Predict: The Next Generation Console Tech

[bgassassin]

Because it is most likely fake ?

Modern CPUs with SIMD units are very effective at traditional DSP tasks. The 360 does all audio processing on the CPUs. On PCs, the dedicated audio card is a dying breed, because the CPUs can do all the processing with just 2-5% of a single core and all you need is a DAC (and you don't even need that for digital out).


Cheers

I've posted the CPU, GPU, and memory in the thread a little while back as that came from target specs. The DSPs are speculation based on Sony's CTO.

http://eandt.theiet.org/magazine/2011/12/maasaki-tsu-interview.cfm

“It took five years before we saw games that used the full power of Cell, so we are used to looking ahead and having capacity,” Tsuruta says. “We are looking at an architecture where the bulk of processing will still sit on the main board, with CPU and graphics added to by more digital signal processing and some configurable logic.”

 

[Mobius1aic]

Creative are responsible for killing the DSP on sound cards
because of their greed, EAX got proprietary after gaining support in games, then a competing standard died and Creative was the only brand that allowed the cool effect in games. (or sound blaster emulation in DOS for that matter)

then microsoft pulled the plug with Vista, when they decided they would support an API that works everywhere. I've never played a game with EAX by the way.
without Creative, we might have had DSP on on-board sound cards . (well, we do have on-board "X-fi" sometimes and they have licensed "X-fi" for some time but it's pointless now. even more pointless is to buy a 300 euro motherboard rather than a 100 euro motherboard and a 70 euro sound card)

Playing BF2 or 2142 with EAX and then without it is a travesty. It made a difference when I built my first PC in early 2007. The $100 Creative Platinum X-Fi was totally worth it in that regard, and other titles like FEAR used it too. Of course, even 5 years ago, on board audio still sucked pretty bad. Even today, I'm happy to have my Asus DS Xonar in my current computer. It's 192 KHz capable and has decent EAX emulation for my older supported titles. You can feel a difference with a dedicated sound card if you have decent speakers, though it is good to see higher end HTPC centric mobos with good sound on them. A desktop I built a couple years ago (and sold to a friend last year) had onboard 192 KHz capable audio on the mATX mobo. It didn't support EAX at all IIRC, but it did open up the "hardware audio" option on plenty of games :smile:

 

[jlippo]

I can certainly see use for hardware video and audio encoding for improved remote play and social applications. (Singstar with HD 3d video export.. etc.)

Also wouldn't audio hardware for mp3/flac decode and sound processing be nice for main ram usage and keeping unnecessary data from reaching CPU cache?

 

[almighty]

I use a £20 ASUS Xonar DG sound card and it's sooooo much better then what my 2500k can do it's unreal!

I don't believe they're dying out at all, many new motherboards are directly fitting X-fi hardware onto the main board itself!

 

[DuckThor Evil]

So 80 million in FY 2013 (ending Apr 2014, right?). Perhaps a few launch titles? Seems pretty cold IMO. Does give a launch schedule though. Sounds like very little investment, meaning not many customers expected to be buying next-gen games up until beginning 2014, meaning sales only up to the year in advance at best. There won't be many XB3's or PS4's in houses by the end of that 80 million investment, unless EA are seriosuly lowballing it.

Ending April 2013. 80 million could be for new engine development, funding initial stages for multiple projects etc.

 

[TheWretched]

Well... EA already has Frostbite 2.0, which can be considered "next-gen" already. So by extrapolation, this money could very well be funneled directly into game development without any fluff needed.

About soundcards... I used to have several Creative cards in my youth, too. Still love my AWE32^^ But with Creative completely ignoring Linux and Microsoft "shutting them out" with Vista and 7, Creative essentially killed themselves with MSs help. I read that with Windows 8, soundprocessing will be "more open" again, but I am not sure what difference that'll make. But I did have my fair share of driver problems, when I went from XP to Vista and later to 7. Officially, my mainboard has a soundchip with DTS-Live (i.e. multichannel "export" to DTS via TOSLink etc), which doesn't work anymore, because of driver problems. If I had known that before, I wouldn't have bought this mainboard (ASUS again... bad drivers)

But for stuff like video decoding, I don't think either console will really need DSPs. PS3 doesn't have any and can already process h264, even when using MVC (i.e. 3D), and there's little on the horizon that would change that. 4k video? Sure, but that's "only" 4 times as complex as 1080P video, if it's using h264 at the same bandwidth/pixel. Leveraging the GPGPU stuff and a newer CPU (whatever it'll be) should be enough, I think. Assuming they'd drop the video acceleration in those chips.

 

[Gubbi]

I use a £20 ASUS Xonar DG sound card and it's sooooo much better then what my 2500k can do it's unreal!

Proper decoupling of power and ground planes is important for good analog output. If you value that it makes sense to invest in a soundcard.

For digital ( HDMI / SPDIF/ Thunderbolt ) it makes zero difference.

I don't believe they're dying out at all, many new motherboards are directly fitting X-fi hardware onto the main board itself!

There is no DSP hardware in those, its just a branding exercise. All the heavy lifting is done in the driver on the main CPUs.

Cheers

 

[Gubbi]

I've posted the CPU, GPU, and memory in the thread a little while back as that came from target specs. The DSPs are speculation based on Sony's CTO.

http://eandt.theiet.org/magazine/2011/12/maasaki-tsu-interview.cfm

“It took five years before we saw games that used the full power of Cell, so we are used to looking ahead and having capacity,” Tsuruta says. “We are looking at an architecture where the bulk of processing will still sit on the main board, with CPU and graphics added to by more digital signal processing and some configurable logic.”

Historic revisionism in full effect here.

Sony didn't look ahead, it took five years to gain traction with CELL because they didn't have a clue as to how to program the damn thing when it launched. At least three different approaches to programming CELL was presented in late 2005/early 2006, it took years before developers settled on the micro batch/shader-like system used now.

If anything, CELL is a lesson in why it is good to keep computing resources as homogenous and symmetric as possible.

Cheers

 

[Shifty Geezer]

Because it is most likely fake ?

Modern CPUs with SIMD units are very effective at traditional DSP tasks. The 360 does all audio processing on the CPUs.

And at a surprisingly massive cost as either joker or bkilian said in a thread (Advantages of Cell thread?). Could have been bkilian, and he said IIRC that a DSP was a requested feature.

Ending April 2013. 80 million could be for new engine development, funding initial stages for multiple projects etc.

A new engine is very limited though. EA games, which are many, run on lots of engines. Could be initial investment, but it's a small figure overall when phrased as "this is what we are investing in next gen". Apr 2013 could mean preparation before investment for releases in Oct-Dec 2013 and beyond.

 

[bgassassin]

Historic revisionism in full effect here.

Sony didn't look ahead, it took five years to gain traction with CELL because they didn't have a clue as to how to program the damn thing when it launched. At least three different approaches to programming CELL was presented in late 2005/early 2006, it took years before developers settled on the micro batch/shader-like system used now.

If anything, CELL is a lesson in why it is good to keep computing resources as homogenous and symmetric as possible.

Cheers

I couldn't really care less about the Cell part of that. I was just showing you the DSP idea came directly from Sony.

 

[Blazkowicz]

Proper decoupling of power and ground planes is important for good analog output. If you value that it makes sense to invest in a soundcard.

yeah I got a Xonar DX, even though a DG would have been enough.
it's so nice as to use power from my PSU rather than from the slot, to cut on the noise.
now I have to make the speakers stand on these little thingies that dampen vibration, for quite some spectrum all this quality is going to waste. after that my room while not tiny is too small.

 

[Teasy]

Because it is most likely fake ?

Modern CPUs with SIMD units are very effective at traditional DSP tasks. The 360 does all audio processing on the CPUs. On PCs, the dedicated audio card is a dying breed, because the CPUs can do all the processing with just 2-5% of a single core and all you need is a DAC (and you don't even need that for digital out).


Cheers

Yeah and it can often take up an entire core to do so.

 

[steviep]

And at a surprisingly massive cost as either joker or bkilian said in a thread (Advantages of Cell thread?). Could have been bkilian, and he said IIRC that a DSP was a requested feature.

bkillian has definitely spoken of a DSP as the most requested feature when they were designing the 360. He's mentioned on this very forum that there are games with a lot of audio streams (I think he used both something simple like Fruit Ninja and another example with racing games - both using up to 200 audio channels) that ate through an entire CPU core.

So yes, a few pennies per console on a DSP that can handle audio tasks is a worthwhile investment in mine and his opinion. For his exact examples, I suppose you can search through his post history.

 

[Gubbi]

bkillian has definitely spoken of a DSP as the most requested feature when they were designing the 360. He's mentioned on this very forum that there are games with a lot of audio streams (I think he used both something simple like Fruit Ninja and another example with racing games - both using up to 200 audio channels) that ate through an entire CPU core.

So yes, a few pennies per console on a DSP that can handle audio tasks is a worthwhile investment in mine and his opinion. For his exact examples, I suppose you can search through his post history.

Audio being run on the ordinary cores is what enables all those streams.

If you dedicate a piece of hardware for something like that you end up setting a hard upper bound on how many streams you can have and how advanced your audio modelling can be. Your only alternative is to throw a lot of silicon at this DSP, and you then end up with a fair chunk of your system being under utilized in the average scenario.

IIRC, bkilian pointed out that the original XBOX had dedicated DSP for audio (Bungie used it, I remember seing a video about the sound system in Halo 2, very impressive for the time). This DSP had poor software support and thus went un-used in many cases, which is the reason it was dumped for the 360.

Cheers

 

[DoctorFouad]

Because it is most likely fake ?

Modern CPUs with SIMD units are very effective at traditional DSP tasks. The 360 does all audio processing on the CPUs. On PCs, the dedicated audio card is a dying breed, because the CPUs can do all the processing with just 2-5% of a single core and all you need is a DAC (and you don't even need that for digital out).


Cheers

thanks, thats what I thought, with the power and multi cores/threads of nowadays CPUs, including DSP silicon is a bad idea in terms of hardware budget allocation for next gen consoles.

The other point why it is most likely fake and just speculation is the absence of any information not only for the frequency of the CPU as I mentioned but for the type of CPU architecture itself, in short saying the ps4 or nextXbox have 4 or 8 or 16 cores is pointless information without specifying the architecture and frequency.

Even the 2 GO GDDR5 is almost pointless information if we dont know the bandwidth, when we talk about 100 GO/s or 200 GO/s it is completely different performance.

 

[Shifty Geezer]

If you dedicate a piece of hardware for something like that you end up setting a hard upper bound on how many streams you can have and how advanced your audio modelling can be. Your only alternative is to throw a lot of silicon at this DSP, and you then end up with a fair chunk of your system being under utilized in the average scenario.

This is the old dedicated vs. programmable hardware debate. If you have dedicated silicon that is being underutilised because it's too powerful for the current task, that's a waste, but if that silicon is a very small fraction because it's very efficient, it's good value. Yes, you have an upper limit, but then hardware always has limits. To get the best overall system you need the right balance. That can include things like custom scaling hardware even though software scaling on CPU is more versatile, and hardware audio even though software audio is more flexible. I always thought audio was a trivial task until bkilian said otherwise, and now I can see the value in a small, potent audio chip could be very value regards it's silicon cost.

 

[Gubbi]

This is the old dedicated vs. programmable hardware debate. If you have dedicated silicon that is being underutilised because it's too powerful for the current task, that's a waste, but if that silicon is a very small fraction because it's very efficient, it's good value.

I don't consider DSPs dedicated hardware. They are still processors that run software, they just have an instruction set optimized for convolutes (typically). Dedicated hardware makes a lot of sense where it solves a problem that is hard to make fast in software, like the entropy decoding of H.264 advanced profile.

I don't think the advantage of a DSP is anywhere near what it used to be. DSPs were nornally designed to sustain one multiply-accumulate per cycle on a convolution loop. This implied one load, one multiply and accumulate and a store op. With automated incrementation (or addition of stride) for both source and destination you're looking at 6 or more operations per cycle, - all packed into a single instruction.

Back in the day, general purpose CPUs didn't even have single cycle multiply, nevermind the instruction decode/issue bandwidth needed to compete with DSPs. It is very different today. We have massive SIMD exec units, wide datapaths and robust memory subsystems.

Cheers

 

[Rangers]

Tim Sweeney made more hints of high spec next gen:

http://gamasutra.com/view/feature/169845/the_future_according_to_epics_tim_.php

Gamasutra: I know it's under wraps, but you guys are probably feeling pretty okay about Unreal Engine 4.

Tim Sweeney: Yeah, we're starting a behind-closed-doors showing of the engine to developers; this is part of our very early ramp-up cycle. We went through this cycle with Unreal Engine 3 starting in 2003 and 2004. At some point we'll make public announcements and ramp up to the point where developers are shipping games, but it's very early right now. We're aiming very high, and the intended platforms this is aimed at haven't even been announced.

So here's a theoretical question for you -- even though I'm sure that this situation has happened in real life. Say Microsoft or Sony come to you and ask, "What do we need to provide you, with our next generation of consoles, to help you make games better?" What would you tell them?

TS: Really for us, there are two things that are going to be essential for the console market going forward. One is that to bring together all of the features and expectations that gamers have built up from all the great platforms out there today, right?

There are great games with Facebook integration that enable you to hook up to social network sites and find your friends in there. To be able to do that from next generation games and consoles will be really valuable.

To be able to go and easily buy and download games like we do in the iOS App Store on future consoles will be incredibly valuable to us as developers, and make it that much easier to get our games out without over-reliance on manufacturing a whole bunch of pieces of spinning plastic to ship to consumers.

So having all of the things that you expect from the game industry as a whole and the best that's been done elsewhere, and to bring that together on a console platform is really important. We saw with the current generation, we went from consoles as a little fixed, TV connected device to an online network of gaming devices where you can play with your friends over the internet, get updates, even watch movies on Xbox 360.

We love that, and I think a huge portion of the business opportunity in the next generation is extending that concept even further forward. So this is a mainstream computing device that hooks into all of your social circles as well.

Number two is raw performance. The thing that separates consoles from FarmVille is the fact that consoles define the high-end gaming experience. When you look for the best graphics available in the whole game industry today, you look at Xbox 360 and PlayStation 3, and those games are the best out there, bar none. And so the big opportunities for future consoles is to bring that to an entirely new level by delivering a dramatic increase in raw computing power.

We measure that in floating-point operations per second, and now we talk about teraflops -- trillions of floating-point operations per second. What we want is as many teraflops as is economically possible to deliver to consumers, because that enables us to create the best quality experience as possible, and that will drive people to buy a new machine. That's a big challenge with a new console -- that you reset your install base from millions and millions of what you have today with current consoles back to zero. Then you have to convince everybody to buy the new hardware.

To do that you need awesome games that provide a level of graphical fidelity that people have just not seen or even imagined previously.

There's almost no doubt in my mind we're going to get really powerful next gen box at this point.

 

[bgassassin]

Even the 2 GO GDDR5 is almost pointless information if we dont know the bandwidth, when we talk about 100 GO/s or 200 GO/s it is completely different performance.

PS4 will supposedly be much closer to the latter than the former.

 

[almighty]

Tim Sweeney made more hints of high spec next gen:

http://gamasutra.com/view/feature/169845/the_future_according_to_epics_tim_.php

There's almost no doubt in my mind we're going to get really powerful next gen box at this point.

But is he talking about console players? In which case I can agree that they'll see graphics that they've never seen before.

PC gamers like myself will be harder to please as although there's not really any 'next gen' games out on PC given the right mods a game can look next generation quite easily.

Skyrim for example below.