Fusion Microsoft = graphics/gaming pwnage?

[BenSkywalker]

AMD has the best combined cpu + gpu engineering resources on the planet.

And based on their track record they have the second best GPU engineering and, being very generous, third best CPU engineering department in the world.

AMD will be able to offer a superior APU in every segment to anything nV/Intel combined can offer because only AMD will be able to offer a cutting edge CPU and a cutting edge GPU integrated on a single chip.

I could say only Cyrix will be going quantum computing for under $1 in the next six weeks, doesn't mean it is going to be close to reality. AMD has proven so far that they can do nothing even remotely in the league of what you are describing, in fact their stated goals aren't anywhere close to it. They are talking about releasing a very low end GPU paired with a low end CPU(must compare it to Intel's offerings in the same timeline) that by the looks of it will struggle to match the current 360s performance. Consoles may not be close to 5890s- but they still slaughter integrated solutions which AMD has clearly indicating they are headed for anything coming out in the next couple of years as far as on die graphics solutions.

Right now, where is AMD's answer to Tegra? We will ignore Tegra2, let's see something from AMD that is remotely close to year old nV low power technology. For that matter, where is AMD's competitor to Atom? AMD has finite R&D resources, if they try to compete in every market Intel and nVidia are in they will get killed in all of them.

By Xbox 720 it is likely minimum HD (SSD) storage will be 120Gb+, broadband access will be a requirement and buying a 'game' at a store will not include the physical media, but an identification code that will lock downloading encrypted gaming code keyed to a specific gamertag/machine. Everything stored on an SSD with an inpenetrable hardware based DRM/encryption scheme. And making the DVD9 limitations and any need to license Bluray irrelevant.

I don't think anyone at MS is stupid enough to even hint at something that profoundly ignorant. Selling a console that can hold a couple of games at a time? Current gen FF13 is just a hair under 50GB, that's a single game on a generation prior to what we are talking too. Console gamers wouldn't dream of tollerating waiting a few hours everytime they wanted to play one of the game they bought for it to redownload. As a general idea, the average 360 user in the US has purchased just a shade under 12 games each- ~700GB would be the minimum for a next gen console for an average gamer. I'm not extreme, but I would need to store at least 50 games on my HD, so 2.5TB minimum for me before I would consider purchasing such a setup.

If it used a Fusion chip and hardwired an SSD onto the same board, it could be remarkably simple and cheap to manufacture and remarkably small as well.

That sounds more like a XBox 240 then a 720.

Right now Ms and Sony have settled into their respective niches

Sony isn't close to mass market price yet, they can't produce enough consoles to fill demand since the launch of the slim(go ahead and look around)- they haven't settled into anything yet. MS is rapidly losing marketshare to both Sony and Nintendo at the moment, they certainly will be looking at improving their standing next generation, but going with a horribly underpowered console with next to nothing for gaming storage doesn't approach being a good idea.

It also means by the time the next generation arrives, it will be incredibly powerful relative to the current generation, backward compatible and have a very long life cycle

The fastest x86 chip in the world can't come close to emulating the 360 yet, and you think that there will be an incredibly low power cheap solution that will do so within 24 months?

 

[cbn]

The fastest x86 chip in the world can't come close to emulating the 360 yet

What is to prevent Microsoft from making a non-x86 gaming console that can surf the net and do word processing?

ARM isn't x86, but I know it can at least surf the internet.

 

[alyarb]

The next generation of performance consoles will continue on the current path to large HTPCs on the POWER architecture with discrete GPUs. They offer much better threadhandling, vector performance and aggregate FPU throughput per watt than x86 and will be backwards compatible with the respective ecosystems that are in place today. There will be more demand for HD content that is both streamed and stored locally, and 1080 will eclipse 720 as the standard resolution for console gaming (if it hasn't happened already). These machines will probably have high density 2.5" hard disks. SSD is too expensive and Llano is too slow for systems like the playstation and xbox.

 

[tommo123]

Perhaps lifecycle would have been a better term to use.

By Xbox 720 it is likely minimum HD (SSD) storage will be 120Gb+, broadband access will be a requirement and buying a 'game' at a store will not include the physical media, but an identification code that will lock downloading encrypted gaming code keyed to a specific gamertag/machine. Everything stored on an SSD with an inpenetrable hardware based DRM/encryption scheme. And making the DVD9 limitations and any need to license Bluray irrelevant. If it used a Fusion chip and hardwired an SSD onto the same board, it could be remarkably simple and cheap to manufacture and remarkably small as well. One small board, it's power supply, cooling, a Natal port, an internet port and an HDMI port (adapters indluded).

Right now Ms and Sony have settled into their respective niches, will both be primarily gunning to take market share from Nintendo with their respective motion controlled solutions, and are more than happy to be cruising along making fat profits from their current generations for the time being. As are all their game devs and game publishers. NO ONE is in a hurry to stop milking the current cash cow.

It also means by the time the next generation arrives, it will be incredibly powerful relative to the current generation, backward compatible and have a very long life cycle. Or maybe only four times as powerful but very power efficient, cool running and quiet.

i personally like the idea of games going back to the cartridge days and coming on flash drives with a small part of the drive writable for saved games etc.

as far as downloading games though - not gonna happen - at least not the only way. 1 simple reason. download limits. PS3 games can take upto 50GB right? next gen consoles could break that barrier and may make use of 100GB disks.

that in of itself is no big deal since i can download that overnight. the real problem (in the UK anyway and i'm sure in other backwards countries as far as internet access is concerned) is download limits.

i'm on an LLU exchange which means my ISP doesn't pay BT for my bandwidth thankfully. other ISPs have 10-20GB limits. some which advertise as "unlimited" have a 80-100GB fair use policy. no downloading games there. Cable is the same. you get throttled down to nothingness.

even FTTC (BTs newest baby) has a stupid 40GB or thereabouts download limit. then there's the actual distance from the exchanges etc etc etc. for some, downloading 1 game could take days - literally due to their poor lines.

some are still on dialup ffs!

 

[ZimZum]

MS is rapidly losing marketshare to both Sony and Nintendo at the moment, they certainly will be looking at improving their standing next generation, but going with a horribly underpowered console with next to nothing for gaming storage doesn't approach being a good idea.

You do realize your describing the console that this generation has sold almost as much as the other 2 consoles combined?

Wii :68,354,047
360: 38,326,691
PS3: 32,616,712

 

[BenSkywalker]

What is to prevent Microsoft from making a non-x86 gaming console that can surf the net and do word processing?

Nothing, that is what I fully expect them to do along with pretty much every industry analyst I've seen comment on it. That would exclude AMD from making the CPU though in any reasonable sense. As far as adding a browser, MS wants you to pay for something else to get that luxury. They are behind both Sony and even Nintendo on that front.

You do realize your describing the console that this generation has sold almost as much as the other 2 consoles combined?

Did I miss a press release talking about MS hiring Shigeru Miyamoto?

On a realistic basis you can approach the market one of two ways looking at all available data. Go after the 'core' gamer who wants the added visuals that is going to require more power and is the larger segment of the market; or you can go after the 'casual' gamer and go head to head with Nintendo trying to make games that must stand up on the merits of gameplay alone. Good luck with that

 

[Rezist]

nV doesn't make CPU's they use the cortex A9. Anyways it will probably be a die shrunk IBM POWER9 derivative in the 720 with an ATI dedicated GPU.

Also why the talk about pwer efficiency? these are consoles probably wil have 250-300W power supplies next gen.

 

[ZimZum]

On a realistic basis you can approach the market one of two ways looking at all available data. Go after the 'core' gamer who wants the added visuals that is going to require more power and is the larger segment of the market; or you can go after the 'casual' gamer and go head to head with Nintendo trying to make games that must stand up on the merits of gameplay alone. Good luck with that

I don't think it would be all that hard to compete with Nintendo on the game play front. As I find the vast majority of Wii games boring after about 5 minutes or so. It looks like with Natal MS wants to go after both markets. Casual and hardcore gamers. Where as Nintendo says, hey we aren't going to get the hardcore crowd anyway so forget them. Lets focus on the people who typically never or rarely play video games and make the game play instantly accessible to everyone.

 

[v8envy]

The x86 code museum is the absolutely worst choice for a console. There's no need to pay the silicon, power and performance cost for maintaining backwards compatibility with the 8088 on a closed architecture like a console. You will *NOT* be seeing Llano or anything remotely like it in the next gen consoles.

I also agree that the game streaming model is not coming any time soon, at least to the continent I live on. For example, while "broadband" penetration in US is currently 12th in the world most of that "broadband" is not set up to handle hundreds of gigabytes of peak downloads per month.

Oh, and my current PS3 console can do word processing and web browsing just fine.

 

[cbn]

Oh, and my current PS3 console can do word processing and web browsing just fine.

I was surprised when I looked up "word processing" and "Play station 3".

Yep, this is a cool feature then. Of course, I have no idea how compatible Open Office is with Microsoft office. Ideally it would be great if someone could transfer a linux file to Windows 7 and vice versa.

 

[v8envy]

There is no such thing as a "Linux file." I edit MS Office documents on Linux all the time -- rather than deal with windows re-activaton I just used a Linux desktop exclusively for about 3 months after my accidental upgrade to an i7. Linux can mount NTFS and FAT partitions for read-write on dual boot systems, and there are utilities for windows to read Linux partitions as well. You can also run either Linux in a VM on windows or Windows on a VM in Linux using the native hard drive partitions, and communicate between the two using windows folder sharing.

Some nutbag also "ran" Windows XP in an emulator on a PS3.

A kernel update broke Eve Online for me so rather than debugging the problem I called Microsoft to reactivate. But until that moment I was 100% functional on Linux including running all the Windows games I cared about, hardware monitoring and overclocking. The 64 bit Flash runs better on Linux than on Windows (there is no 64 bit flash for windows afaik)! Any PC with a 8 series or newer nvidia graphics card will make a perfectly acceptable Linux desktop machine for 99% of the users out there.

Latest Open Office on Windows, Linux and Mac is fairly compatible with MS Office 2003.

Also, I'm hoping that Google Docs will function via the PS3 browser before too long.

 

[cbn]

Latest Open Office on Windows, Linux and Mac is fairly compatible with MS Office 2003.

Also, I'm hoping that Google Docs will function via the PS3 browser before too long.

A few months ago, I wrote a thread asking about running PC games (which are often times console ports) on Linux. People said running the game through wine works, but not without a performance penalty.

So by using a console with Linux vs a PC with Linux any issues with gaming are completely eliminated.

1. The console is often times cheaper than a PC

2. The console processor itself can be faster than x86

3. The console doesn't suffering porting and emulating losses like a PC would. (This further increases the processor advantage of the console)

This begs the question, what could a future desktop computer do better than a future console? At what price?

 

[alyarb]

emulating x86 on POWER is not fast and you can't virtualize it on nonnative architectures.

 

[cbn]

emulating x86 on POWER is not fast and you can't virtualize it on nonnative architectures.

So Open Office will not work on PS3? Is this why v8envy mentioned Google docs?

 

[alyarb]

There is only one version of OpenOffice that is compiled for the POWER architecture and it runs only on Mac OS, for running on old PowerPC macs. OpenOffice will also work on SPARC systems (using the solaris OS). My point is that you can virtualize whatever you want in linux as long as it's all x86. Of course windows games can be emulated inside linux. They were compiled for x86 and they are being virtualized within an x86 linux environment. You cannot, however, take a SPARC or PPC machine running linux and expect to get any x86 emulation to be usable. On that same principle, you cannot take a PPC console, no matter how fast it is, and expect to be able to emulate windows AND a directx game inside that, that's crazy.

cloud-based productivity apps are universal and this would be the best way to go for basic users doing basic document stuff - as long as they can access their work on their primary computer, which 99.999999% of the time is going to be an x86-64 machine running OS X or windows. "computer people" really shouldn't care that their consoles will never natively run their windows apps. computer people should have a separate, superior machine for running windows apps. they will let the PowerPC do what it does best which is vectors and FPU(gigantic asterisk here) and the hundreds of games compiled explicitly for the architecture; convenient since fewer and fewer games are being made for x86 as the years pass. what basic console users need is a transparent way of interfacing with their stuff like docs, email, http, movies, and music. normally, on a normal machine yes you would do this within windows. but all of this can (must) be achieved without windows if you're on some non-x86 computer.

If microsoft wanted, they could easily make a word processor that runs natively in the xbox os that can open and edit office2003/2007 files. They won't do it though because it will compete with their x86 products like windows and office. If people start working on their documents under a playstation-linux environment, then that is even worse and sure microsoft will eventually respond. The main question would be is it better to write so many native applications for these two different architectures or run everything in some virtual web machine or whatever. Will either side come out with badass linux environments that play native xbox/playstation games and run premiere/photoshop/openoffice and all this other stuff on their superfast console CPUs? no.

 

[taltamir]

This was triggered by a story I read in which Microsoft demonstrated a game on Win 7 phone/Zune, Xbox 360 and Windows that shared 90% of it's source code.

It occured to me AMD's upcoming Fusion platforms could provide a rather striking dovetail synergy with Microsoft's strategy.

Why would it? MS's strategy is to sabotage the PC gaming industry so that they can sell more XBOX games, because they charge an extra 10$ each, and cram it with extra DRM. This sort of demo is just an MS stunt that will not materialize into anything worthwhile...

plus, why in the world would fusion CPUs have anything to do with it? they will be on the extreme low of graphic for xbox and PC with them being IGP... and they will be way too big and power hungry for cellphones/zunes.
PC and Console gaming will have discreet GPUs for the foreseeable future (next 20 years)

I would give long odds AMD and Microsoft are in extremely close collaboration in developing Bulldozer and Northern Islands and the Fusion platforms that will be developed from them.

MS knows jack squat about CPU and GPU development. So it makes no sense. plus this is AMD tech and they wouldn't want it getting out.

It's a natural fit. Fusion can provide a common, highly scalable, highly energy efficient hardware platform able to supply a set of integrated, and very advanced, cpu/gpu solutions for all of Microsoft's hardware needs.

It is only able to supply the most minimal of IGP performance... nothing "very advanced" there.

A set of solutions neither Nvidia nor Intel will be able to supply.

Intel timeline shows it having their own fusion well ahead of AMD. Intel has already released i3 and i5 CPUs with on chip (but not on die) CPUs. They are about a year and half ahead of AMD in terms of "fusion".
Nvidia recently bought VIA (after years of hiring CPU hardware designers), when you buy a via nano netbook with its via made x86 CPU, you are now buying an nvidia CPU. It is only a matter of time before nvidia comes to market with their own fusion.

Microsoft could supply a set of platforms tuned to Fusion

no, MS is completely unable to do that. fusion is cheaper IGPs, period. Gaming PCs will have discreet graphics still, consoles will continue having discreet graphics (unless you expect the xbox4 to have wii level graphics), and phones will not be running fusion. Heck, AMD has NO market presence in phones, at all... the intel atom was unable to penetrate the phone market (but thanks to asus created the netbook market)... nvidia actually has a better offer on that regards with their tegra platform (nvidia igp + two cortex A9 ARM CPUs)

which AMD has indicated will be it's focus in the future.

Because cheap low end stuff sells... don't expect any innovation there. it is a better focus then crossfire though... what a waste that has been.

I can see, in three to four years

hahahaha

Windows 8, Windows 8 Phone, Xbox 720 and Fusion coalescing in a unique synergy that will be very difficult to compete with.

how in the world would that happen?

And game developers/publishers will absolutely love.

popcap games will be about the only one releasing games to that... the biggest problem with such cross platform games is that you have vastly different screens and controlls... a TV + controller is very different then a monitor + mouse/keyboard, which is in turn very different than a 4 inch touchscreen. There is absolutely no synergy there, and MS is not going to be doing any innovation or driving in this regard.
oh, and the second biggest problem is vastly differing hardware levels... simple flash games are the only thing that phones can run... which limits the xbox and pcs quite a whole lot.

 

[tommo123]

Why would it? MS's strategy is to sabotage the PC gaming industry so that they can sell more XBOX games, because they charge an extra 10$ each, and cram it with extra DRM. This sort of demo is just an MS stunt that will not materialize into anything worthwhile...

see, this is what i don't get. a lot of people now are sticking with windows over linux since gaming is too much of a chore on linux and not all games work.

if PC gaming ever truly dies then i expect a huge shift of gamers to linux. why run windows anymore?

these are also the people that friends and family ask for tech support and may follow suit

 

[taltamir]

see, this is what i don't get. a lot of people now are sticking with windows over linux since gaming is too much of a chore on linux and not all games work.

if PC gaming ever truly dies then i expect a huge shift of gamers to linux. why run windows anymore?

these are also the people that friends and family ask for tech support and may follow suit

the execs that run those companies are out of touch with reality.
Gaming is "a niche market for children", comics are "a niche market for children", porn is "something shameful that nobody uses".

they fail to realize that the ONLY reason MS has maintained its market dominance is because of their video games monopoly (granted, it gained it due to actual quality). They fail to realize porno drives the adoption of new technology. They fail to realize that comics and cartoons are amazingly popular (holliwood finally realized it though, so now we have transformers, GI joe, fantastic 4, ironman, watchman, etc etc etc etc).
To them the notion that games have anything to do with anything is a shameful and immature thing that they must pretend does not exist. It will cost them, heavily.

I actually have a completely feasible plan to breaking the intel and MS monopolies with one fell stroke (not something I can implement, but someone can). take over the gaming industry EA style, partner with a company to develop a new CPU that does not use x86, develop a new desktop OS that supports that (or use linux), and have a slew of new games come out ONLY for that new platform. intel and MS will collapse like a house of cards. But neither company is willing to admit that significance. Of course, it would be many years of hard work and successful business (assuming I am successful) before I ever got the chance to do something like that myself, and by that time MS And Intel seem intent on committing suicide.

mmm. actually, MS does realize the value of this now that I think about it. This is how they set up the xbox, they bought a bunch of game studios, and then released their PC games (ex: halo) as xbox exclusives, fueling sales of their machines.
http://en.wikipedia.org/wiki/Microsoft_Game_Studios

I don't know why MS thinks it can get away with sabotaging its number one seller (windows) for their xobox department (that loses billions of dollars per year)

 

[v8envy]

There is only one version of OpenOffice that is compiled for the POWER architecture and it runs only on Mac OS, for running on old PowerPC macs.

Wow, no kidding. I must have imagined

openoffice.org-base-core
openoffice.org-calc
openoffice.org-common
openoffice.org-core
openoffice.org-dmaths
openoffice.org-draw
openoffice.org-filter-binfilter
openoffice.org-gnome
openoffice.org-gtk
openoffice.org-impress
openoffice.org-math
openoffice.org-pdfimport
openoffice.org-style-human
openoffice.org-style-tango
openoffice.org-style-galazy
openoffice.org-writer

All in the PS3 ubuntu repository, compiled and available for installation on my PS3. Also mythical is the ability to just run the damn thing. Feel free to hit psubuntu.com and poke around for other people discussing this very package.

 

[alyarb]

i just checked what was available on openoffice.org for the sake of the discussion of emulation on nonnative architectures. perhaps openoffice was the wrong example but do you really think i care? i was trying to stay within computer bottleneck's context.

thanks for the spirited correction, though. (there is a tempting joke about plugging leaks that i won't use but i want you to still see the opportunity for humor because it's funny)

 

[v8envy]

A few months ago, I wrote a thread asking about running PC games (which are often times console ports) on Linux. People said running the game through wine works, but not without a performance penalty.

So by using a console with Linux vs a PC with Linux any issues with gaming are completely eliminated.

Er, what? No, you still have the same issues and even more of them. The console has relatively feeble general purpose CPU so isn't great at running a general purpose OS like Windows or Linux (or BSD or OSX or...). In addition, the PS3 only has 256 megs of ram (plus another 256 megs of framebuffer memory you can configure as swap). It's by no means a fast box for general purpose computing. In fact, it makes netbooks look snappy in comparison. I have mine using NAS (iscsi) over gig-e for storage instead of the glacial onboard hard drive as well.

There's a reason Apple ditched power for x86, and it's not because Power was a faster general purpose CPU.

As far as the penalty of running games with wine under linux: if you have nvidia hardware the performance loss may not be that big on a fast CPU. If the game is a directx game there is a layer converting directx API calls to opengl, yes. But NV hardware degrades performance far less due to not having game-specific driver optimizations than ATI, so the losses are on the order of 10-20%.

2. The console processor itself can be faster than x86

Only for specialized tasks, like FPU math. For encoding blu-ray rips the PS3 eats my i7 alive. At everything else it's 1/10th to 1/100th the speed. And I don't have access to the 3d functions of the GPU. Probably good enough for Grandma to browse the web and write email, but I'd go nuts if this was my only computer.

3. The console doesn't suffering porting and emulating losses like a PC would. (This further increases the processor advantage of the console)

This begs the question, what could a future desktop computer do better than a future console? At what price?

Everything, at a higher price.

 

[v8envy]

i just checked what was available on openoffice.org for the sake of the discussion of emulation on nonnative architectures. perhaps openoffice was the wrong example but do you really think i care? thanks for the spirited correction, though.

That's the beauty of open source. Porting it to hardware and software platform of interest can be done by the end user, not the vendor. It doesn't matter that Sun doesn't care about the PS3 as a platform -- someone else can do the work.

And emulation speed on non-native architectures can be practical. My i7 has enough horsepower to emulate the power architecture at least as fast as the last powermacs. The powermacs themselves emulated the 68k instruction set for compatibility with the older macs.

Transmeta even managed to emulate the x86 instruction set (much, much harder to do well than power or 68k) well enough to be almost competitive on their hardware.

I'm not sure how we got off on the entire emulation tangent however.

 

[alyarb]

i think it's because computer bottleneck is really excited about Llano and hopes one day that we will have wii-sized APU systems that can run ps3/xbox/windows stuff smoothly. except it's just not fast enough. POWER gives you enough performance to run games while using ~150 million transistors or less. meanwhile x86 is getting into the billions and they haven't even hit the 100 gflop mark yet. i doubt we'll see x86 in consoles ever again.

it makes you wonder if altivec will ever support DP FPU.

i think transmeta was a little ahead of their time and i'd like to see what someone could do with the concept today.

 

[v8envy]

Oh, from a performance/power usage standpoint you're absolutely 10000% correct. Non-native virtualization, while possible, will never be practical. What makes it worse is the x86 is a code museum, and it's much easier to emulate a clean architecture on a dirty one than the reverse.

Eventually consoles will get fast enough to emulate computers of today. But it'll be at least a decade, and general purpose computers of THAT day will be faster still.

Transmeta's basic premise was flawed. You'd need to violate the laws of physics and thermodynamics to expect an emulated environment to be as fast as a native one. It's a safe bet that an emulated CPU will always be less efficient than a native implementation, and thusly always inferior: anything virtual can be implemented in fixed function silicon instead. That hasn't changed since the day of Transmeta, and I doubt their approach will ever be competitive.

 

[taltamir]

anything can emulate anything else given enough power to account for overhead, and a dedicated software team to implement the emulation.
the problem with such emulation is that its years behind, expensive, and lower performance than natively running things.