton of power to spin up, but it doesn't HAVE to be that way. A few years ago, Seagate drives regularly guzzled upwards of 50 (!) watts during spinup for example, while at the same time Hitachi drives were much more frugal (as mentioned in previous posts.)

So it would be a design issue, not anything inherent to getting a light platter spinning.

Ah. hm... so what makes them different :?:

4cm case fan. It was always going to be this way!

Think the WiiU has 4 USB 2 ports, right? So that's 10W allocation iirc (or maybe more if they want fast charge over USB). And yeah, not just Wifi but video streaming to two pads.

Running the 4 cm fan at 12v couldn't keep the system cool enough!

Just a note: the number of pads that video is being streamed to is irrelevant - apparently it's all multiplexed in the same stream. We can conclude as much based on the '1x pad @ 60 fps, 2x pads @ 30 fps' info we got some time ago.

Yeah, I'd worked out that there was some limit on the WiiU meaning 1 x 60 went to 2 x 30. I didn't know what it was, but using the same wireless channel seems entirely reasonable and I guess it's cheaper in terms of hardware.

Yes, but probably only incidentally, as wireless video is probably not trivial to get working reliably, and having two parallel streams would be (at least) twice the engineering headache, risk of interference etc. Plus, as mentioned, power use as well... Damn that tiny little metal tin can! :lol:

I was thinking a bit more about the CPU setup, trying to reconcile the initial announcements about the Power7 relation, the asymmetric L2, the Wii HW BC, and the supposed ease of 360 portability.

How about a Power7 main core derivitive + 2 "enhanced" Broadways? :p i.e 4-way SMT + 2 cores -> 6 threads *ahem*

Sounds ridiculous, I know (45nm Broadway would be lol). >_>

Don't worry. You're not the only crazy person around here.

http://www.neogaf.com/forum/showpost...ostcount=12624

heh, that's what I get for not following thousands of posts. :p

I can't believe you missed that one post out of 100,000.

I need to go back making games, I'm bored of even considering to try to guess, I want to know :p

Could the e6760 be the WiiU GPU?

http://www.em.avnet.com/en-us/design...rocessors.aspx
http://www.amd.com/us/Documents/E676...duct-Brief.pdf

480 shaders
3DMark™ VantageP score of P5870
DirectX® 11, Shader Model 5.0, OpenGL 4.1
35W
40nm

Still seems a bit to high on the power draw for the WiiU.

Well if it has that gpu and 2 gb ram I'm buying it:)..believe it when I see it mind..

35W is based on GDDR5 memory. Is it possible that GDDR3 would bring it down?

I considered the option earlier (based on the difference in the amount of L2) but discard it.
I remember discussing here with 3dilletante which knows a lot more than I do on those matters about the odds about having for example a power7 and power a2 cores in the same chip a bit like the small/big ARM a15 and a7 philosophy.
He told me that actually the powe7 and power a2 may have slight difference in implementation of the ISA that could hinder the implementation. The A15 and A7 were developed hand in hand.

Looking at Nintendo I think it's unlikely, putting together a CPU is complex and costly enough, 2...
The whole thing sound like quiet an expanse.

Honestly I still believe that the best bang for buck would have been a redesigned Cell (like a cut power7 + 3/4 SPUs). Either way using low power cores and a SMP set up they could have gone with more cores (like 6/8).

To me the best system with their high constrains (cost and power) would have been sonething like that:
Cell 2 Power7 + 3/4 SPu, would end barely above 100 sq.mm on IBM 45nm process.
A ~redwood (so 4/5 SIMD) including 16MB of scratch pad 9edram) memory on the same process (around the size of MS vahalla chip?).
the gpu is the north bridge, 1 or 2 gb of ddr3.

We are going to find out soon how much silicon Nintendo invested on WiiU, that's gonna prove interesting.

no way in this world. Goes against every piece of info we have on wiiu. Just the kids in gaf wishful thinking. Same way the gpgpu rumor started and just about everything else. It funny watching happen on gaf. :lol:

What is this "every piece of info"? The rumor mongering which was going towards soon suggesting it's in fact weaker than original Wii? :roll:
(of course it won't be off-the-shelf part, but other than that those shader counts etc could be close to truth easily)

Can you elaborate?

It's not exactly an earth-shattering set of specifications. 8 ROPs and 24 TMUs is pretty tame.

Just the fact that we have the dev specs and none match this gpu family. Matches exactly r700 series core. We had many dev and insiders confirm the leaks.

Bg also even stated the wiiu is still r700 based today.

" Idon't see it as a base. Just what Wii U's GPU would have similarities with. Istill believe it will have R700's base architecture. Beyond that it could resemble the E6760."

Here is the back story behind this rumor.
http://www.neogaf.com/forum/showpost...postcount=6275

What do you think base off the info we know?

I'm just looking at the raw specs* and not the GPU feature capabilities with respect to DX, so pretty much what bg just said. So sure, the WiiU GPU might not be DX11 compliant (so that'd apply to everything past the rv7xx generation), but then why did you even bother listing the specs? It's the rest of the hardware you're looking at, no?

*480/24/8 (ALU/TMU/ROP) is the most likely configuration for the E6750 given the existing 6570/desktop (650MHz) or 6750M/mobile (600MHz) products.

If you just want to stick purely to existing DX10.1 products, there are parts configured as 320/32/8 as well (though you wouldn't be able to tell too much about power draw for WiiU speculation since they're 55nm parts).

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

So basically... ballpark speculation that we've had before (and even since the original article here with 400/20/8 - Redwood). *shrug* These aren't really outrageous throughputs we're looking at here.

Yes a Redwood PRO underclock would work. I was more speaking about the feature set that went along with the rumors.

From a raw spec or performance that rumor would be the high end of the scale right now. I do agree with the point you made in your post.

Too bad we are going to have to wait until these things to go on sale to find out really anything. I love to get the power consumption numbers.

Yes if you just took the shader count part of the rumor. From a raw performance this would be the high end of the scale.

I think that is just silly to say its weaker than the wii. From just speaking of the gpu side it is unlikely weaker than the ps360. I dont even think that was up for debate here.

Nah. That started with us on GAF. A poster named Fourth Storm brought it to my attention through a PM and then I posted it a few months ago stating how I could see Wii U's GPU having similarities to it. Every now and then I touched back on it after seeing something related to passed things we discussed there. Apparently a site consolidated the info we (I) have been posting and made their own article or whatever.

Your "best bang for the buck" idea is similar to what my thinking was in that the modified POWER7 core would be the "PPE" and the two 476FPs would be the "SPEs". Just OoO.

Too be fair, there have plenty of points on GAF where the rumor mongering was basically leading to the point it was unquestionably more powerful than what they were speculating for Durango (Eg, Durango= 480 SP 6670, while Wii U=800 SP 4850+)

It actually fits very nicely with all that is known since its an VLIW5 chip.

And whatever chip goes into WiiU it has GPGPU support 100% since r700 line was able to do that kind of tasks allready.

Btw, not saying this is what WiiU will use as a GPU but "No way in this world?" or "Goes against every piece of info que have on WiiU" are both incorrect.

"Was able to do" is not the same as "actually being good at it". Which wasn't the case for the 4xxx series Radeons. We've been over this already.

Well from the reading between the lines that I have so far, that the gpu is out balancing the cpu... and that improving difficult.

So I have no idea how that is possible...how the hell can you make a cpu in late 2012 that is weak?

An quad OoO power 7 running at 2.5gHz would be more than enough for a decent cheap console...and the expected 45xx gpu wouldn't out balance that would it? 2gb ram with 50gb/s would also hit the sweet spot very nicely.

If they serve something like that which is entirely reasonable then I will take the bait.

More like "not being as good as GPUs really made to use GPGPU tasks" OpenCL can do some tasks better in r4000 series than CPUs so the point remains. Besides Im not talking about performance just noting that the above comment regarding GPGPU in WiiU more or less as nonsense its wrong.

And to clarify myself further Im not saying GPGPU is a key to WiiU development or anything, just that the chip itself its capable of it, nothing more nothing less.

How is a quad core power 7 in any way reasonable?

It can't be. The power requirements would be far too high...

http://www.youtube.com/watch?v=I9Ajm_6Vz8M&t=0m26s

Are these rumors new?
Courtesy Wiiudaily.
__________________________________________________ ___________________
The Wii U GPU is made by AMD and is based on a modern Radeon HD design. Nintendo has been working with AMD for two console generations now, and this will be the third time AMD/ATI supplies the graphics processor in a Nintendo console. The Wii U CPU will be made and supplied by IBM. According to reports, the Wii U GPU is based on the Radeon HD 5000 series GPU, which was introduced to the PC gaming market in 2009. However, while the based architecture will be the same, the Wii U GPU will be custom made with additional features. For more on the console hardware, check out the complete Wii U system specs.
Wii U GPU specs

The Wii U GPU specs have yet to be announced, but based on developer comments and on the already available design sheets from AMD, we can estimate the following features for the graphics chip:

Based on Radeon HD 5000 “Evergreen” series
Built on 40nm manufacturing technology
Unified shader architecture
GDDR5 memory support (memory is likely to be shared with CPU and system)
Shader count: 400 unified shaders (rumored)
75 GB/S bandwidth
Low power design, 50W TDP
Full 1080p, 60 frames per second support

The Wii U GPU has been in development since 2010 according to sources, and it’s likely that AMD created a highly specialized chip for Nintendo, with many modern features incorporated into the 3 year old Evergreen design.
Wii U GPU power and performance

While the above specs don’t sound much compared to the current generation graphics chips from AMD, where up to 2000 unified shaders are possible, the specs are feasible for a console. Due to pricing, power, and resource issues, the GPU is estimated to have 400 unified shaders with extra features added. This would still make the Wii U GPU many times faster than what’s found in the Xbox 360, which only has 48 shader units. It has been rumored that the Wii U GPU includes a small amount of RAM (32 MB) embedded with the graphics processor and aside from the main system memory, although this report hasn’t been confirmed. This RAM is supposedly used as a framebuffer and to assist with some of the other GPU features.
__________________________________________________ _________________

The Wii U CPU is based on IMB’s latest PowerPC technology, the POWER7 architecture. The processor will use most of the modern features from POWER7, including multi-threading and power saving controls. The Wii U CPU speed is estimated to be at 3 GHz by developers who have access to dev kits. The processor is the key part in the Wii U system specs, which include a Wii U GPU from AMD, and an unspecified amount of RAM. According to reports obtained by Wii U Daily, the console will use a processor based on the “IBM 710 Express”, but will be heavily modified to suit the needs of the console and video game software.
Wii U CPU specs

The Wii U CPU uses most of the features from the POWER7 architecture, the rumored and leaked specs so far indicate:

4 CPU Cores and 2 MB shared L3 cache
3 GHz clock speed
Multi-threading: 4 threads per core, 16 threads in total
45 nm process
Advanced power savings features and design
256 KB L2 cache per core

The processor of course includes a whole slew of additional features, but the main attribute is the addition of multi-threading: the Wii U CPU can handle up to 16 threads at the same time. Compared to the older Xbox 360 chip, which can only run 6 threads at a time, the new chip is much more capable when it comes to multi-threading and simultaneous executions of code. Production of the Wii U processor will take place at IBM’s facility in New York, USA, while the final console and components will be assembled in China and shipped worldwide.
Wii U CPU power and performance

The power output of the Wii U CPU is debatable at this point, as no clear technical specifications are available. The IBM Power7 architecture is rated at maximum 33 GFLOPS(Giga FLOPS) per core with the full amount of cache available at max clock speed. Per CPU, the maximum performance output is over 260 GFLOPS, albeit this is based on a fully specced out 8 core chip at 4 GHz. Realistically, the Wii U CPU power output could be at around 50 GFLOPS, with 4 cores at 3 GHz. This would make the Wii U CPU over 20 times faster than the old Wii Broadway processor, and twice as fast as the PlayStation 3 Cell processor, and 2.5X as fast as the Xbox 360 chip.

Those rumors don't sync with the the 75 Watt power brick rumor.

With the GPU being 50 Watts that leaves less than 25 watts for everything else in the system. A 16-thread 3Ghz 45nm CPU using less than 10 Watts? And that's with assuming Nintendo would be crazy enough to draw 100% power.

They definitely don't fit one another.

It's from WiiUDaily. That's all you need to know. But just in case don't waste your time with their site.

Too bad though it sounded "right" too me. I don't mean I read the stuffs as it was true but the choices and the system they describe make sense to me.

Well I guess that in your view the system is heterogeneous, so the code running on the main core can't always run on the other ones which is not the case in the a7 a15 where it's just a matter of performance.
Looking at the throughput of the 476 I'm not sure it's worse it. Moving away for SMP for the sake of 2 476 cores doesn't sound like a good compromise to me.

The thing with SPU is that at least you get quiet some rewards from moving away from SMP, not too mention that SPU are tinier than a PPC476 + its L2. They may have more than 2 for the same "price".
Then there is the (sustained) throughput and even clocked lower than in the ps3 they are not in the same ballpark.

If I look at an engine and a game like BF3 and the matching FB 2 engine, I'm convince that a lesser cell would do the job (especially once you remove lot of the graphical tasks the GPU would have handled) for now and even for quiet some years coming (assuming devs have intensive to continue to develop for the arch ie the hypothetical WiiU sales well).

Anyway we are sure Nintendo didn't take that road but to me it was the best one if they wanted to be "in between" or remain at the tail (viable) end of PC gaming for a few years.

Well I can tell you now if those specs were anywhere near real then we would be looking at games much much better than what they hag shown..I don't care about early software...we are talking 2-4 times ps360.

Oh and it make s an obvious mistake...xenos has 240alu against 400 in the supposed wii u gpu...not 48..

If that were really the specs then I would snap that console up when it comes out...certainly devs and iwata would be saying it blows current gen omit of the water...they don't.

Yes there are obvious mistakes in that so called report, indeed Xenos is 48 "vec4 +1 scalar unit" so 240 alus.
As said by others 4 power7 cores at 3GHz migh consume more than the whole power budget for the wiiU.

It sounded what a low-spec, modern console should offer, except a bit too many threads on the CPU and not enough L2. The 360 has 1MB for 6 threads, which is not really enough, Wuu would have 2MB for 16 threads? No I don't think so, that's even worse cache/thread ratio...

Not to mention that the power7 may need 1 MB of L3 per core. Actually with 256 KB of L2 512KB of L3 per core doesn't make sense.

I have thought of a scenario that no one has put forward yet....what if it is a 4 thread....but single power 7 with 4x smt @ 3.2ghz??

In that scenario the complaints about wuu having less cpu power/threads than 360 and having to use gpgpu would make sense...however I could see that cpu still being much more powerfull than xenon.

Still think everything we have heard so far is complete rubbish...
I'm going for a single core 4x smt power 7 with 4mb. L2...maybe 32mb edram shared between gpu and cpu....a hd 4570 class gpu...custimised and 1-2 gb ram.

That set up mentioned would yield the early game demoes seen and hit tdp/cost targets @ 45nm

Well dimishing the number of cores has a non linear effect on power consumption.
basically 1 power7 core doesn't consume 4 times less than a quad core.

Your prediction is more than unlikely in my opinion.