Fredrik Aldhagen :
Don't get me wrong, I like Chrome OS, I like ARM. This whole thing came up when you mentioned gaming, which I don't think Chrome OS is suitable to.
I just don't see Chrome OS as a viable gaming OS. As for unifying mobile and desktop, some games that are suitably to one are not suitable for the other. I don't want to play Temple Run 2 on my desktop, neither do I want to play Guild Wars 2 on a mobile.
Yes, I know about the relation between Linux and Android.
What I'm calling a niche is Linux gaming, not Android, and most Linux users are running x86 devices (not counting Android devices)
I don't want gaming to be x86/Windows exclusive either, but x86 is the most powerful platform at the moment. I would love for ARM performance to be competitive with x86. As for data centers, those usages differ from home computing, you usually have very parallelized workloads, so you make up for a lower core performance by having more cores.
NVidia is making ARM chips because they don't have an x86 license. I can't speak for AMD, maybe they find ARM a better choice than x86 for low energy devices?
As for Macs going ARM, I've heard that rumour too but not seen anything concrete. I admit it would be interesting to see.
Intel is highly profitable because they can charge a premium for their product, their only other x86 manufacturer is AMD which is seen as something of a budget product.
I don't think ARM CPUs will be high margin items like x86 CPUs, there are too many competitors, good for consumers not so good for manufacturers. Now combine that with the R&D costs of making a high end CPU.
As for Google making an ARM CPU, it's really the only reasonable choice if you were to make a CPU.
You misunderstand me, I didn't say 2GB RAM would kill Android devices, but that people don't want to be forced to buy a new tablet when they bought one last year. If Google makes it so only devices with 2+ GB can run KitKat that would leave out a lot of existing devices, including their own 2012 Nexus 7.
Don't forget that mobile game development is CHEAP. However, if you aim to make the sort of game people are used to seeing on consoles development costs go up considerably.
I mostly agree with you but I don't think we got a viable alternative to x86 for the high end at the moment. If every developer decides to make Linux ports available then yes we can ditch Windows.
I like ARM, I just don't see it replacing x86 any time soon.
As for Windows, Linux has made very large advances when it comes to user friendliness in the last decade, meanwhile Windows went from XP (which most people liked) to Vista (sluggish with lots of unnecessary visual features) to Win 7 (what Vista should have been) to Win 8 (tablet OS on a desktop) and Win RT (Win 8 without backwards compatibility).
somebodyspecial :
A good hockey player plays where the puck is. A great hockey player plays where the puck is going to be.
Wayne Gretzky
Quick looking at TODAY. Look at the last 5yrs, and start thinking about where we'll be in 2018-2019. What will 10nm do to a soc's power? How will 64bit (already in iphone which sped up everyone's plans on mobile) be on mobile in 5yrs? In 5yrs will my phone/tablet have 6-8GB of ram or more? That's just two $20 samsung modules even at todays size (between 20-30nm said toms). I'm sure at 14nm/10nm these may become 1 module, or at least brings this bill down to $30 total for 6GB. I'm thinking 6GB is norm at 14nm, and 8-12GB at 10nm.
My comments are where the puck will be soon, not today
I'm not saying the PC will die, but it is morphing at the least, and I don't believe it will be a WINTEL monopoly shortly. I say you are looking at a different world for these two in 5yrs or less as we already see adobe etc inching to mobile and games are going to be there in droves next xmas with 60% making something for them right now. GDC 2014 will show higher than 60% making mobile I'd guess and depending on Q1 console sales (GDC 2014 in march IIRC), we may see a smaller than ~12% planning anything for consoles, which may cause a PC rise above 40% also. Yes I'm crossing my fingers but really see it as a foregone conclusion.
WOW, should have cut out a lot of this, but oh well...Most are done reading this thread anyway
This is hard to answer. If you look back 5 years and compare computers with those of today, a lot of people are still happy with the specs of 5 years back.
Of course there will be high end mobiles, and we may very well see 8GB mobile phones in the next five years.
And yeah, long post is long.
I cut a bunch above, and hopefully didn't totally mess up the post...LOL.
It seems like we're pretty much saying the same thing (what we'd like to happen), just disagree on time of it it will happen or how much will happen at all.
I think in 5yrs Wintel is much less important (with many games being made on mobile now first, apps will be late and prolong this story some). Agree chrome may not be the best, so as I said maybe its a dual/tri boot etc with others on the system to cover all segments needed to make Wintel less important. I see it happening though no matter how it happens. My main point is that without Wintel doing something to stop it, I expect Arm to eat a REAL chunk of the x86 pie (clearly pro apps will keep the platform alive longer, but games moving to mobile will lessen the need for x86 required for gaming as it is today almost).
I think SteamOS will bring more games to OpenGL etc which are easily ported to any other linux base (android, ios etc) or even as I said before a dual/tri boot of things that cover all x86. Maybe you fire up chromeos for safe browsing, you fire up android/linux for games, then linux for apps? No idea how the end will shape up, just that it will be some combo or one rising to the top to face x86. As an example google has no FABRIC for a server cpu but can easily partner with anyone who does to provide it (I really though when it was low Google would have picked up NV, say the $8-12 range, but at $15-16 maybe it's out of reach). I think at some point Apple buys IMG.L (for the gpu/mips maybe), so google would need a full mobile solution and the SOC is easily had by buying NV before they explode into mobile gaming or x86 space replacing wintel as the only option.
I fully believe GPUS will be our new kings going forward, not x86 cpus (already made it to the top of the supercomputer list with Tesla, about to take over Green500 list with the same, and next year or 2015 will feed them with Tegra5/T6). In order for Google & NV (or anyone in this ecosystem) to succeed they have to make DirectX (games) and x86 (games and apps/office) worth less to end users so you can migrate perhaps without the pain
Both will assault games first them hope apps follow. Games will go where massive unit sales are (already have as GDC shows), while PC's could survive for longer on niche apps. 3dsmax, adobe etc are not used by 90% of us or more. Handhelds are already are dead IMHO & consoles day of reckoning is probably next xmas as they get "good enough". Maybe BIG PRO apps will never get to arm (but games alone can make it viable and kill consoles/handhelds in the process, and can dock at home for basic kids homework, browsing etc). We are doing more and more offloading to gpu these days and all the fun gets made or played on a gpu. I need a gpu upgrade during a PC cycle, I don't upgrade cpus these days. I used to upgrade both during a PC cycle, now just gpu upgrades. I mean, during my PC's life I usually upgraded both at 2-3.5yrs or so, and another gpu at 4-5yrs (1st gpu bought in 2nd-3rd year and 2nd gpu upgrade in year 4-5 to extend life one more time). That 2nd gpu may end up out-gunning the old cpu in the end, but carries right into the new board/cpu in year 6 or something (note the cpu is now bought much stronger - like planning a 4770 or broadwell, which is top end). I used to buy the best cpu OCer in mid-range ($200 style, now buy $340 from day one pretty much top chips). Now I just buy near top and OC in the end as it slows perf wise compared to the 2nd gpu I upgrade to. If that makes sense
I'm morphing to fit a slightly different cycle where gpu is ultra important.
I wouldn't mind an amped up temple run (Temple 3??) with higher textures etc, and wouldn't mind guild wars 2 on TV via HDMI out from a tablet/phone or just a android console etc (steambox?). I see all mobile as portable consoles out to TV, I have no intention of playing anything on 5in screens
Hopefully that clears some of my point up (output to 65in TV! and mobile only in a doctors office etc).
Totally agree x86 is most powerful at the moment. Hence my stream this device to other device hooked to TV, like PC streaming to shield, which I think NV will roll out to every tegra device as they perfect this technology. Right now they stream to their own device and gpu so they can troubleshoot a known commodity, then branch to all others.
Agree, NV heading to ARM due to lacking x86. I think AMD sees x86 becoming less important and also they clearly aren't going to knock Intel off the crown of x86. But if android becomes more important than Windows, clearly NV will benefit from ARM's side not x86 who already have a GOD called Intel right?
A GPU can make you the god on ARM's side where Intel is a dominant force on x86 with the help of windows keeping them there to some degree (apps etc, as you say we currently have no other real option). AMD has to run there, as ARM/Android will be a real option in under 5yrs. If Adobe gets the full suite either in a browser (anywhere) or totally ported to Android/linux, and games are there already moving massively, and Google gets a REAL office package for home you're going to want to be chasing SOCS over there where WINTEL doesn't run the show. NV had their chipset business shut down by Intel. There motive for profits and revenge is clearly ARM 64bit either in SOCS or a CPU for desktops paired with an NV gpu that at some point won't work on x86 (assuming AMD gets weaker, you block your gpus from x86 side to bring all games to your side). Leveraging your GPU to make a new cpu king. If Intel fields a great gpu (as good as NV/AMD) they can change the story but they have proven they have no idea how to do this (or all the good engineers already work at AMD/NV etc). NV will use the gpu to kill the cpu, just like Intel used the cpu/windows to kill NV chipsets. Karma is a ....
Mac - Apple has a FAB (bought UMC) and is making another soon. They bought PA Semi so they could design swift etc. Next step is owning a gpu but they may already have people working on their own, but I thought IMG.L would be logical as they are broke and used in every iDevice (ipod, iphone, ipad). They could easily own them for $1B-2B. Net income never over 60mil, this year 6mil, had to borrow 20mil just to buy 100mil purchase of mips. Either apple is working on their own GPU or they have an idiot running the company who should have already bought IMG.L for the coming gpu war. They don't have a modem either AFAIK which is a mark also on AMD, TI quit socs due to no modem in a war where everyone else has one 2014. Apple likes to dictate things, so they have to can INTEL. Samsung makes $8B/Q now due to 65% of their phone coming from IN-HOUSE products (memory, soc, glass etc). In order for apple to compete, they need the cpu, gpu, modem, and anything else they can get in house.
Unfortunately for Wintel everyone is banding together as a group to kill them. FABS with Global Consortium, android/steamos with games, apps that work for most users (except pro apps), chips for low power servers etc. The whole group and everybody's plans are making x86 worth less yearly and it seems to be accelerating. Android is so big WinRT can't even get off the ground. Everyone has left (all oem's, only MS is left) and they already had to write down 900mil. Buying Nokia to get win phones out won't help stop android as it's too late (maybe 2yrs ago this would have worked). Intel is too late too. They should have moved mobile to 22nm first not last as AMD was already pounded into submission. But they let ARM get too big. Now it's just a question of how much of the pie will ARM/Android/Linux steam together? It is no longer a matter of WILL THEY, just how much and how fast
MS is the most screwed as they have no hardware to stop other hardware from making the ENEMY OS's stronger, and even Intel has joined making android stronger with x86 on android. Windows/Office is left in the cold as games leave and so does Intel. It's hard to sit on a stool with one leg...LOL. They had games (gpus), apps, Intel. Now they have apps. Intel on the other hand moved to fielding a modem (needed to field a phone where the unit sales are), and a SOC but socs came too late and modem is a year late from stopping the ARM party too (maybe 2yrs late with modem aspirations).
If I triple the size of the ARM soc I charge you $200 and pit it against Haswell desktops at $340 (4770). NV etc won't sell you a desktop chip for $25 like socs. They will be made for $100 (like a Richland/Trinity) and charge $200 (remember 3x the silicon as stated before so much more complicated and larger than 80mm of T4). Basically they will just move ARM into AMD product stack APU pricing (probably better perf than AMD then too).
http://www.pcmag.com/article2/0,2817,2427509,00.asp
Last month, you see Intel feeling pain, in denial on tablets, pulling in modem rev2 etc. They sound like AMD "we were two years late to mobile". This is BS as its really 4yrs+, but Intel really is 2yrs late and at least they have a modem that will be on chip in 2014 probably but they needed it in 2012/2013 to stop what we see now and about to happen. You needed a modem ON DIE before they got a Desktop 4ghz chip out for a 400-1000w box that isn't aimed at tablets/phones. It looks like that race will be a TIE which is bad for Intel, as they needed a WIN but are late.
"Eul and Intel CEO Brian Krzanich reiterated the company's commitment to being nimbler and faster-acting in getting its best platforms for mobile devices in the hands of smartphone and tablet manufacturers. Intel chairman Andy Bryant went even further, saying he was "personally embarrassed" about the company's sluggishness in catching up to mobile rivals like Qualcomm, which supply smartphone and tablet makers with processors based on the ARM architecture."
That about sums up why I think they are about to get a HUGE dent in their pie
If Baytrail had been the first 22nm chip to pop off the line, they might have slowed this party down. If Sofia was out NOW, same story. But it's not and by the time it hits how much further will mobile be entrenched in ARM and games? No LTE from Intel until 2015? Only 3G in 2014 and that says SHIPPING the soc not the device with the soc? How much better will Adobe's Photoshop be on mobile next xmas? How much better will apps be on android etc next xmas when Sofia has any chance of being bought in a device and rev2 with LTE 2015? They are still talking TABLETS here (because they have no PHONE-able parts yet), when they should already have a PHONE part IN a phone today. Mobile is making x86 worth less, and Wintel is too late to stop it (already killing console/handhelds look at wiiu/vita sales). Apple let Android get too big, and now margins will suffer because we have a "good enough" android option (actually in some cases BETTER option right? Memory cards, replaceable battery etc). If Qcom had started building a GAME gpu and gaming focused announcements 3yrs ago (instead of just hiring 30devs last year), we may not be talking about tegra today. But NV has the best foothold on the future since games are worth more than modems shortly. AMD has a great gpu but I think they're far too late as they have no modem and are worse off than Intel here (have no fabs either now, no modem, nothing but Seattle on the roadmap and its for servers). I hope AMD gets bought to save the gpus before they end up like the cpus. AMD cpus are dead IMHO - x86 AMD party is over, though someone else buying soon could fix this with no 200mil/year going to interest on debt, and billions in the bank, no ties to GF (which gets AMD 200mil+ fines yearly, Dec31 another 200mil due, last year more, year before that 750mil fine etc) - might fix it with that a company like that buying AMD. Could Apple/Amazon/Google buy AMD and make x86 cpus? That would change my whole story
Regarding forcing OS restrictions to higher memory: Phone makers already move on and many force a new purchase to get the new OS. You really don't get out of that much unless you buy a NEXUS and even then maybe you're left hanging one model later maybe. So I'd rather see them force 3GB+ on us since 1B units are replaced yearly now anyway and it's only $10-20 more. As google, you don't want to give MS/Intel more time to stop your train. You need devs making REAL apps on top of games coming next xmas etc, where 512MB OS enabling low-end models forever will never get you there. You need to have apps that are USING 64bit and large footprints. I'm kind of talking the ancient art of war here, not really about how I want it to turn out. If you let MS have the time to figure out WinRT/Win8 merging, or Intel get a modem+better gpu into mobile this war can come out very different. If you get apps over to android soon you're going to get some x86 pie no matter what Wintel does. Chromebooks are already great for people not needing office, pro apps or high end games (which is a HUGE portion of the world - what like 80%+?). A large portion of america connects to the internet with only a mobile device so many in the future may never see a Wintel device and be fine as we already see now to some degree. That's bad for wintel.
Witcher/Witcher 2 cost 7mil Each. Torchlights even less. When you remove EA/Activision/Microsoft (and the overhead organizations this big cost) from the picture a game can be made cheaply. Check out Tides of Numeria, Shroud of Avatar etc on kickstarter. Most made for under $4mil or so. We are moving from large devs back to indie types like Origin (ultimas), SirTech, Bullfrog, Interplay etc from years ago that were 5-20 people making the game. Instead of what we have today with 100 people and huge management/ceo staff, marketing teams to pay on top (you really just need steam etc today for that-they put indie etc front and center instead of someone paying millions just to get a retail shelf space), shareholders pushing beta crap etc to meet a deadline that only makes a shareholder grin and a gamer wince in pain from patches. The funny part is, a lot of those little devs are the same guys that were running the old companies who went through the EA/Activision/MS buying/merging spree. Now they have money to fund their own projects largely, and many with the same teams as 15 years ago. AWESOME stuff coming soon from these guys who are reverting to old school quality and gameplay. Carmack himself spends all his time on mobile programming now, none on x86. Sign of the times ahead? Unreal Engine runs on mobile now too and ported in 4 days by a small team.
I guess it's just down to what either of us calls "SOON". I say 5yrs and I think you'll see it happening at some point if we have this conversation again next xmas. You may say 5yrs by then
Anyway, great conversing with someone on tech without it resorting to name calling etc that usually happens. For the record, I have no love for any of these companies...I just like the ones that make me the most money (stocks), or give me what I want as an end user
I enjoyed the back and forth despite differences on where we see the puck and how soon it will end up somewhere else
I wish more forum discussions went like this and I hope you're in here next xmas, as it will be interesting to at least see what has changed by then (no matter how it works out, its certainly getting exciting these days!!) since things are certainly happening faster today. It took Rimm less than 5yrs to go from king to crap (1% share now), and the downfall started with an iphone getting exchange. 10yrs ago if you'd have told me Wintel had any way of being taken down I'd have laughed at you probably. But then you look at Rimm, Nokia, etc etc and how fast they implode and think...Hmmm... I'm sure Apple thought they'd have more share than android today too
Parting thoughts: What happens to all phone maker's MARGINS as Amazon enters phones and gives them away for less than Google just to sell a book/movie or ads? Amazon seems content to make $0 on margins of hardware for the foreseeable future (google almost does this now - both apple/google supposedly have consoles coming too-maybe amazon too if rumors are true). I see Apple dropping more in stock price without some massive game announcement or great console soon leveraging the apps store that might again make them special and samsung weathers this storm better as they make a ton of components. Sorry if stocks aren't your thing...I leverage my tech knowledge for stock trades
Today I'd rather be invested in a component maker than an OEM like apple who makes the mint on margins of the entire device (a Qcom/NV etc who make the most important parts OF the device). More margin in the whole device but you can quickly become Rimm, Palm etc if that company isn't thinking longer than next year. Services survive better here too. Google/Amazon can wait out everyone else in the war of hardware as they make most of their money from content/eyeballs etc. Strange times ahead but good all around IMHO
Have a great Christmas