• 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.

WiiU technical discussion (serious discussions welcome)

Thraktor

Member
I have nothing more to add than either of these

http://www.youtube.com/watch?v=m_mDTLphIVY
http://www.youtube.com/watch?v=Hj5vERbLtYI&t=49s

Please advise. Over.

Broadway outperforms Bobcat by 35.7% per clock while running SIMD-heavy code, which would generally be considered the chip's weak point. Furthermore, use of paired-singles by Broadway improved performance by 167.8% over simple scalars.

Edit: Interestingly, Bobcat performs 96.3% better in the same test when utilising its SIMD capabilities compared to operating on scalars. This means that Broadway's paired-singles is actually more beneficial than Bobcat's full-fat SIMD instruction set, in this case at least.
 

ozfunghi

Member
Broadway outperforms Bobcat by 35.7% per clock while running SIMD-heavy code, which would generally be considered the chip's weak point. Furthermore, use of paired-singles by Broadway improved performance by 167.8%.

TY. I assume that's per core, per clock?
 
Broadway outperforms Bobcat by 35.7% per clock while running SIMD-heavy code, which would generally be considered the chip's weak point. Furthermore, use of paired-singles by Broadway improved performance by 167.8% over simple scalars.

Edit: Interestingly, Bobcat performs 96.3% better in the same test when utilising its SIMD capabilities compared to operating on scalars. This means that Broadway's paired-singles is actually more beneficial than Bobcat's full-fat SIMD instruction set, in this case at least.
Wow, Broadway is not nearly as weak as some people made it out to be. I'm curious on what we will find out about Expresso when we get the die pictures. One reason why people are making theories of Latte having 160sp is due to them considering the CPU to be weak and unbalanced with a stronger GPU.
 

ozfunghi

Member
So, this is actually Wii's CPU, not WiiU's CPU... any idea what the performance gain might be? Not much i take it, except for the larger cache? Still, that would mean 1 U-CPU core at 1.25 GHz isn't necesarilly underpowered compared to 1 Bobcat core at 1.6 GHz. And if Durango is only using 6 cores, that would mean the gap is only 2x roughly? But how representative is this specific test?
 
So, this is actually Wii's CPU, not WiiU's CPU... any idea what the performance gain might be? Not much i take it, except for the larger cache? Still, that would mean 1 U-CPU core at 1.25 GHz isn't necesarilly underpowered compared to 1 Bobcat core at 1.6 GHz. And if Durango is only using 6 cores, that would mean the gap is only 2x roughly? But how representative is this specific test?

Nah there's more to Xbox 3's CPU than just plain Jaguar cores apparently.
 
So, this is actually Wii's CPU, not WiiU's CPU... any idea what the performance gain might be? Not much i take it, except for the larger cache? Still, that would mean 1 U-CPU core at 1.25 GHz isn't necesarilly underpowered compared to 1 Bobcat core at 1.6 GHz. And if Durango is only using 6 cores, that would mean the gap is only 2x roughly? But how representative is this specific test?

U-CPU's cores are clocked ~70% higher than Broadway, has 2x cache for two of the cores, and 8x(!!) the cache in the other one. Perhaps the switch of L2 cache to eDRAM would give it performance boost too. It is probably ~6-7x Broadway.

How much more capable are Jaguar cores compared to Bobcat's?
 

USC-fan

Banned
U-CPU's cores are clocked ~70% higher than Broadway, has 2x cache for two of the cores, and 8x(!!) the cache in the other one. Perhaps the switch of L2 cache to eDRAM would give it performance boost too. It is probably ~6-7x Broadway.

How much more capable are Jaguar cores compared to Bobcat's?

On that test about 2.3x... ;)

The biggest difference really is the HSA support on jaguar. I think that is one of the main reason they both went with that chip.
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
If I'm not mistaken, doesn't Bobcat do 128-bit SIMD through two passes of a 64-bit ALU? Not to take anything away from your results, but worth remembering in comparison to Jaguar, which has a proper 128-bit SIMD unit.
You are not mistaken. Effectively both CPUs do 2-way SIMD, though one of them pretends it's 4-way. Jaguar should be 4-way proper.

Broadway outperforms Bobcat by 35.7% per clock while running SIMD-heavy code, which would generally be considered the chip's weak point. Furthermore, use of paired-singles by Broadway improved performance by 167.8% over simple scalars.

Edit: Interestingly, Bobcat performs 96.3% better in the same test when utilising its SIMD capabilities compared to operating on scalars. This means that Broadway's paired-singles is actually more beneficial than Bobcat's full-fat SIMD instruction set, in this case at least.
Part of the high gain for Broadway comes from the overall better code produced by the newer compiler. I can run the test sans paired-singles, so we can try to figure out what comes from the paired singles alone.

On that test about 2.3x... ;)
Have you actually run it?
 

Van Owen

Banned
So now that Aliens is getting panned and it was revealed Gearbox didn't even work on the main game for the main SKUs, can we finally admit Pitchford was spouting PR bs?

(Like I said all along)
 
Broadway outperforms Bobcat by 35.7% per clock while running SIMD-heavy code, which would generally be considered the chip's weak point. Furthermore, use of paired-singles by Broadway improved performance by 167.8% over simple scalars.

Edit: Interestingly, Bobcat performs 96.3% better in the same test when utilising its SIMD capabilities compared to operating on scalars. This means that Broadway's paired-singles is actually more beneficial than Bobcat's full-fat SIMD instruction set, in this case at least.

Do we know how much of a difference the edram as cpu cache makes in performance? Or a good estimation at least?
 

tkscz

Member
So now that Aliens is getting panned and it was revealed Gearbox didn't even work on the main game for the main SKUs, can we finally admit Pitchford was spouting PR bs?

(Like I said all along)

Sure, but answer me this, why are you so determined to make the WiiU seem weaker than current generation? Same goes to USC-fan.
 

ozfunghi

Member
So now that Aliens is getting panned and it was revealed Gearbox didn't even work on the main game for the main SKUs, can we finally admit Pitchford was spouting PR bs?

(Like I said all along)

No. Just to spite you.

Besides, What did he say exactly? That WiiU had a great CPU? Well, Blu just proved him right, lol
 

KingSnake

The Birthday Skeleton
So now that Aliens is getting panned and it was revealed Gearbox didn't even work on the main game for the main SKUs, can we finally admit Pitchford was spouting PR bs?

(Like I said all along)

Not that I care too much for this game, but have we seen the Wii U version yet?
 

USC-fan

Banned
Sure, but answer me this, why are you so determined to make the WiiU seem weaker than current generation? Same goes to USC-fan.

Who said i was? I been getting flak from the fanboy since way before e3. Crazy thing I even way over shot the performance of the wiiu. I was called all kind of name so it nothing new.

Facts speak for them selves in the end.
 

deviljho

Member
What does the Aliens being critically panned have to do with system power?

I don't follow Van Owens logic

it's just an unreliable comment randy pitchford almost a year ago where he made some vague claim about the wii U version of Aliens: CM being graphically superior due to "newer hardware" - people have been clinging to it for various reasons.

Facts speak for them selves in the end.

oh...
 
it's just an unreliable comment randy pitchford almost a year ago where he made some vague claim about the wii U version of Aliens: CM being graphically superior due to "newer hardware" - people have been clinging to it for various reasons.


So are you saying it will look better due to competent developers?
 
So are you saying it will look better due to competent developers?

Given the way the 360/PS3/PC versions look, the messy development history and the news (well, to me anyway) that this wasn't really a Gearbox project, but another dev doing the bulk of the work - along with the disappearance/delay/whatever of the Wii U SKU - it doesn't look like anyone should be expecting anything much out of this game anymore.
 

Effect

Member
His comment was unreliable and PR nonsense from the start. The power of the Wii U shouldn't even be apart of the issue I feel.. He never backed up his claims with anything. No one should have taken what he said seriously without any proof at all.
 

wsippel

Banned
So now that Aliens is getting panned and it was revealed Gearbox didn't even work on the main game for the main SKUs, can we finally admit Pitchford was spouting PR bs?

(Like I said all along)
I'm not sure how that particular game being shit or TimeGate doing most of the development work has any relevancy to the topic at hand?


We know the GPU is weaker than the standard desktop variants of what its based on
After looking at the die, we don't even know what it's based on.
 

Schnozberry

Member
Perhaps it will take Nintendo's reveals at E3 to really find out what this thing can do. The only thing that seems certain is that we don't know what to expect out of it performance wise.
 

Vagabundo

Member
So now that Aliens is getting panned and it was revealed Gearbox didn't even work on the main game for the main SKUs, can we finally admit Pitchford was spouting PR bs?

(Like I said all along)

This isn't even true. While TimeGate think they did about 50% of the final SP game (plus some additional content) they haven't even played the finished product, so they don't know what content is in there.

Gearbox made most of this game.
 

deviljho

Member
This isn't even true. While TimeGate think they did about 50% of the final SP game (plus some additional content) they haven't even played the finished product, so they don't know what content is in there.

Gearbox made most of this game.

None of this is even relevant. This studio did this. The other did that. Demiurge is actually doing the Wii U version. Randy makes some comment about a game that's on the back of his mind for PR (cause he's focusing on Borderlands 2 at the time) before the final Wii U devkits are even out, and the Wii U version is being outsourced. And then you don't hear from him again about it. Oh wow, a motion tracker on the gamepad... neato.

This is 100% off-topic.
 
Who said i was? I been getting flak from the fanboy since way before e3. Crazy thing I even way over shot the performance of the wiiu. I was called all kind of name so it nothing new.

Facts speak for them selves in the end.

I'm not. I think it's the same mostly.

I believe the issue is that you two spend way more time "poo-pooing" on Wii U than you do talking about the other consoles.

For example in the two current Wii U threads, Van Owen you have 14 posts while USC has 68. In four or five of the most recent PS4/Xbox3 threads, you two have a combined five posts that USC made in only two of them.
 

Van Owen

Banned
I believe the issue is that you two spend way more time "poo-pooing" on Wii U than you do talking about the other consoles.

For example in the two current Wii U threads, Van Owen you have 14 posts while USC has 68. In four or five of the most recent PS4/Xbox3 threads, you two have a combined five posts that USC made in only two of them.

Not a bad ratio for me.

I wouldn't say I'm poo pooing Wii U, just being realistic about its capabilities, as I've always been while others think is a noticeable leap over current gen. Alrhought the current lineup does blow until Pikmin and 101 and its troubling that its struggling to get current gen ports.

Can't really hate on something like Orbis with what we know now. It seems to be very capable.
 

nordique

Member
We know the GPU is weaker than the standard desktop variants of what its based on

No, we don't. We don't even know what its based off of yet. In fact, there seems to be less certainty about it than before, plus more mystery about it.


Who said i was? I been getting flak from the fanboy since way before e3. Crazy thing I even way over shot the performance of the wiiu. I was called all kind of name so it nothing new.

Facts speak for them selves in the end.

No no no, you were supposed to respond to this:

Have you actually run it?

since its actually relevant to both the relevant discussion and your claims
 
Not a bad ratio for me.

I wouldn't say I'm poo pooing Wii U, just being realistic about its capabilities, as I've always been while others think is a noticeable leap over current gen. Alrhought the current lineup does blow until Pikmin and 101 and its troubling that its struggling to get current gen ports.

Can't really hate on something like Orbis with what we know now. It seems to be very capable.

At this point "realistic" is subjective just like saying there will be a noticeable leap is subjective. But with these being technical threads, only saying "it's on par with PS360" and not giving a technical basis behind it is not really technical. No different than someone on the opposite end and giving no reasoning behind why it would a noticeable leap.

And the point wasn't only about hating on consoles, but no discussion at all about the others. Why is Wii U tech discussion and being "real" so important, but not for the other two? And before you take that the wrong way, I'm not saying you or USC shouldn't post in these threads. Just why is Wii U the focus and not the other two. Because the ratio is rather absurd to me.
 

Van Owen

Banned
No, we don't. We don't even know what its based off of yet. In fact, there seems to be less certainty about it than before, plus more mystery about it.

See, this is kind of ridiculous. We can be certain it's not more powerful than a desktop equivilant, and there's not some secret Nintendo magic inside that is going to suddenly increase visual quality in games to any discernible level.
 

Goodlife

Member
See, this is kind of ridiculous. We can be certain it's not more powerful than a desktop equivilant, and there's not some secret Nintendo magic inside that is going to suddenly increase visual quality in games to any discernible level.

But we haven't found out what the desktop equivalent is yet.....
 

Goodlife

Member
Maybe there isn't one.

Who cares. Either way, it's not a big enough unknown that we can't estimate what Wii U is capable of overall.

Well there seems to still be a fair amount of debate about it, even after the die shots, so I'd say it still is a big enough unknown.
 

tipoo

Banned
So, this is actually Wii's CPU, not WiiU's CPU... any idea what the performance gain might be? Not much i take it, except for the larger cache? Still, that would mean 1 U-CPU core at 1.25 GHz isn't necesarilly underpowered compared to 1 Bobcat core at 1.6 GHz. And if Durango is only using 6 cores, that would mean the gap is only 2x roughly? But how representative is this specific test?

Durango would be using Jaguar, not Bobcat, yes it's derivative but Jaguar has a lot of FPU enhancements. The FP unit gets 128-bit processing and a 128-bit wide data path, and 256-bit AVX processing. The boost to FP performance should be substantial, it will be more like the big chips (Core/FX) than a low power core in that regard.
 

nordique

Member
See, this is kind of ridiculous. We can be certain it's not more powerful than a desktop equivilant, and there's not some secret Nintendo magic inside that is going to suddenly increase visual quality in games to any discernible level.

No one said anything about "secret" Nintendo magic.

If anything, that statement just implies some prejudice (what are you trying to prove exactly?)

It may very well be weaker than X desktop card, but superior in some other way. Its so customized that its seemingly really hard to discern anything about it.


The people who are following the GPU discussion don't care about its performance as much as figuring out what exactly the GPU layout is and what its capable of...

I don't have anything against you, Van Owen, but it seems everytime you come in to post its either attention seeking or it leads to derailing a (Wii U) thread. There isn't anything constructive you contribute that I can pin-point, no offence.

Wii U has become an huge mystery, and some people are just interested in finding out what exactly its capable of; if you aren't willing to help out with that with your technical know-how in a constructive fashion, then what exactly are you bringing to these threads?
 

Van Owen

Banned
The people developing games for it don't think it's a huge mystery and think its equivilant to current gen. Why would anyone ekes think otherwise?
 

wsippel

Banned
The people developing games for it don't think it's a huge mystery and think its equivilant to current gen. Why would anyone ekes think otherwise?
Some do, others disagree. Some (like Omega Force and 4A) say it's weaker, others (like Sumo and Rebellion) say it's clearly more powerful. Nobody has detailed hardware docs. We've seen what's inside, and got more questions than answers.


Durango would be using Jaguar, not Bobcat, yes it's derivative but Jaguar has a lot of FPU enhancements. The FP unit gets 128-bit processing and a 128-bit wide data path, and 256-bit AVX processing. The boost to FP performance should be substantial, it will be more like the big chips (Core/FX) than a low power core in that regard.
And Espresso has lots of cache, potentially even atomic and multiversioned.
 
Top Bottom