Lower is better in this case? I understood it as you did, no need to feel bad, man.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Wii U CPU Benchmarks (highly unscientific)
- Thread starter wsippel
- Start date
Lower is better in this case? I understood it as you did, no need to feel bad, man.
First of all, it supports more than those platforms. Also, they managed to speed up that particular effect by a factor of 1.1 to 2.8 on various other platforms. Which platform is 1.1 and which one is 2.8? And what are the baselines on those platforms? Not mentioned, therefore worthless.
CaptNfantasy
Banned
By now its just assumed.
Sure is, but not without a 360 devkit and a Wwise 360 license.
I'd just like to add that one of the beauties of consoles is that they are running on specialised Operating Systems. The OS plays a big role in performance and that's why the PS3 and 360 have reached graphics not achievable with the same hardware running Windows 7 with four times the RAM.
The specs for next-gen consoles don't need to compare to the top-range PCs to deliver comparable performances since their architecture, both on the hardware and software level, gives them an advantage over PCs.
The specs for next-gen consoles don't need to compare to the top-range PCs to deliver comparable performances since their architecture, both on the hardware and software level, gives them an advantage over PCs.
Eh? You can't just compare raw numbers like that, especially for a mobile component and a desktop component.
Hmm... I wonder to which extent we can use those benchmarks to measure WiiU's CPU capabilities.
I mean, just having a quick look at the Release Notes documents by Audiokinetic itself, one can read about WiiU-specific optimizations that appear to give different improvements from a platform to another.
For example, they say something like this:
It sounds (no pun intended!) as if the middleware doesn't behave the same on every supported platform... or, at the very least, that the code optimization worked better on the WiiU CPU compared to other platforms. That says little about relative performance, I think. But, by the way, I'm definitely not 100% sure this is the correct way to read that small (?) bit of information.
I mean, just having a quick look at the Release Notes documents by Audiokinetic itself, one can read about WiiU-specific optimizations that appear to give different improvements from a platform to another.
For example, they say something like this:
Wwise release notes said:
It sounds (no pun intended!) as if the middleware doesn't behave the same on every supported platform... or, at the very least, that the code optimization worked better on the WiiU CPU compared to other platforms. That says little about relative performance, I think. But, by the way, I'm definitely not 100% sure this is the correct way to read that small (?) bit of information.
Its not surprising that these optimisations are giving bigger performance improvements on WiiU than on any other system. They've been working on platforms like 360 and PS3 for years now so most of the optimisations have already been made there. WiiU on the other hand is a new platform so the code is much less mature than the code they're using on PS3/360. Also WiiU itself is still improving (SDK revisions ect). Its extremely likely that future releases will also see much bigger jumps in performance on WiiU than on the likes of 360/PS3.
I agree its not a perfect comparison even if we had 360 numbers due to the less optimised code. But it would be a decent indication of where developers are at right now with WiiU's CPU performance vs the likes of 360.
Moral Panic
Member
Have you seen the dark souls PC thread? Lol
blu
Wants the largest console games publisher to avoid Nintendo's platforms.
Perhaps the SIMD performance might be of more interest here (just speculating, I'm not familiar with wwise's code).
Zoramon089
Banned
More likely that they see smaller improvements because they're old established hardware while the Wii U in this very much in flux in terms of the SDK.
nordique
Member
that's not how it works...not at all
I'll let someone else explain that; but all I will say is CPUs cannot be compared like that.
edit: I saw you said you know that; so it is pointless to compare the Vita's mobile CPU directly with the Wii U's in the first place then
I doubt the Vita has a stronger CPU in it
Graphics Horse
Member
The links seem dead but doesn't the Wii U have a DSP, and if so are they using it like they do with the Vita and Wii?
Yep, my point is (or, well, was meant to be... sorry, I'm no native English speaker), is it "fair" to compare performance data when you have a very well known HW against a platform still in its early lifecycle, taking into account what this translates into in terms of code optimization, and so on?
Lord Error
Insane For Sony
The optimization increase they're quoting is not the important thing there, the benchamrk they ran and shown results of is, and this is what OP quoted and compared to his own CPU.
Yeah like I said I agree it wouldn't be a perfect comparison of the true performance of the CPU's even if we had 360 numbers due to the less optimised WiiU code.
But it would at least be a decent indication of where developers are right now with WiiU's CPU performance vs the likes of 360.
I think his point is that the large increases in performance are indicative of immature code. Until those gains start to reach a more stable level like they have on 360/PS3 (gains from 1.1-2.8x rather than 4.4x) we can't be sure that the WiiU version is at a reasonably optimised stage. Which could make any direct comparison misleading.
Personally I still think its info well worth knowing, as long as the person reading accepts that its only a guideline and that performance is likely to keep improving faster on WiiU than any system its being compared to, but I do see his point.
radioheadrule83
Banned
No offence, but this 'information' is useless, as evidenced by the discussion so far.
Not in the Wwise lib AFAIK. It's a cross platform library so need to work the same on all platforms. I have no idea if they'll do a DSP version, but it'd be nice of course.
Well sort of yeah since we don't have the comparative numbers for 360 or PS3. But if we could get that info it would become very useful, though far from definitive, information.
blu
Wants the largest console games publisher to avoid Nintendo's platforms.
No offence, but gaf threads are hardly an indication of the usefulness of any information ; )
Actually, this is the most concrete piece of performance info we've had on the WiiU yet. It's just too isolated to be of much value to the general populace.
GreggTheGrimReaper
Member
How many Bits? Can it do blast-processing?
Srsly? Last time I worked with ARM cores they were pretty awesome. Cortex A9 (which is Vita's base design) is pretty aggressive, OOOE, branch prediction and all that.
http://www.arm.com/images/A9-Pipeline-hres.jpg
An order of magnitude is 10x. It's impossible to have a gap that large. Anything beyond 2x is a stretch. If the Hollywood@~twice the clocks thing holds true, I'd expect them to be really really close actually.
DaSorcerer7
Member
Power 7
Yeah that was a typo!
You mean Broadway, its not true though. We already know 100% that its not just using overclocked Broadway cores by looking at cache sizes. Could it be descended from Broadway?, possibly but that doesn't just mean Broadway performance x3 + extra Mhz.
Stephen Colbert
Banned
So...
Wii U is less powerful than Ouya confirmed!?
Wii U is less powerful than Ouya confirmed!?
nordique
Member
Wii U is basically less powerful than paint
I thought that might be some kind of silly dragon ball Z character until I googled it 
Can't believe I've never heard of it before, interesting console, obviously quite weak in comparison to even current gen systems though, the GPU is under 10Gflops AFAICS. But I suppose power isn't the point of that system, cheap access to all the free/cheap mobile phone games with a controller on your TV is the point I suppose, and quite a good idea.
Can't believe I've never heard of it before, interesting console, obviously quite weak in comparison to even current gen systems though, the GPU is under 10Gflops AFAICS. But I suppose power isn't the point of that system, cheap access to all the free/cheap mobile phone games with a controller on your TV is the point I suppose, and quite a good idea.Canis lupus
Member
Here's the prototype:
Orange gamecubes? Never knew they existed!
Could just use Geekbench numbers. Not entirely comparable (cause there's other factors like memory bandwidth) but it's probably the most widely used computation benchmark.
Anyway the iPad gets around 750 with its dual 1ghz A9, so ballpark the Vita around 1500 (4x375)? I'm seeing 800mhz and 2ghz around the web for it, but I don't know if the source for the latter is Sony or just sites regurgitating the theoretical max of the A9 design (which I'm not sure any company has actually reached yet). For 800MHz it'd be around 1200 (300/core).
For Gekko/Broadway, the closest thing I can think of is the G3 Macs. On here there's a few 700mhz G3 iBooks that got around 350 with a single core. Of course that doesn't have the extra bits added on for the Nintendo chips, and I remember Macs having pretty slow memory systems for a while, so Gekko/Broadway could potentially be a good chunk faster taking those into account.
And since we know so little about the new CPU it doesn't mean too much. Higher clockspeed plus newer architecture can make a big difference, like a single 1.8ghz G5 is in the 1000 range. Three cores of that and you're in 2ghz Core 2 Duo territory...of course IBM failed and couldn't get power consumption down enough for Apple, but hey that was 7 years ago. Then there's the fact that this is all pretty much peak performance with ideal loads, not everything can multithread nicely to 3-4 cores and vector units or whatever.
Not just orange.
SPICE ORANGE!
blu
Wants the largest console games publisher to avoid Nintendo's platforms.
While I'm also skeptical about a difference as large as an order of magnitude, last time I worked with an A8 it was neck-and-neck with an e300 at half the A8's clock, neon non-withstanding. Just saying. Of course the e300 was also drawing about 2x the power (very loosely estimated, as both parts were sitting in SoCs), but at that price its fpu was managing to make the A8's vfpv3-lite look silly.
I respect the Cortexes for what they are (heck, I fancy them), but let's not pretend they can somehow defy the laws of physics and compete with parts that draw much more power at similar fab nodes.
Cortex A9 is a bit better than A8 in IPC though. Of course it most likely won't beat a similarly clocked Power-based CPU (nevermind a higher clocked one!), but speaking about a factor of a dozen or more is wild exaggeration.
The only thing the discussion so far is in evidence of is that most people fail at interpreting the provided information correctly. It's still better than pretty much everything else we have.
The only bit of info we have about the Vita being clocked at @2Ghz was an interview were it was stated "It's not going to run at 2 GHz because the battery would last five minutes and it would probably set fire to your pants".
So, no, Vita's clock is probably far from being 2Ghz.
So, no, Vita's clock is probably far from being 2Ghz.
It is strange that this was mentioned under a "Wii U:" heading. I also read it as "between 1.1x and 2.8x the wii u figures for other platforms" Not sure if it's referring to how much they've optimized the Peak Limiter (4.4x over previous version) or the 0.06ms mono sound (other platforms are 1.1 to 2.8x more -which is slower in this case).
Link?