Slashdot Mirror
Intel Discrete Graphics Chips Confirmed
Arun Demeure writes "There have been rumors of Intel's re-entry into discrete graphics for months. Now Beyond3D reports that Intel has copped to the project on their own site. They describe it as a 'many-core' architecture aimed at 'high-end client platforms,' but also extending to other market segments in the future, with 'plans for accelerated CPU integration.' This might also encourage others to follow Intel's strategy of open-sourcing their Linux drivers. So, better watch out NVIDIA and AMD/ATI — there's new competition on the horizon."
Nothing for you to see here. Please move along.
Competition is almost always good, so I look forward to this. I'd like to see Intel push ATI and Nvidia to create more power efficient chips, as it's quite rediculous right now.
What is so discrete about their graphics? They are telling about it in a press release? Or does this mean that they no longer restrict themselves to the continuous graphics market?
And if they enter the gaming video market, I can assure you that my next videboard will be an Intel one.
Intel drivers for Linux Just Work(TM). I installed Ubuntu 6.10 on my Acer notebook, with a i915g video adapter, and everything worked without any extra effort. And I'm even able to use Beryl/Compiz as my default window manager, without any stability issues.
Both nVidia and ATI should learn from Intel.
---- You know how some doctors have the Messiah complex - they need to save the world? You've got the "Rubik's" complex
The summary doesn't link to the proof, which can be found on the careers section of Intel's website.
Intel is years behind in this market. And they tried this once before, with dismal results: http://news.com.com/Intel+retreats+from+graphics+c hips/2100-1001_3-230019.html
If anything the graphics market has gotten even more specialized since then. I don't know why they think they can succeed this time.
The more you regulate a company, the worse its products become.
I want to get a motherboard with Intel onboard graphics (that has free Linux drivers). I've heard of the G965 chipset; is that the one to go for? I would prefer to buy a 'workstation' rather than 'consumer' motherboard but they tend not to have integrated graphics, no?
Are Intel's own-brand motherboards worth it? In the past I've bought Asus but that was for AMD-based systems.
-- Ed Avis ed@membled.com
...who can compete with ATI and Nvidia.
Intel has technology, has brains, has money, has plants. They can do something "as good as" the two others. Competition is a good thing (prices falling, etc); only two main actors for videocards is a bad things.
S3 can't compete. Matrox can't compete. 3dfx can't compete (they're dead). Others can't compete. Intel is our only hope.
-- Rastignac was here.
If you look at the vast majority of chips either ATI or nVidia sell, they're actually pretty efficient.
But they invariably _have_ to have some benchmark-breaking super-card to grab the headlines with. The way it works is that while only a minority of people will actually buy the top-end graphics card, there are millions of people who just need a reminder that "nVidia is fast" or "ATIs are fast". They'll go to some benchmark site to see some "nVidia's 8800 GTX is faster than ATI's X1900XTX!" article (not entirely unexpected, it's one generation ahead), end up with some vague "nVidia is faster than ATI" idea, then go buy a 5200. Which is the lowest end of two generations behind the ATI, or 3 behind that 8800 GTX.
Both ATI and nVidia even went through times of not even trying to produce or sell much of their headline-grabbing card. And at least ATI always introduces their latest technology in their mid-range cards first, and they tend to be reasonably energy efficient cards too. But it's like a chicken contest: the one who pulls out loses. The moment one of them gave up on having an ultra-high end card at all, the benchmark sites and willy-waver forums would proclaim "company X loses the high performance graphics battle!"
I don't think Intel will manage to restore sanity in that arena, sadly. Most likely Intel will end up playing the same game, with one overclocked noisy card to grab the headlines for their saner cards.
A polar bear is a cartesian bear after a coordinate transform.
If Intel can make a graphics card that is better than my current GeForce FX 5700LE in all areas (including shader performance) I am sold. :)
Especially if they have open source Linux drivers for the thing
Will Intel be clever enough and innovative enough to have a "GPU" socket on such motherboards? Maybe even GPU-specific memory sockets rather than shared memory?
One can always hope.
34486853790
Connection too slow for X forwarding? Try "ssh -CX user@host"
I doubt this will eliminate onboard graphics. At the low-end price range and in the light-weight mobile market, they're simply necessary. But if Intel could produce an onboard graphics chip that would compete with the 300-series (low-end discrete) from Nvidia and ATi, that could change the game.
It's also unlikely Intel boards would have a GPU slot that's not PCIe (or PCIe 2.0), since no one would buy a motherboard that locks them into only Intel. Even Crossfire/SLI boards allow you to have one of the other guy's card.
That socket is usually called a "PCIe slot" these days. If you use a socket instead of just integrating the graphics chip into one that is onboard anyway, you might as well use the established solution.
Another interesting approach (albeit not for high end machines and somewhat OT here) is AMD's plan to integrate the GPU with the CPU. That way, you might have some more choice than with a soldered in chip, and GPU cooling could profit from the availability of decent CPU coolers.
C - the footgun of programming languages
Intel has a chance. Intel has the experience with cpu's. Intel can also interface with their new processors. I think Intel could atleast put up a good fight. Why do you think Amd bought ati? They know that intel can do gpus and really good ones if they tried and the only way amd would be able to compete would be buying a gpu maker wich they did.
Don't forget that AMD was also talking about opening up the hypertransport spec and their cpu socket (from the 4x4 line?) so that you could use a cpu and another chip, either graphics or specialist, in a dual-cpu board. Would be interesting to see a socket 940 graphics processor, that's for sure.
I've never met an Intel graphics solution that could play anything more intense than Solitaire. Is the G965 any good?
I just upgraded my sister's mobo + CPU. It had embedded graphics, so I figured it would be comparable to her 2 year old nVidia AGP card. Nope. I had to buy a new PCIe nVidia card to handle Sims 2.
On a side note: Has anyone noticed that the extremely popular family-friendly 3D games are the worst performers? Sims 2 and RCT3 both take eons to load - much slower than Q4 or HL2.
"No matter where you go, there you are." -- Buckaroo Banzai
you missed the point entirely.
This doesn't take up one of your expansion slots, since you already have the graphic-out ports on the motherboard in such solutions. Meaning in a small-form-factor machine, you have one more option for tweaking the system to what you want/need.
34486853790
Connection too slow for X forwarding? Try "ssh -CX user@host"
I've been waiting for years for such kickass videocards. I've seen running protoypes in labs/universities; quite impressive videos. After a few years, now, the technology should be ready for the big market ? Pixar-like technology at home !
Real-time raytracing needs a lot of power; so, a multicore videocard is a great idea ! With raytracing, each core can compute one part of each picture. Better than SLI.
Using their knowledges, Intel can build a very fast multicore real-time raytracing videocard. It will be "something different", and it will compete with ATI and Nvidia in a new innovative way...
-- Rastignac was here.
I wish this one comes with complete documentation so I can fully use it when I make my own OS. I couldn't find one for G945. :(
It's performance is on a par with the IGP's about 3 iterations back from ATI- mostly due to immature drivers. It's closer in performance to the previous generation of integrated graphics (Which happens to be a chip from the previous era of GPUs with a vastly lower power consumption due to process shrink and logic improvements...)- some things it bests ATI's chips, other things ATI's chip with it's current drivers pastes it all over the place. The chip's capable of quite a bit more, but it's hampered by an immature driver (gee, this sounds familiar...) so it's a mixed bag right at the moment. The Larabee Group, depending on what they do, might actually give the other players a run for their money.
I am not merely a "consumer" or a "taxpayer". I am a Citizen of the State of Texas
Until this new hardware will let me display fractional polygons I'm sticking to my continuous graphics board.
Tsunami -- You can't bring a good wave down!
And you missed the point entirely as well.
Unless a specialized connector offers some particular advantage (e.g. like AGP did over PCI), it's best to stick to the general-purpose connector, due to manufacturer familiarity and economies of scale. Sure, you lose an expansion slot, but that slot has motherboard real estate available to it, because it's not being used up by some special graphics socket.
But will they include DVI? Better yet, dual DVI for those who run either dual monitors or really large monitors which require dual link?
The Christian Right is Neither (Christian nor right). See: Matthew 23, Matthew 25, Ezekiel 16:48-50
Intel's previous foray into the Discrete Graphics Market was the Intel i740. I got one, agreeing with PC salesman "Hey, you can't go wrong with Intel can you?" It was quite a decent chip for its time, and the driver was very stable. I don't ever recall graphics hanging once! It was disappointing when Intel bailed out of the 3D market, but to their credit they continued to update the drivers whenever a new version of DirectX rolled out.
Intel have already made a return of sorts to 3D with their Media Accelerator 9XX series chips you'll find in many Intel laptops. It's funny, because you'd expect an embedded chipset to be lame; lowest common denominator, shared RAM and akk. But this lappie has it and the graphics scream. It's faster than my nVidia 5700 which is two years old. The driver is stable too; never crashed. If they can do this with an embedded chipset 3d, imagine what they can do when they really put their mind to it?
nVidia and ATI have the market to themselves these days. nVidia has got pretty lax regarding driver stability for these days, and it's damned near impossible to get support out of them. They've fobbed off support to OEMs, who slap electronics onto cards and are in no position to help with driver problems. That's the sort of thing that happens when a company dominates a market.
If Intel can come out with some high performance electronics and stable drivers, well, Welcome back, Intel! I for one welcome you as my new Overlord!
Has anyone considered that the reason ATI/NVidia won't open source their drivers/firmware is because there are blatant copyright and patent violations in their code? I'm not saying there are violations, but if there are, then I would expect each to violently defend against anyone seeing their source code. To date, the best argument heard is that access to the code would provide their competitors an unfair advantage into their optimization techniques, which most of us recognize to be hog wash. At worst, they wrap it up in "we have licensed proprietary algorithms" declarations and refuse to give the community a chance to work around those algorithms.
There is only one way forward. NVidia should fund the effort to rewrite their firmware/drivers, providing only the hardware register descriptions and nuances. I'm quite sure others have asked NVidia to do this already, but Intel moving forward with this plan should force the other's hand. I'm surprised that Microsoft hasn't chimed in here because for every open specification we get in the OSS world, they also get. That's where all those Microsoft drivers come from. And only on occasion is a vendor-supplied driver better that the Microsoft one. Open sourcing any drivers also helps Microsoft support more hardware out of the box, without a multitude of licensing agreements and royalty schemes.
And of course, NVidia (and now ATI) have been adding more treasure to their war chests with the PCIe motherboards. I just bought a new motherboard and it's extremely hard to find a new board with PCI-Express that doesn't have an nForce or ATI chipset.
It's going to be a tough game for Intel because it's not just graphics drivers. AMD could play into this game if they took a decisive maneuver with their GPU integration into the CPU. Remember that AMD now owns ATI.
But nVidia bought 3dfx, not ati...
XML is like violence. If it doesn't solve the problem, use more.
When you write a patent you have to describe exactly how it works (well here in the UK anyway). That makes all patents open source, so if I can just look up the S3TC patent and get the algorithm why would it force ATI/NVidia to close source their drivers? That doesn't make sense.
thank God the internet isn't a human right.
This would be nice in a laptop. Nvidia's effort to standardize MXM (Mobile PCI Express Module) hasn't been widely received by ODM's, or it was dropped, or it wasn't properly adhered to, etc... Having a solution in the iMac, Mini, or small form factor would make sense. I think engineers could design a board with this kind of integration without increasing its 'footprint' so to speak.
If you lend someone $20 and never see that person again, it was probably worth it.
I've been monitoring this thread with some interest. I'm looking to build a new home computer that will run Linux exclusively (most likely Kubuntu). Mostly, it will be my personal workstation but I do plan to install some games - mostly 1st person shooter types. While I don't require "cutting edge", I would like decent performance. Can this chipset handle things like the latest UT or Doom III on Linux?
I mean, I like nVidia, but if Intel is supported out-of-the-box with open source drivers, then that works for me as well.
A goal is a dream with a deadline
Considering most motherboards are sold already in PCs I doubt most people even know or care that they are locked into Intel only and anyway motherboards don't lock you in that much, you can just get a new motherboard without worrying too much. Intel boards don't allow you to plug in an AMD processor chip after all and I dont hear many people complaining about being locked into a processor manufacturer. The performance, efficiency and manufacturing gains might well be worth producing powerful on board graphics, the wheel of reincarnation revolves onwards.
If this were really happening, what would you think?
I believe in competition being good, but I'm not sure it's all about just competition. This likely could be the move to save PC Gaming as a whole. Technology-wise PC's will always have superiority over consoles, but there are rare arguments to the economics of top-end gaming PC's against consoles. Microsoft and Sony take huge losses to push their hardware, and slowly but surely it does pay off - Gears of War on the Xbox 360 has sold 3 million copies in just a hair over 60 days. Name one PC title that is using every bleeding-edge technology and has sold that many copies that fast. You won't find it, because the segment of people who will pay between $2500 and $5000 for a PC to play those kinds of games (Crysis, Oblivion) is so small you can't hope to sell that many copies. Intel knows how to make computer chips quickly, and on the cheap. That is what I feel they are bringing to this contest. I think Intel believes they can make a graphics platform just as powerful or more powerful than Nvidia/ATI and can do it for less cost. That is how you generate competition not just in the graphics sector, but you make PC's more competitive against the consoles. The PC has endless amounts of good games to sell, the problem is there aren't cheap PCs that will play them with the slickness that consoles provide. Ultimately this move should make the top-end PC cheaper, which is good for everyone because the inherent competition will force Nvidia/ATI to lower prices. I like this move. Go Intel!
Even Intel doesnt get it perfect in the fist chips.
Another considerationis they may have to emulate someone elses API. Too much software out there to have a new one.
Muppet
So, what we really have is one person saying, "it's better than my nVidia 5700" in a non-troll-like manner vs. you saying "you're full of shit" in a...well..toll-like manner.
So, we have two opposing opinions, one with only anecdotal evidence and one with....bupkis.
Why don't you do the "uncritical and accepting" masses on Slashdot a favor and point us to some data we can use?
In fact, maybe you should, otherwise, you might start to look like the one who's "full of shit".
A goal is a dream with a deadline
The i740 was a very decent for corporate use. It was inexpensive, had stable drivers for both Windows *and* Linux, and had decent performance for everyday tasks. It worked well without muss or fuss. I would definitely call that decent.
It wasn't the greatest gamer's chip ever made for sure, but most tasks on a computer aren't games, especially in a business.
Patents would not require the code to be closed source.
You are probably confusing patents with copyrights on the submitted code.
The only way patents make code closed is when a company thinks they may be violating a patent and wants to hide it.
A HTX slot video card is better.
I meant it's marketting/advertising/PR, by any other name. It's not that everyone uses _only_ those 8800 GTX benchmarks to choose a lower end card, it's that it's there to bombard you with "company X is better/cooler/higher-tech/whatever than company Y" until it hopefully starts to create a subconscious bias. It's not the only criterion, but for enough people it ends up being _a_ criterion whether they acknowledge it or not.
Sure, we all like to pretend that we're, like, all intelligent and stuff, and would never let advertising rule our lives. In practice, we already know that marketting works.
See for example Coca Cola's "New Coke" disaster for the prime example of how much of a bias marketting can produce. It surprised even Coca Cola.
You can read a more detailed account on Snopes, for example, but basically it goes like this: in double-blind taste tests, not knowing which is which, the vast majority of people preferred the New Coke taste. It would stand to reason, then, that they'd go buy the drink whose taste they prefer, right? Well, wrong, as Coca Cola discovered. When they _did_ know which is which, the majority demanded the old Coke back. A helluva lot decided that they hate the New Coke without even tasting it, just because it wasn't the product that ads had told them to prefer, year after year.
So, well, maybe you are one of the select few who genuinely aren't swayed by marketting. Kudos and more power to you, then. But in more average people marketting can and does create pretty strong biases.
And benchmarketting is just one way to fight that war for mindshare, really.
A polar bear is a cartesian bear after a coordinate transform.
The Cell CPU of the Playstation 3 will change that. This Intel announcement is basically mimicking what the Cell CPU does, only in a much much slower way (going over ANY peripheral bus will be much slower than the on-die method used by the Cell). Nonetheless, the games that will be coming out over the next few years for the PS3 will demonstrate exactly what the possibilities of multi-core distributed processing can achieve. Can you say "Gran Tourismo Raytraced"?
"I could agree with you, and concede the point, but I consider using words which mean the opposite of what you are trying to say in normal conversation to be extremely silly.".
Oh yeah, that makes sense.
Or even bettter a mother bord with several HTX connectors, in which user is free to put whatever he want, eihter "slotckets'"-like cpu board (Cpu board with a AM2/3 socket and DDR2/3 memmory socket), or even low-cost all-in-one CPU modules ("HTX module with Sempron64 and 1GO DDR2 already soldered") or high speed graphic-cards.
Or high speed task-specific coprocessors like massive vector engine for physics (Ageia's PhysX) or like fast string pattern matching for anti-virus and spam filters (Like the boards that ClamAV supports) or Cells.
Or volume-processing graphic cards (used for display 3D medical image like MRIs)
offers a lot of possibility, without needing to put a large collections PCIe slots, each with it's own chipset controlling it (like on some current 2x PCIe16 board with two northbridges, each controlling it's PCIe 16 line).
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
And I described an importance, a big one in fact. It's the same importance that has people making new mobo form factors... You missed the point, entirely.
34486853790
Connection too slow for X forwarding? Try "ssh -CX user@host"
I'll welcome Intel back the day they add hardware transform and lighting to their graphics line. Seriously folks, vertex processing has been on NVIDIA and ATI GPUs since the Geforce 256, yet Intel is still running *ALL* vertex processing in software.
Perhaps the addition of a discrete line of cards will force them to reconsider hardware T&L for their integrated solutions. Once the Media Accelerator 9XX (or its successor) include hardware vertex processing, my interest will be piqued.
I work as a tech at a place I will call OverpricedRetailComputerStoreUSA. Before this I was working at their competitor across the street. I know I personally have done a lot of video card installs over the years, and two recurring trends I've noticed:
"I'm a Laver, not a Phyto[plankton]"
If nothing else the excitement in and around the Linux community over these OpenGL possibilities on the desktop has brought forth a need for better drivers and acceleration. Nvidia has the most feature complete drivers on the Linux desktop, but they are not free in the licensed sense and therefore have to be installed by the user (bad in many ways). Ati's official drivers are just as closed with worse performance and features. For ATI cards there is a reverse engineered free driver that has more features than the official drivers but still lacks the performance of an official driver. Its a great effort and by far one of the best things so far, but it does not work with the newest Ati cards- meaning that Linux is always one step behind. This leaves a vacuum in the market for a discrete card maker to put out hardware that is nearly feature complete with open drivers before or on the day it's released to the public.
Intel so far looks to be that company, and now should be considered the best hardware friend so far to the future of the Linux desktop. Give credit where credit is due and buy their stuff and mail them and tell them you did. This is the best shot so far....
Open Source Sushi
The GMA950 is a crap 3D card. Even the most basic google research shows that it is NOT a return of Intel to 3D and no reviewer worth a dam has said the graphics "scream". Poor performance and incomplete 3d support are the hallmarks of the GMA950. If you play nothing but Quake II than yea, the GMA950 is for you.
4 ,00.asp7 &p=3- gma-950-terrible-opengl.html
http://www.extremetech.com/article2/0,1697,182181
http://www.anandtech.com/video/showdoc.aspx?i=242
http://everythingapple.blogspot.com/2006/03/intel
If you wanna get rich, you know that payback is a bitch
YU0 FAIL IT!
You're overlooking that the socket for a graphics processor will require room on the board and the socket itself will be a part that costs money. If the graphics processor is a high end model that draws a lot of power, you will need a voltage converter and cooling solution similar to what the CPU needs.
Overall I see this causing even more problems than using a PCIe socket and a graphics card that carries a voltage converter/cooling solution that are adapted to the needs of the processor on the card... You still missed the point, entirely.
C - the footgun of programming languages
None of those things make it impossible, and none of them make it require a significantly larger form factor than micro-atx/dtx/itx board (except the dedicated video memory slots, which could be optional in smaller boards). So, no I didn't miss the point your trying to make. I'm just suggesting that some people have different priorities. I would love to have a board like the one I described.
34486853790
Connection too slow for X forwarding? Try "ssh -CX user@host"