Slashdot Mirror
AMD, Intel, and NVIDIA Over the Next 10 Years
GhostX9 writes "Alan Dang from Tom's Hardware has just written a speculative op-ed on the future of AMD, Intel, and NVIDIA in the next decade. They talk about the strengths of AMD's combined GPU and CPU teams, Intel's experience with VLIW architectures, and NVIDIA's software lead in the GPU computing world." What do you think it will take to stay on top over the next ten years? Or, will we have a newcomer that usurps the throne and puts everyone else out of business?
I predict wrong predictions.
Damping absorbs vibrations. Dampening is caused by moisture.
With greater personal power, we won't have Microsoft dictating what 3D features we can have. With individuals become supercomputers, these three companies will be out of business. However, personal survivability and power will be sufficient that former employees will be fine.
What?
"I like to lick butts!" by MobileTatsu-NJG (#32700246) (Score:5, Informative)
All three will be marginalized by the ARM onslaught. Within 10 years, smartphone will be the personal computing device, AMD and Intel processors will power the cloud.
In short: make your time.
Mod me down, my New Earth Global Warmingist friends!
Do you mean it will become an integral part of the chip? Or that it will be basically unused by standard compilers and only see use in hand-optimized libraries that are used through APIs? Or both?
$_ = "wftedskaebjgdpjgidbsmnjgcdwatb"; tr/a-z/oh, turtleneck Phrase Jar!/; print
It seems like VLIW is a bit like nuclear fusion -- in ten years, people will still be talking about how its practical realization is ten years away.
Let's look at x86. It's dominant because of mostly inertia. Millions of chips must be taken into account when deciding whether to try a new architecture. A personal supercomputer--a self modifying brain--will not have limitations on trying new things.
Or look at directX--this is a cooperative library for ensuring games can work with Windows. Personal supercomputers will be able to establish higher level descriptions when interacting with other supercomputers (posthumans). Without burned-in chips or IP (intellectual property will seems silly in this era), it's about the stories you can tell.
The use of futurism has been thoroughly discredited.
I predicted that years ago.
On the contrary, I think the CPU will go the way of the coprocessor. The humble Atom may be enough CPU power for most people these days, but you can never have enough GPU power... at least not until your po-- I mean, games, are photorealistic in real time.
A slashdotter who didn't build his own computer is like a Jedi who didn't build his own lightsaber.
Well... There's 2 ways of looking at it. Either the GPU and the CPU will be merged into one beast, or there will be further segregating of tasks. In terms of price, what's more efficient: Having 1 chip that can do everything (Picture a 128 core CPU, that has different cores optimized for different tasks. So 32 cores optimized for floating point processes, 32 for vector processes and 64 for "generic computing") or having multiple chips that are each fully optimized for their task. Actually, now that I think about it, I'd probably say both. Economy computers would be based off the "Generic" cpu, whereas performance computers and servers would have add-in modules that let you tailor the hardware more towards the task at hand. So the motherboard could get an additional 8 sockets (similar to DIMM sockets) that would let you plug in different modules. So if you need to do graphics heavy processing (video games, movie rendering, etc) you'd add 8 GPU modules to the motherboard. If you needed floating point capacity, you'd add 8 FPU modules... Etc... The advantage of doing it that way over the current PCIe method, is that you get to skip the southbridge (So these modules would have full speed access to system memory, hardware and each other). Of course, there are a lot of hurdles to implementing such a thing...
I am not an engineer, these are just thoughts that rolled off my head...
If a man isn't willing to take some risk for his opinions, either his opinions are no good or he's no good
There's a new GPU company called Zig? Where?!
You can always spot a sensationalist post when part of it predicts or asks who will go out of business. Or what thing will disappear.
For example, in his post, ScuttleMonkey asks this:
NNote, the post is a good one - Im not being critical. But change in the tech industry rarely result in big companies going out of business - if they do, it takes a long time. I think sun is the canonical example here. It took a long time for them to die - even after many, many missteps. Sun faded away not because of competition or some gaming changing technology, but simply because they made bad (or some would say awful) decisions. Same for Transmeta.
People have been predicting the death of this or that forever. As you might imaging, my favorite one is predicting Microsofts death. Thats being going on for a long, long time. The last I checked, we are still quite healthy.
Personally, I dont see Intel, AMD, or NVIDIA ding any time soon. Note, AMD came close this last year, but they have had several near death experiences over the years. (I worked there for several years...).
Intel, AMD and NVIDIA fundamental business is turning sand into money. This was a famous quote by Jerry Sanders the found of AMD. Im paraphrasing, but it was long the idea at AMD that it didnt matter what came out of the fabs as long as the fabs were busy. Even though AMD and NVIDIA no longer own fabs, this is still their business model (more or less).
I think its interesting how a couple of posters have talked about ARM - remember, AMD and NVIDIA can jump on the ARM bandwagon at any time. Intel already is an ARM licensee. Like AMD, they are in the business of turning sand into money - they can and will change their manufacturing mix to maintain profitability.
I also dont see the GPU going away either. GPUs are freakishly good at what they do. By good - I mean better than anything else. Intel flubbed it badly with Larabee. A general purpose core simply isnt going to do what very carefully designed silicon can do. This has been proven time and time again.
Domain specific silicon will always be cheaper, better performing and more power efficient in most areas than a general purpose gizmo. Note, this doesnt mean I dislike general purpose gizmos (like processors) - I believe that the best system designs have a mix of both - suited to the purpose at hand.
-Foredecker
Jibe!
I find it funny when Science Channel has one of those "Future" shows on, and you get some asshole talking into the camera with the little caption under his name pegging him as a professional "futurist".
He gets paid to make wild guesses. ::golf clap::
Living With a Nerd
That is what Transmeta thought. Intel proved more agile than they predicted. AMD, Intel and NVIDIA can move faster than people think. i suggest that it is their market to loose, not others to win. -Foredecker
Jibe!
The article is about the GPU business, which, IMO, is the less interesting aspects of these companies. The new hotness is all in the mobile space. You want interesting, read about the upcoming Cortex-A9-based battle going on for 'superphones', smartbooks and tablets.
People have been predicting the death of the desktop computer for a long time, but the problem is that there hasn't been anything realistic to replace it. We're within sight of that replacement. Once everyone has a superphone that can do the bulk of what people normally do (either when in mobile mode, or when docked to a nice monitor/keyboard/mouse at home/work), THEN (and only then) will the desktop PC fade away for the majority of people's home use.
I give it 2 years before this is practical, though of course, the ecosystem to support such a change is the hard part. Can you get access to all your data, etc., from your docked mobile? That's gonna be the key. What about when you want to take a call at home and use your docked mobile as a computer?
You want to make money - solve these problems now.
Actually, usually it's nVidia or ATI dictating what 3D features we have, with the other immediately implementing the same thing to keep up.
GLaDOS for President 2016! "Well here we are again. It's always such a pleasure." -- GLaDOS, 2011
Batteries limit mobile devices. Heat limits supercomputer servers.
The first low power CPUs like the Atom were lame. Better devices on the way.
If he is getting paid well, he doesn't give two shits what kind of clap you are doing.
Nerd rage is the funniest rage.
http://javadong.blogspot.com/2010/03/ftl-not-faster-than-light.html
Damping absorbs vibrations. Dampening is caused by moisture.
When singularity come, electronics end up all over.
Nerd rage is the funniest rage.
I disagree. Floating-point coprocessors basically just added some FP instructions to a regular single-threaded CPU. There was no parallelism; they just removed the need to do slow floating-point calculations using integer math.
However, GPUs, while they mainly do floating-point calculations, are essentially vector processors, and do calculations in parallel. They can easily benefit from increased size and parallelism: the more parallel processing capability a GPU has, the more realistic it can make graphical applications (i.e. games). And with all the GPGPU applications coming about (where you use GPUs to perform general-purpose (i.e., not graphics) calculations), there's no end to the amount of parallel computational power that can be used. The only limits are cost and energy.
So if someone tried to fold the GPU into the processor, just how much capability would they put there? And what if it's not enough? Intel has already tried to do this, and it hasn't killed the GPU at all. Not everyone plays bleeding-edge 3D games; a lot of people just want a low-powered computer for surfing the web, and maybe looking at Google Earth. An Intel CPU with a built-in low-power GPU works fine for that, but it won't be very useful for playing Crysis unless you think 5 fps is good. People who want to play photo-realistic games, however, are going to want more power than that. And oil exploration companies and protein-folding researchers are going to want even more.
GPUs aren't going anywhere, any time soon. Lots of systems already have eliminated them in favor of integrated solutions, but these aren't systems you're going to play the latest games on. For those markets, NVIDIA is still doing just fine.
Consoles come out with 1080p/DX11-class graphics. Graphics cards for the PC try to offer 2560x1600+ and whatnot but the returns are extremely diminishing and many current PC games will come to a console with keyboard and mouse. GPGPU remains a small niche for supercomputers and won't carry the cost without mass market gaming cards. The volume is increasingly laptops with CPU+GPU in one package and there'll be Intel CPUs with Intel GPUs using an Intel chipset on Intel motherboards - and they'll be a serious player in the SSD market too. AMD will do the same but suffer from being behind on manufacturing process and continue to struggle but survive like Macs do in a Windows market. nVidia will lose their way if they haven't already lost it, everything they've said so far about Fermi makes me think they're heading down a dead-end street. No i7/i5 motherboards with nVidia chips, new Atoms which kill ION, nVidia is being forced to go discrete in a market that is increasingly more integrated. The good new for them is that Intel will continue to flop on 3D performance including Larrabee so there'll be a market for discrete cards a little longer.
Live today, because you never know what tomorrow brings
The (I assume automatically targeted) ads were awful. They had one for an add-on sound card on the second page of the article, "The Death of the Sound Card". Lovely pictures and remembrances of old stuff though; I still remember my first 3dfx Voodoo, and sending the daughter-card back for a Monster when I found out the frame rate in GLQuake was weak sauce. Good times.
(1) Low power consumption that avoids the use of a traditional clock. (2) Stack machine architecture - produces dense code. (3) Multiple stacks with shared hardware and compiler support.
Lol, good point. Reminds me of a conversation that took place at our dinner table last night:
Wife's Uncle: "I wonder how Michael Phelps can swim for a living. Doesn't that get boring?"
Me: "Sure...if you consider it boring to be a millionaire."
Living With a Nerd
Mobile phones don't have the shelf life required to replace the desktop. I like having my PC at home simply because it cannot easily break like a phone can. I don't worry about getting hit in the holster with a door from a careless worker and shattering my touchscreen on my home PC, and I don't worry about someone crushing it when they carelessly sit on a table that my PC rests on. Everyone I know with a smart phone replaces it every couple of years - but I have desktops that last 5 years before I move it to a test bench position or give it to a nephew as an upgrade for him ( even then, I only replace it for gaming purposes and not due to any hardware failures).
I disagree. It seems CPUs and GPUs are designed and planned well ahead of time. Tapeout occurs many months before products hit the market. Intel's Sandy Bridge apparently taped out in June 2009 and won't be released until 2011. Yonah taped out in October 2004 but wasn't released until January 2006. If it appears that these companies are responding quickly with new, competitive designs, it's because they correctly predicted the market direction and planned accordingly.
The companies can only really move fast in adjusting pricing, marketing, availability, and SKUs.
At some point, the GPU goes on the CPU chip, and gets faster as a result.
Maybe.
GPUs need enormous bandwidth to memory, and can usefully use several different types of memory with separate data paths. The frame buffer, texture memory, geometry memory, and program memory are all being accessed by different parts of the GPU. Making all that traffic go through the CPU's path to memory, which is already the bottleneck with current CPUs, doesn't help performance.
A single chip solution improves CPU to GPU bandwidth, but that's not usually the worst bottleneck.
What actually determines the solution turns out to be issues like how many pins you can effectively have on a chip.
I am still waiting for OpenCL to get traction. All this CUDA and StreamSDK stuff is tied to a particular company's hardware. I think there is a need for a free software implementation of OpenCL with different backends (NVidia-GPU, AMD-GPU, x86-CPU). Software developers will have great difficulties to support GPUs as long as there is no hardware-independent standard.
Mobile phones don't have the shelf life required to replace the desktop.
This would be part of that ecosystem mention I made.
If you dock your mobile phone to access the 'big' data (videos, music, etc.), and other stuff is in the cloud, then replacing the phone itself isn't going to be a big deal, and with cell providers going the hwole 'go with us for 2 years and get your phone for half price' thing, that will incentivize people to upgrade their shit more often, which, from a web developer's perspective, I find quite nice. :)
Move zig!
Damping absorbs vibrations. Dampening is caused by moisture.
True, but cloud computing has a serious obstacle to overcome in this area - the ISP. Every ISP is working on a way to charge per gig per month right now (I know, I work for one that actually has been doing it in the central US for over 3 years). ISPs are one of the greatest threats to tech like this simply because they want to find a way to make more money off providing the same service which is slowly choking the customer out of wanting to use the internet.
A patent dies everytime you mentaly masterbate on slashdot. Oh the horror.
That depends on the market.
If we are talking about the PC/Laptop market once integrated graphics are good enough at 1080p it is game over. In the workstation market I would agree with you.
Even photorealisim may not be worth the cost. Also that depends on A class games being optimized for the PC and not just console ports.
The way I see the future is this.
nVidia goes for the Supercomputer, workstation, and embedded markets. Their biggest product in numbers and revenue will be the Tegra line if they are lucky.
ATI and Intel duke it out in the X86 market with integrated GPUs and they will be on the die.
Intel and ATI will see their market share shrink in the mass market because they can not complete with ARM+GPU in speed vs cost vs power use. as Smartbooks, tablets, nettops, and Smartphones move more and more into the PC space.
I could be totally wrong and probably am but it is as good as a guess as any.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
will we have a newcomer that usurps the throne and puts everyone else out of business?
Unlikely.
After all, any newcomer would have to buy licenses for all those patents owned by the existing companies. That's a lot of money, and it isn't even spent for realizing the newcomer's innovative and great idea. That requires another big load of money. Good luck finding enough investors for your business.
I am not really here right now.
As bus speeds go higher and higher, you want components closer together, not further apart.
Change your model to putting specialized components on the die, and youa re on the right track. Unless something like optical busses come mainstream, in which case multiple sockets make some sense. Multipurpose sockets that allow you add vector-specific processing to your system for gaming, vs. I/O or GP processing for servers, vs leaving them empty for the budget.
But I think on-die specialization is the near future. Optical seems a long ways from here relatively speaking.
deleting the extra space after periods so i can stay relevant, yeah.
Yes - it takes about two years (or more) to go from a white board to first silicon. Until I worked at Microsoft, I worked at hardware and silicon companies. But remember, the competition to Intel, AMD and NVIDIA will be other silicon companies - not software companies. The new compitetion will have the same constraints. This is also a small industry - its very difficult to do someting both major and new in secret. When I was at AMD, we knew about Transmeta's plans when they were still in stealth mode. It wasn't because of anything nefarious - the community is small and leaky. -Foredecker
Jibe!
You just said people will never want photorealistic rendered porn. You have got to be the worst predictor ever.
Have you ever even met a person?
The GPU will go the way of the coprocessor
The GPU is a coprocessor
I disagree. Floating-point coprocessors basically just added some FP instructions to a regular single-threaded CPU. There was no parallelism; they just removed the need to do slow floating-point calculations using integer math.
However, GPUs, while they mainly do floating-point calculations, are essentially vector processors, and do calculations in parallel. They can easily benefit from increased size and parallelism: the more parallel processing capability a GPU has, the more realistic it can make graphical applications (i.e. games). And with all the GPGPU applications coming about (where you use GPUs to perform general-purpose (i.e., not graphics) calculations), there's no end to the amount of parallel computational power that can be used. The only limits are cost and energy.
So if someone tried to fold the GPU into the processor, just how much capability would they put there? And what if it's not enough? Intel has already tried to do this, and it hasn't killed the GPU at all. Not everyone plays bleeding-edge 3D games; a lot of people just want a low-powered computer for surfing the web, and maybe looking at Google Earth. An Intel CPU with a built-in low-power GPU works fine for that, but it won't be very useful for playing Crysis unless you think 5 fps is good. People who want to play photo-realistic games, however, are going to want more power than that. And oil exploration companies and protein-folding researchers are going to want even more.
GPUs aren't going anywhere, any time soon. Lots of systems already have eliminated them in favor of integrated solutions, but these aren't systems you're going to play the latest games on. For those markets, NVIDIA is still doing just fine.
Why Can't they dedicate 1/2 the die space to x86 cores and the other half for a big simd processor
While I am not a fan of porn I can assure you that almost none of it is currently "rendered" and frankly from what little I have seen the last thing anybody really wants is higher resolution porn.
In that category of video even HD maybe a step too far.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
True, but that ignores the economies of production. Right now, most CPUs are available in three flavors (Budget, Mid level, and High End). So if you look at the permutations of adding these specialized modules (100% CPU; 50% CPU 50% GPU; 50% CPU 25% FPU 25% GPU; 50% CPU 50% FPU; etc), you'd have potentially 50 different CPU designs... So the design challenge becomes much more difficult (Both the core design, and the production run and testing design). If you specialized everything, the number of designs for each part drops drastically (So instead of needing 50 models of CPU, you'd only need 3 or 4. Plus 2 or 3 GPU modules. Plus 2 or 3 FPU modules, etc)... The performance (and energy consumption) benefit of pushing everything onto one die would be offset by the reduced difficulty (and hence) of production for each... Again, I'm not saying it's the only way (or even the best), but it's one way that I wouldn't be surprised if we see in the not so distant future...
If a man isn't willing to take some risk for his opinions, either his opinions are no good or he's no good
What I find interesting is the overall lack of game-changing progress when it comes to non-3d-or-hd-video-related tasks. In March 2000, i.e. ten years ago, top of the line CPU would be a Pentium III coppermine, potentially topping out around 1 Ghz. I could put Windows XP on one of those (with enough RAM) and do most office / browsing tasks about as fast as I could with today's top of the line CPU. Heck, it would probably handle Win7 okay. Contrast the period 2000-2010 with the period 1990-2000. In 1990 you would be looking at a 25mhz 486DX.
Because die space is expensive. Why would someone with a low-powered notebook or netbook want to spend extra money on 3D graphics capabilities they don't need? And why should someone who's going to buy an NVIDIA card waste money on a CPU that's twice as large as it needs to be, because it has a built-in GPU they'll never use, when they could have double the number of CPU cores instead?
This is exactly the sort of progress that will NEVER happen unless Windows is no longer the dominant OS, and to a larger extent proprietary programs in general. With OSS, this sort of thing is just a recompile away. Maybe not for full optimization, but at least to get up and running. For example going from 32 to 64 bit, or x86 to ARM is generally possible with very few (if any) changes. But this is impossible if the source is unavailable.
Mac isn't as bad as Windows for this, since Apple writes the OS and many of the core apps, and has shown to be willing to provide a comparability layer for older software. But with Windows? It'll never happen.
"If you specialized everything, the number of designs for each part drops drastically (So instead of needing 50 models of CPU, you'd only need 3 or 4. Plus 2 or 3 GPU modules. Plus 2 or 3 FPU modules, etc)... "
Um, right now, you have 50 or so CPU models to accomodate the perceived value of specialization. This impacts choices of CPU speed, caches and their sizes, FSB size, etc.
How this is improved by making 50 or so SKUs of various modules is beyond me. Not to put too fine a point on it, but instead of 50-60 CPU SKUs, you think 50-60 SKUs of various components is a better idea?
Now, the idea of choosing and changing the mix, to add more vector modules and then add more FP modules for other purposes sounds coo,, but honestly, how often do you bother to upgrade your CPU now? Don't you take advantage of system board improvements often?
And we still haven't dealt with the problem of distance. When the FSB starts going over GHz, you will have lots of problems with interconnects. Optical is the future, but it doesn't seem to be in sight yet. So I think this modular concept will play out on-die, not as multisocket options.
If at all. Intel is banging out higher- and higher-speed chips, AMD is getting better fab ability daily, and ARM chips are knocking on the door and want in on the action. This is a great time to be building systems. Let's see, faster or cheaper? Both? Remember the 486 days?
deleting the extra space after periods so i can stay relevant, yeah.
Only if it's cubed.
If you don't know what AltaVista is (was), get off my lawn.
...which has a full-sized keyboard, a mouse and a huge display? I bet you call it 'a computer' or 'PC'. Because editing documents, watching videos, even web surfing on a 5 inch display with a micro keyboard sucks.
I just wish that AMD will finish its open-source drivers and ditch the infamous proprietary ATi legacy. Than Nvidia will have no other choice than to open their drivers or go bankrupt. I know, I am a dreamer.
Now you just sound educated stupid.
Mod me down, my New Earth Global Warmingist friends!
Actually the original 8087 had true parallelism. That's why there was the FWAIT instruction: It told the 8086 to wait for the 8087 to finish its computation, and wold be used before accessing any results from the 8087.
The Tao of math: The numbers you can count are not the real numbers.
Desktops will always be much more powerful than phones, and there will always be ways for the "bulk of people" to utilize that extra power.
Why on Earth won't the future be like now in this sense - mobile small computers (smart phones, netbooks, tablet pcs) that you can lose or break reasonably easily. Bulkier boxes sitting at home, that never go anywhere (not exactly a huge drain on the space inside people's home...where's the push to plug a tiny phone into your huge monitor)?
And in the middle laptops.
And why won't there be some "cloud" computing and some decidedly non-cloud computing? Do people in the future lose all concept of redundancy?
If anything, I expect a more heterogeneous range of computing devices/methods in the future, not less.
Azural - instrumentals
Unless those big dogs wake up soon from their stupor, an unknown startup will sneak behind them and steal their pot of gold.
You are going to need that pot of gold to fund your project.
Research, Engineering. Fabrication. Marketing. The barriers to entry here are not trivial.
Really? You know, searching for "HD porn" results in 11,8 million hits. There's obviously a market.
Dilbert RSS feed
the ecosystem to support such a change is the hard part. Can you get access to all your data, etc., from your docked mobile? That's gonna be the key.
And it's going to be why it doesn't happen for a LOOOOONG time. In the US, Wireless gives you a choice of AT&T, Verizon, Sprint, or T-Mobile. Wired gives you choice of your phone company or cable company. None beat having your data local, and as long as you're keeping a machine around with your data on it, why not pay ~$50 more so it can drive your keyboard, video, and monitor without plugging in your mobile?
Probably because you are constantly posting your views without ever producing any real working products that prove your view has any merit. Quit talking and get to work.
Phone as a Terminal
The best solution would not be to run apps on the phone at all. It would be to get always on bandwidth from a PC at home to your phone that was fast enough to do remote desktop at a speed where you couldn't tell that you were working remotely. Once we have that kind of bandwidth, the phones are basically done. The phone as a terminal. With this configuration, you get:
* Massive upgradeability on the phone since to make your phone faster, you just upgrade the PC in your home.
* Far greater battery life, as once the phone is a good terminal, adding more processing power to the PC will add power, but since that part is plugged into the wall, it won't drain your battery at all.
* Losing your phone does not effect any of your data.
* Replacing your phone is simpler.
* You can access the same application from a desktop, TV, or the phone, and there is no reason the interface cannot change for each.
* Better utilization of processing power, since people will end up with a home server anyway, for running their home media servers, security systems, home automation, etc...
* Cheaper. It will always be more expensive to build these things smaller, so putting it in a PC makes it cheaper.
* Faster to market. It takes time to shrink electronics.
* Possible functionality that is impossible on the phone. We are getting to the point where we may be limited by physics on how small transistor can become. This means that the amount of processing power that would be supplied to the phone as a terminal may be impossible to have in a handheld device.
atom is too lacking in cpu features to take over like that and low end amd chips are much better and they have a good low end video plan.
Your post is based on several assumptions that make no sense to me as a student of human nature, and an engineer.
1. 1080p is current technology. Even if we assume we will not have hologram visual output within the near future, there will still be some new technology that the powers that be will sell to the masses. It may be an incremental improvement, but it will still be enough to drive the markets.
1a. As long as it's new and shiny, there will always be someone to buy it.
2. Consoles use GPUs and CPUs the same as PCs do. There is a longer update cycle in place, but whenever each cycle ticks they adopt all the new technology that has been developed during the lifetime of a console. As such, it makes sense for the console makers to encourage such development.
3. Intel would have to shut down all of their operations to let nVidia claim the workstation market. Like it or not, Intel still makes pretty hefty CPUs, owns the workstation market, and has more disposable cash, and a bigger engineering staff than any other chip maker. The embedded market has even more competition for its crown, so I will not go there. The supercomputer market, while good for satisfying the nerd bragging rights quota, is not know for being an amazing source of profit.
4. The AMD vs Intel battle for the mid-range market is actually something I can see coming to pass. I would not be too surprised if this market gets a third player as the line between computation devices becomes blurred.
5. ARM is not the only company in the world that can make a low power chips. Worst case, ARM has a few years of dominance before the other guys catch up. Also, as the article pointed out, integrated CPU/GPU has several obvious advantages over discrete CPU + discrete GPU.
In all, while I am not ready to make my own predictions, yours could use a bit more analysis and tweaking.
You stand to lose much more by using multiple chips instead of multiple cores. As we get faster and faster clocks, the distance a signal can travel in a clock cycle gets smaller and smaller. Even with modern technology trying to access something off chip is likely to cost you hundreds of cycles. As such, you want to minimize the amount of off chip communication that needs to happen.
I think you hit the nail with the different module types, though I would implement it a bit differently. First, the motherboard system is likely to become a thing of the past in the next few years. There was already an article on here recently showing off a pluggable module system, which allows for standalone computational modules, which can then connect to other similar modules in a grid system. When you need more power, you could just go to the store, get a new module, and swap out or plug it into your existing system. The actual modules would be similar to what you described, though a bit more organized. For example: it would make no sense to have pure GPU modules since they would still need a CPU to manage them, but nothing would stop you from having a module with 1 generic CPU and 127 vector GPUs tailored for the rendering customers, or a 128 generic CPU for the servers, or even a 32 generic/48 vector/48 fp unit for the standard market.
Best of all, these same modules could be used for IO, extension cards and whatever else have you.
What are you talking about?
The current version of OSX can run apps as far back as 2001. Apple does not provide any official way of running apps older than that.
The current 64 bit versions of Windows can run apps as far back as 1993. Microsoft provides an official way of running even older apps (XP Mode). 32 bit Windows can often run 16 bit apps without the emulator.
Microsoft has lots to fault them for, but their record on backwards compatibility is WAY better than Apple's.
That's because the CPU was crippled by Intel (I think it was the 386sx that didn't have a FPU, the external 387FPU was simply a 386 with an internal FPU). In fact, I think the FPU on the 386sx was fine, it was just disabled by a bonding option. Quite the scam - sell 2 chips instead of one.
No, I don't trust in god. He'll have to pay up front, like everybody else.
Word.
"Draco dormiens nunquam titillandus."
bandwidth
netbooks benefit from gpus in the video decoding aspects. Especially with google implementing h.264 codecs for some of their youtube videos there's a big performance advantage to having a gpu instead of an additional cpu core.
"And we have seen and do testify that the Father sent the Son to be the Savior of the World"
1 John 4:14
You just described the business model for FPGAs, and reconfigurable computing. At some point, we'll have CPU+GPU+FPGA trifecta as standard operating environment, with FPGA dynamically updated as needed per application, to hardware-accelerate some particular software-intensive cycle-churning computations.
s/386/486/g
It boils down to the ascendence/dominance/supremacy of the financial sector. Finance ubber alles. Companies had nothing new to offer so they borrowed money (billions in some cases) to buy out their competitors. When the mergers failed to result in the massively greater efficiencies, they started selling off divisions and the shareholders got screwed. (What good did buying Saab do GM, for instance?) None of this mattered to the finance guys. They got paid on the front end. I need to get Paul Krugman's book. Damn, it's expensive.
Damping absorbs vibrations. Dampening is caused by moisture.
I have only one thing to say about NVIDIA: nvlddmkm. That says it all. I won't ever buy NVIDIA again.
My opinion is the best bang-for-the-buck would be to integrate RAM into the CPU. Perhaps a decent GPU with lightweight CPU added in would make sense if netbook continues as a separate niche, but again it might be better to put the RAM in as well. Between allowing multiple memory paths (no longer restricted by pins on the chip as much) and presumably lower power required to access the bits, it seems like a reasonable path to take. Imagine if the RAM were fast enough that little cache were needed on the CPU for added benefit.
pimtamf
I think any hardware company can survive until either their products become a part of another system (e.g. Sound Cards) OR seriously fail to compete against another product (not a good e.g., but MIPS architecture... which lost to x86 in desktop sector).
I am confident all three players (INTL, AMD, NVDA) will survive this decade with the exceptions:
1. AMD financially collapse
2. INTL builds a competing CPU+GPU product, which will cut lose NVDA
After all, in the PC targeted semi-con industry, there are not many players but still.. its a growing market!
I seriously doubt NVDA can win a x86 processor battle against INTL... unless they acquire AMD. Which will be an interesting tech-fight we can tune in. But still, AMD is a toxic asset with all the debts and mismanagement in later half of naughties.
Obviously, I want to see a CPU+GPU product. It can be a real game changer in this decade considering the upcoming trends like e-readers, powerful netbooks, smart phones so on.
Clearly, hardware companies are lucky compared to software counterparts. As in, software market can change overnight and that's it. If Intel was a pure software company and did a mistake like Netburst, they will be niche within months. And Microsoft.. they survived vista debacle plainly because of market share.. but if they continued until 2011-2012.. they would've been doomed and apple would've easily take over OS market.
3. Intel would have to shut down all of their operations to let nVidia claim the workstation market. Like it or not, Intel still makes pretty hefty CPUs, owns the workstation market, and has more disposable cash, and a bigger engineering staff than any other chip maker.
By workstation he likely meant professional scientific data/visualisation, which nvidia currently owns basically with cuda, yes the main cpu is intel chips, but we're talking gpu's here.. when was the last time you saw an intel gpu being used for massively parallel vector math or someone that cared a lot about their visualisatoin?
Also, as the article pointed out, integrated CPU/GPU has several obvious advantages over discrete CPU + discrete GPU.
except that whole.. ability to upgrade gpu business, while integration yields many advantages on the low/medium end, integration can restrict flexibility with the hardware though.
On the high end side the very idea of having the gpu on the same die as the cpu is ludicrous, modern gpu's already have transistor counts way surpassing that of cpu's
I hope that the opposite will happen, i.e. the CPU will go onto the GPU.
Let me explain: code is much smaller in size than data. It makes sense not to move data from memory to the CPU, but move code from CPU to memory! a big step in processing speed will come if each memory chip is also a CPU with a separate buffer for code. Instead of copying data, code will be copied to each memory chip's code cache and then applied over the local data.
This will considerably increase performance, as well as provide the basis for safe concurrency, since the data processed by a memory chip can't be processed by another chip.
A somewhat misleading comparison. Try instead to run Windows NT 4/Alpha apps on Windows 7/x86-64. Not possible? Well, what about running Windows NT/PowerPC apps on Windows 7/x86? No? What about OS X 10.0/PowerPC apps on OS X 10.6/x86-64? Oh, that one works fine.
Oh, and with Darwine and DOSBox, I can also run those old DOS and Windows apps quite happily on a Mac (and quite a few newer Windows apps). With Basilisk II, I can run those old m68k Mac apps on a Windows machine. Beyond a certain point, backwards compatibility is irrelevant.
As someone who ran NT4 when it was new, I can tell you that there were a lot of apps that didn't work correctly on NT, but did on Windows 95. When I got Windows 2000, a few NT4 apps stopped working and a lot of 9x apps still didn't work. Apparently XP and Windows 7 have better compatibility, but I recently tried some of those apps that didn't work in 2000 on my Mac, and WINE runs them perfectly.
Oh, and when you say 'without an emulator' you mean 'without a third-party emulator'. The WOWEXEC process is an emulator (or, if you want to use the WINE definition of an emulator, it's a translator).
I am TheRaven on Soylent News
In fact the combined Microsoft+Intel backward compatibility record is way better than anyone else, which is certainly one of the main reasons that the combination is #1.
"His name was James Damore."
You seem to have you timelines messed up. The 8086 and 8088 (16- and 8-bit bus versions of the same chip) did not have built-in FPUs. The 80286 did not either. The 80386 came in sx and dx versions that were similar to the 8088 and 8086; they had 16-bit and 32-bit external busses, respectively. The sx was initially much cheaper because you could use most of the same components for their motherboards as you used for a 80286.
The first i486 came with an FPU. This line was then split into the i486SX and i486DX because the yields were low and could be increased by selling the ones with faulty FPUs as i486SXs and the ones with faulty CPUs as i487s. Later, the yields increased, and they started just disabling working FPUs.
The i486 came a decade after the 8086 and 8087 though, so had absolutely no influence on the design of these chips.
I am TheRaven on Soylent News
Because it's spam. You post links to your ramblings repeatedly. You never back up your assertions that everyone else is wrong with evidence (or, in many cases, with anything like a coherent argument). You are convinced that you are right and everyone else is wrong, yet you don't ever produce any evidence that supports this hypothesis.
No, I didn't read your link - I've read other links to your blog and didn't want to waste any more of my time reading more. If I'd had mod points (and hadn't already posted in this story), I'd have modded you down too.
If you want to look a bit less like a spammer, at least try summarise your points (if you ever have any) in your post, then link to your blog for further reading. Don't just post a spammy link to the blog and an unsupported assertion.
I am TheRaven on Soylent News
I'm not sure why you think this is a property of the ecosystem. The handheld that I use is a Nokia 770. It's five years old, still works fine, and is now quite out-of-date. My laptop is a bit over three years old and will probably be replaced some time this year. Desktops? I've got a few of them in the attic somewhere.
The handheld has no moving parts, so it's a lot less fragile than the laptop. I've dropped it a few times without any problems. The CPU is around 200MHz, and I'll probably replace it soon with something in the 600MHz-1GHz window and a decent OpenGL ES 2.0 GPU. At that point, it goes to my mother to use connected to her stereo for listening to Internet radio stations and checking online TV guides and will probably see several more years of use. Not bad for a machine that cost around £70 when I got it in 2005... The laptop that I bought a year earlier cost almost £2,000 and saw three years of use before being replaced.
I am TheRaven on Soylent News
I pondered the feasibility of a stack machine revival about a year ago. Stack machines lost over register machines because out-of-order execution of stack-based instruction sets is really hard. Very low power chips, however, tend not to have out-of-order execution (although the Cortex A9) does. If you discard OOO and, for the same transistor count, can get two stack-based cores and enough i-cache to keep them fed or one register-based core and enough i-cache to keep it fed, then it's a clear win for the stack machine.
It's not clear, however, whether this would be the case. ARM's Thumb-2 instruction set is incredibly dense and you'd have to design a stack machine very carefully to make it denser. Register machines are easier to optimise code for than stack machines.
Stack machines also don't really play nicely with pipelining. You generally have to wait for one instruction to complete before you can start the next one. Register machines can start one instruction then if the other one has independent operands you can start the next one next cycle. You could address this by having a lot of SMT, so you overlapped instructions from different threads in the pipeline, and every individual thread would execute one instruction and wait for it to complete.
It might be an interesting architecture for running highly-parallel Java server code on, but it wouldn't be particularly good for the majority of existing code.
I am TheRaven on Soylent News
You are correct with what I meant about workstations. I see nVidia pushing the idea of the personal super computer more and more. Yes it will probably have an X86 CPU but will probably have several nVidia chips as well.
As to the gpu and cpu on the same die I do think you will see it very soon. Just as the FPU has moved on die so will the GPU. As the low end gets better and getter the high end will be used by fewer and few people. Just as it has happened with audio.
They will still be available but less people will use them.
And lets be honest. Every game maker on the planet will jump for joy the day that every $299 PC and netbook can run games at 1080p with all the eyecandy maxed out.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
"1. 1080p is current technology. Even if we assume we will not have hologram visual output within the near future, there will still be some new technology that the powers that be will sell to the masses. It may be an incremental improvement, but it will still be enough to drive the markets."
Someday but not for a long time. 1080p will be a video standard for a long time. The production of 1080p panels will become mainstream and create huge economies of scale. Also 1080p really is good enough. Of course we may see game makers and video card makers start chasing frame rate instead of resolution at that point. You may start to see cards listed at HD, HD120 and HD240.
Or you may see 3d take off but 1080p is going to be the mainstream standard for a very long time.
"1a. As long as it's new and shiny, there will always be someone to buy it."
Yes but that will be pushed farther and farther upstream. You will always have the super high end CAD/CAM, workstation, Video production market. And you will always have the nut case gamers. You know the ones that are playing with LN2 cooling.
Those are outside of the mainstream. Look at the title. There are still coprocessors and FPGA cards even now to handle super high end math on PCs. They are just rare. You also see them on servers for encryption.
3. Workstations will still probably have Intel or AMD cpus. But it will not really matter much. On workstations more and more of the high end lifting will be done by nVidia. Of course Ibm may offer Power based workstation and maybe are will offer one with 32 ARM cores with a huge number of GPUs. Who knows for sure but I think nVidia will fight hard to keep that market.
4. I don't know where the middle range will fall. In servers I think AMD will keep fighting it out. On the desktop they will for a while but the margins in that are going to get even thinner over time.
5. ARM doesn't make any chips at all. The just design the cores. ARM cores dominate the embedded space. You will not find a single Smartphone that doesn't use an ARM. MIPS has become an also ran. It is still in there but they are few and far between. PPC is still around in that space. They are doing very well in the game console market and automotive.
ARM has does so well in the low end that I think it will be very hard to beat them.
When you say you are also saying.
nVidia Tegra and Tegra 2
Ti OMap line.
Marvel
Qualcom "Snapdragon"
Samsung
Apple the new A4 in the iPad.
AVR makes some ARM core systems.
and goodness knows how many more companies all making fast low power and cheap ARM based systems.
Of course every thing changes. Windows may be so entrenched that we never get away from the X86 on desktops and laptops. Which I feel is a shame.
Or Intel may throw enough money at the X86 to make a fast low power X86 that competes with the ARM. But the ARM will still probably be cheaper to make.
My prediction is that things in the GPU market will not change a whole lot. The majority of PCs will ship with integrated graphics just as today. The difference is that they will no longer suck.
Once they no longer suck people will buy more games and fewer medium to high end video cards because there will be no good reason.
That that do will probably use them to drive multiple cheap 1080p displays. But most people will stick with one.
I do think that wireless HDMI will become a popular option if we ever get a good cheap standard for it. Maybe a from of streaming over wifi to wifi enabled TVs.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
Yes, it was the 486, not the 386 - after around 20 years a detail or two gets lost. And you are right about the 8086/8087 being an additional decade earlier, and unless Intel had a time machine...
Intel has always played hardball, and as unfairly as they can. It's in their institutional DNA to be FIGs...
No, I don't trust in god. He'll have to pay up front, like everybody else.
As to the gpu and cpu on the same die I do think you will see it very soon. Just as the FPU has moved on die so will the GPU.
FPU's only became integrated when it became a comparatively small portion of the die, I have no doubts that things like intel's gma line would be possible on die, but to say that everyone would be content with that is just silly.
Eye candy means maths, more maths capability means more eye candy, people want things shinier so they want a computer that can do maths faster, which in gpu's case means more parallelism and more transistors.
On die gpu's are for people who don't care about shiny, fps or resolution... people may be out there like that, but those people aren't buying graphics cards anyway, instead using the integrated stuff on the motherboard.
They even adopt new technology inside a cycle, though it's often used to improve power efficiency and lower cost.