ATI Mobility HD 3870 Coming in March

AMD's latest roadmap (from late January) shows the mobility HD 3870 to be released in late Q1:



http://crave.cnet.com/8301-1_105-9871416-1.html

 

hmmm, can't wait for the "puma" launch. Somehow I doubt it will be double the graphics performance of the previous generation.

 

Nice find! Bet HP will put this in the HDX.

 

Nice, I hope it is MXM-2able.

What's Puma anyway? A new IGP solution?

EDIT: It's the new platform for the Turion Ultra. Features DDR2 800 and HT 3.0

 

i agree....it will not double...want to team up and sue for false advertising?

 

Yep.....I agree that HP will most likely incorporate this chip in the HDX "Dragon" especially for the Montevina platform.

Good looking out Yokozuna and thanks.

 

Maybe he though it was 3870x2 they are releasing for laptops Hope to see some benchmarks soon on this one and it would be interesting to see if they can beat the 8800m GTX.

 

Actually, I'd expect it to. The 8800M GTX is basically a hamstrung G92. The RV670 is more efficient and draws less power than the G92, which throws heat like no one's business.

Because the die of the RV670 is actually comparably small, I'd expect it to materialize with all of its shaders intact.

 

*Yawn* Too little too late again for ATI/AMD.
Atleast thanks to them, nvidia continues to release new hardware, otherwise we'd be stuck with the geforce 7x series still...

 

Well Pulp if that is the case that is only good for us, though I prefer Nvidia´s hardware.

 

I wonder if this is worth waiting for? I just want something thats faster than 8600m GT and fits into 15.4". Will this chip even fit into that? What do you think?

 

Hmm my theory is that these goes into 17" but since the 8800m GTX is in a Aleinware 15"4 and these cards probably runs cooler then why not?

 

I'm looking for something better which fits into a 14.1".

 

Sweet. Looking forward to the 8800M GTS/GTX and Mobility 3870 comparisons.

 

Absolutely. Shaders for the 3600 and 3400 mobilities are the same as their desktop counter parts. Where as with the 8800M GTS it's only 64 and with the 8800M GTX, 96. Their desktop equivalents have 96/128 and 128 shaders respectively. And higher clocks and larger memory interfaces to boot.

What's more, the RV670 is DX10.1, shader 4.1, and PCIe 2.0 compliant. Nvidia doesn't have anything in the immediate pipeline that will be. Also, as a definite plus, the 3870 brings with it a 256 bit memory bus.

This will be the top notebook card in the single GPU category. When the R700's for mobile debut perhaps later this year, they will answer SLI with crossfireX.

All good and healthy competition.

 

If its even more efficient than the 8800GTX, maybe we'll see this one in 12" notebooks!

 

This sounds promising! I hope they release some new notebooks with this chip soon!

 

Probably with Montevina launch. We might see someone at CeBit early next month debut a working prototype lappy with it. Maybe Clevo. One thing I'm fairly certain of is HP using it in the next iteration of the HDX - assuming that they will continue the HDX through Intel's Montevina launch.

I wouldn't be surprised if Dell looked to do something 17" with it.

 

Okay, so probally not anything to count for as 15.4" then.

 

Don't know. There was some news a week or two ago that Dell and HP are committed to using ATI cards in their lappies. The ATI cards have better thermal and power effeciency properties than Nvidia's comparable cards. It may be possible to get one into that form factor. If not, then they'll default to the 3400 and 3600 Radeons.

You can check the specs of those two on ATI's website.

Big reason Dell and HP want to use the ATI cards is not just because of the better thermal and power properties, not to mention better pricing which gives them better profit margins, but because thanks to Vista SP1, they don't
want to slap Nvidia's non-DX10.1 compliant (read obsolete) cards in 'em.

 

Mod this into the Asus EEE..lol

 

.and then use Crossfirex and mod 2 one

 

If Pulp is right then I'll take the 3780 over the 8800 even if the 8800 is faster , ofcourse assuming it will provide a better value for the $ .

Problem is AMD/ATI are known lately to release more press then actual hardware .

 

i think u have a miss understanding of what DX10.1 and SM4.1 actually entail..

10.1 is a hardware block to let new games/software force 4xAA to always be on kind of like OSX(a software setting will be used for DX10 cards in SP1 vista.. thus making it the same performance impact but making the non 10.1 ones be able to turn it of via drivers)

SM4.1 is a small adjustment to the Shader model however its considered software based and ATI have just slapped it on along with 10.1 .. Nvidias shader clocks massively out perform ATi's atm.. regardless of what model is used. 3,4

Its not like having to set 4xAA manually is going to effect performance hell it makes the non 10.1 cards at a advantage..


be really nice to see a 3870 running in a macbook pro but... 8600M is getting a tad old.. but honestly in all the Cases the 8800M GTX are in the gfx card hardly runs at high temps even in Sli in the Dell 1730 and has lots of overclocking room.

also the 3850 and 3870 in desktops are horrible over clockers... so i doubt any miracles in the laptops

But most importantly you can not compare ATI temps to nvidia temps.. because Nvidia cores can simply run at MUCH high temps stable ATM due to their cores quality and long/not rushed development.

the amount of G80's and G92 cores in gamers PC's atm thanks to the massive amount of games released at the end of last year has pretty much made them the new standard... its the 9800pro of 2004 in 2008.

however looking at the 9300m and the 9500m is think its pretty clear they are scaling back the 9series massively due to no competition from AMD/ATI.. and i'm sure the next generation of high end ATI cores will prob have to be released before Nvidia respond with any GT200 stuff.

for high end gaming you going to have to wait till the r700 or GT200 get cooled down for laptop use... which.. will be a long long time (they have temp Health warnings lawl)

 

Actually in almost every review the 8800GT smokes the 3870 by 20-50% depending on the game. This is hardly comparable.
The chance that the mobile 3870 will overtake the mobile 8800 GTX is very unlikely.

 

Desktop 8800GT specs are not quite the same as mobile 8800M GTX

8800 GT (Desktop)
112 streamers, 600mhz core, 1500mhz shader ,900mhz mem,256bit

8800M GTX (mobile)
96 streamers, 500mhz core, 1200mhz shader,800mhz mem,256 bit

Looks closer to the ...
8800 GTS 640 (desktop)
96 streamers, 500mhz core, 1200mhz shader, 800mhz mem, 320bit.

Radeon 3870 (desktop) beats or matches the 8800GTS 640 (desktop) in most reviews depending on game and drivers.

And since the ATI 3870 is a 55nm die, I wouldn't think that there would be all that much stripped from it to make it notebook friendly. It should pretty much make it into the notebook "intact." Not only that but should Montevina for notebooks use the PCIe2.0 bus, the ATI 3870 Mobility would be able to take advantage of it while the 8800M GTX would not.

 

and at the same time due to it being rather new along with the new platform (that probably should land on the shelves near half the summer) the drivers might ruin a lot not to mention the $$$ you pay for the newest and latest gpu... but let's not be hasty - why can't we be offered with them already i'd love to see the tests

 

I think YOU misunderstand what DX10.1 entails.

D3D 10.1 DOES NOT FORCE AA TO ALWAYS BE ON !

It forces IHVs to ensure that their hardware supports at least 4X programmable AA as an OPTION available to the application to guarantee a minimum level of support. And if the application is DX10.1 only then you will not be able to run in with DX10 hardware (other than hack pseudo support).

Got something to back that up, especially the slapped it on part?
Otherise it just looks like a reactionary Fanb0i statement.

Why don't you explain to the class why features were dropped from DX10 requirement and made optional and how that was because it couldn't be 'slapped on' by one IHV?
It is a small change, but all of the changes are beneficial changes like Cube Map Arrays for global illumination.
As for the 'shader clocks' yes they are faster, but they are also out-numbered by the AMD hardware, so it often balances out, and it depends on the specific shaders that are run as to which is more performant. The peak pixel fillrate on the HD3850 is higher than that of the GF8800GT and GTS-512, with the HD3870 being closer to the level of shader power of the Ultra than the GTS is to it.
If they 'massively outperform' then why is it not the shader intensive situation front-end situations where they differ so much, but instead the back-end focused texture and AA situations where the GF8 outperforms the HD series?
If nVidia 'massively outperforms' then why don't they keep pace when it's shader based AA instead of the back-end oriented AA where only they have hardware resolve in the ROPs?

A better view of the balance of the shader power between both can be seen from The Tech Report, as you cans see they are both powerful but in different situations (the common occurrence of which can be debated);
http://techreport.com/articles.x/12458/3

Saying either is 'massively' better would be like saying 6"3' is massively bigger than 6"2' or $255 is massively more expensive than $250.

No it doesn't, especially since those that do it manually aren't designed towards shader based AA and as such don't perform when it's called for (as is the option in DX10) in applications like Call of Juarez.

As for the rest of your junk, seriously, could you be more one sided? FUD much?

I don't think DX10.1 will make much difference to people's choices, not because it doesn't add anything, but because it won't have much support in many apps for a while, other than Futuremark's Vantage (which will sell a ton of cards to n00b knobs who give it credibility) for a long while. However, I always find it entertaining to see the instant reaction though to DX10.1 that people like yourself think is so important to come out against it even posting incorrect information about D3D just to try and make it sounds as if it's a disadvantage to be both DX10.0 and DX10.1 compliant.

 

Yeah, Nvidia's shader processors have an approximately 3:1 advantage over ATI's. That's why ATI usually has 3 times as many shaders just to keep up. The 3870 has 320 shaders which is roughly equivalent to 106 Nvidia shaders. Give or take.

On DX10.1, not sure what you think I misunderstood. Nvidia plays it down, naturally, since they haven't any wares that are ready-set-go. While ATI plays it up because they do. Happy medium is, potentiality.

In any event, think like a manufacturer. Would you want to sell DX10.0 cards when Vista SP1 has DX10.1? Not likely. That's partly why Dell and HP have committed themselves to ATI cards in their next iteration of their lines to coincide with SP1 and Montevina launch.

See here:

http://tailrank.com/4932095/AMD-to-grab-over-50-discrete-notebook-GPU-market-share-in-2Q08

Actually, temps and power usage are important when it comes to notebooks. And you can and should compare this metric. The trend is towards sleeker, more power effecient notebooks that run quieter. Big bulky gaming rigs like the XPS 1730, Clevo MR570U, Clevo 9xxx and Alienware's latest 15 and 17" are geared towards the gamers who don't mind sacrificing style, loudish fan noise, and who throw sensible economics out the window.

Yeah, the 9300M and 9500M seem to be Nvidia's new naming convention. They's both just 8400M and 8600M's respectively. But high end gaming is already here. 8800M GTS/GTX's fit the bill. They only need some healthy competition. ATI should finally break through that ivory tower NVidia's been enjoying in the notebook segment- yeah, the one where they sit back and smile over the $1000 price premium Dell charges for Dual 88's in SLI.

Let's face another fact - the desktop is going the way of the dodo. Notebook sales growth largely outstripped desktop sales last year and will continue to do so. The average lappy is about as powerful and the average desktop and is getting closer to price parity. (High end excluded.) Game developers are becoming more aware of this new paradigm and will begin to design their games for the installed base of the majority - which will be notebooks. (Crytek's Crysis excluded.)

Here's the rub - most notebooks have integrated graphics - mostly because of Nvidia's price premium for discrete doesn't leave maufacturers all that much wiggle room on margin at the low to mid spec segment. With Intel's new integrated graphics set to launch with Montevina and the reasonable priced ATI's 3400/3600 Mobility discrete options, that installed base's anti just got upped. Upped enough that game developers won't have to dumb down their code to make their games run on these systems at acceptable resolutions/options and frame rates.

 

aww shucks Fanboi and FUD?
i'm a jobless student no magical insider knowledge from me but from all the Microsoft press releases it seems what your saying is wrong ..

"According to Microsoft’s Sam Glassenberg, DirectX 10.1 will be fully compatible with all graphics cards supporting DirectX 10. He told that the current updates are very similar to those performed for DirectX 9 back in the days."
http://www.techspot.com/news/26640-microsoft-directx-101-fully-supports-directx-10-hardware.html

are the internet's lieing to me is so im sorry for my level of sureness but i've read the above imformation more than once?

this hole "ATi are going to be better for DX10/10.1 and nvidia are best for DX9 stuff and wont be able to run DX10/10. is lies" is pritty fanboi if u ask me.. both cards will prob end up bench close and 3870 will def be able to be put in smaller and thin laptops.

case closed till bench?

 

That's called backward compatibility. But DX10 cards will not fully support DX10.1 games as they haven't the hardware to do so.

If you quote the next few sentences in the article you linked, you'll see it says that.

"However, Sam confirmed that existing graphics cards may still not be able to use all the new features of DirectX 10.1."

 

The only disclosed exclusive feature is the option for programmers to use forced 4FSAA i don't see how that is remotely appealing let alone needed or not needed or not already possible.. all driver revisions already have a toggle for applications controlled FSAA levels.

the fact is it does not lock people out like SM3... and people are treating it like it is. sure we we need a more aggressive market but lieing to current or future nvidia users suggesting being shader model/Dx10.1 obsolete is simply a total fabrication based on current information. agree?

 

Wait a second. Is that how you think it works? I think you might want to read up on how DX10.1 works and what it actually adds. It adds a bit more than that.

Here's a fun thing we can do. ATI released a DX10.1 whitepaper. I have it as a PDF file. You can google"ATI dx10.1 whitepaper" to find it. We can critique it and see what's hype and what's meaningful. This way at least we're on the same page (no pun intended.)

 

Actually if you read it it doesn't contradict anything I said. What you're quoting is the reaction to people's "the Sky is falling my hardware won't work" fears which were fostered by the mistatements of people, kinda like the whole "DX10.1 makes it run 4XAA all the time".

It is incremental, just like SM3.0 was incremental over SM2.0, but there are benefits to both. Memory virtualization should actually be great for laptops moreso than desktops IMO.

If it's the same places you're getting that junk about overclocking and overheating, then maybe the internet is lying to you, or it's just the voices in your head. I suggest you read harder Homer, and try starting at MSDN, not some tabloid tabloid site.

Depends on what statements you're going off of, if it weren't for M$' optional change in DX10, then the GF8 series would still be DX9+, but overall regardless of DX9 or DX10.0 both will be very capable, but when it comes to DX10.1 there is only one solution. The main thing is there's nothing DX10.1 worth talking about right now, and the future may or may not change that. The main thing is for people to focus on performance in games once the hardware actually arrives because that's when it will matter.

You can easily stick to the facts and show that DX10.1 isn't that important, no need to going into all that other stuff.

 

Once again, read harder homer. Focus on shader based AA (easily found by looking for pixel-correct or HDR-correct AA), and then figure out why devs want control.

Sure it does, just not yet.
The first SM3.0 games like FartCry didn't lock out the other cards, it was later. The question is will we ever see a DX10.1 minimum spec game? Probably not for a while, just like it took quite some time for SM3.0 games to come out.

People are confusing, minimum requirements with useful feature.

It won't be a minimum for a while, but may be useful as long at TWIMTBP program doesn't get too involved in that decision making process. And that has more to do with the install base you talk about.

It's in nV's best interest to resist development of DX10.1, and in all likelyhood will see DX11 being the support level of the T200, not DX10.1 to avoid any focus on it, thus making it also like DX8.1/PS1.4 (GF3/4 vs R88500) which was limited in focus, but did show up as a nice feature in games like Morrowind, and as a minimum spec in games like BF2.

When this becomes a 'limiting' factor is long after it may become useful.

 

This is incorrect. Comparing shaders between the two vendors is apples-and-oranges; ATI's architecture is VERY different from nVidia's. I'm not even sure exactly how to count ATI's; they count them as "stream processors" but it's my understanding they only have something like 64 fully unified shaders (divide the 320 by 5, and I fully admit I could be wrong on this detail), with the remaining shaders in each cluster being somewhat more specialized. It's a really weird architecture, but this comparison isn't worth making.

While I agree on your first point - of course nVidia's going to downlplay DX10.1 when they have no cards capable of it - I think your second point has a hell of a lot more to do with the fact that ATI parts are cheaper and draw less power than DX10.1 support. Vendors don't really care about screwing us out of features, they care about their bottom line. ATI gives the best price/performance for them.

This is absolutely true. The RV670 die is worlds smaller than the G92; in addition to being produced on a smaller process, it also has almost 100 million LESS transistors. It draws less power and produces less heat than the G92, which had to be cut down for the mobile market, so it's probably a safe bet we'll see the full RV670 in mobile form instead of a cut down part.

While in the desktop market ATI is still fighting an uphill battle with nVidia, due largely to the failure of the HD2900XT, the mobile market looks like it's gonna get a lot more interesting, where efficiency and power consumption are king.

This is patently false, and I'm disappointed you'd say this since you've oftentimes been on the mark about other things. Notebooks will NEVER obsolete desktops. The amount of power in my custom desktop is all but impossible to duplicate in a notebook right now, and when it IS duplicatable, I'll probably be driving a new desktop with faster hardware still.

Notebooks will always have to worry about power consumption and thermal constraints that desktops don't have to deal with. Notebook sales have been outpacing desktops because notebooks have been a lot cheaper lately, and because some users may not necessarily have the space to devote to a desktop.

But notebooks are NOT going to obsolete desktops.

They never dumbed down their stuff to run on Intel's IGPs to begin with, so your argument isn't sound. And it's not like gaming technology just sits still; while IGP performance (outside of Intel's parts) has been ramping a bit faster than the technology curve as of late, it's still been so far behind that it'll be some time yet before an IGP can truly be considered the minimum of a modern game.

 

But just dividing by 5 isn't accurate either, it's more like 256+64 because it's a Vec4 + 1 design, which isn't really weird it's closer to the way things have always been done in the previous vertex shaders. Check the TechReport link and it will give you some insight, but Beyond3D probably has the most detailed 'layman' explanation if you want to go more granular;
http://www.beyond3d.com/content/reviews/16/5

Which is nice after nVidia completely flipping the mobility dominance almost overnight during the GF6 vs Radeon X generation. A return closer to parity would be nice to get some R&D working again.

Yes they will, along with game consoles, WHEN is the only question. And then the big game consoles will be replaced with smaller portable game consoles, it's the way of the world. How many arcades do you still see out there?

You seem to miss that high end gaming rigs are a very small part of the overall market, and while they may even eventually grow to a larger part of the desktop market that's because the overall desktop market is dissapearing. As that occurs and more iMACs and ONEs are produced and counted as part of those few remaining 'desktops' and not the true stationary laptops that they are, you will see less and less products being developed for the desktops because the market will be too small to support it from just desktop gamers alone. It's not going to happen immediately but it will happen sooner rather than later just like LCDs replacing CRTs it was slow at first, but tell me where you can buy a CRT in standard retail. It was still better technology until LEDlit LCDs and HDR LCDs starting maing an argument, but despite that they were going going gone. With laptops expected to be more than 75% of purchases by 2010, they will follow the same path where even with a large minority the development motive won't be there. Unfortunately the by this time (likely shortly after 2010) you will probably see some killer console replacement for our current gen and voila, the death of big desktop gaming. however expect the consoles to continue to become more desktop like though too so that at least the software components to modding likely remain, but the hardware upgradability will likely dissapear. And no matter how passionate the desktop crowd, it's eventually gonna go, just like CRTs.

As for the minimum spec issue at the end there, expect Fusion and Larrabee to change how with things of that as the usless GMA and X1150/GF7150 market simply because 'processing power'.

I don't see anyone predicting a long life for large desktops.

 

If nothing else, I would think desktops would remain for ergonomic factors. While personal users may be purchasing more laptops, that doesn't mean they are replacing their desktops with laptops as their primary computers. I think it's more a matter of convenience and maybe for kids. A laptop is still much more awkward to use for extended durations of time than a desktop with external peripherals. Even if you purchase separate keyboard, mice, etc., you would be taking up the precious few ports the laptop already has and introducing enough clutter to obsolete the fact that it is a laptop and meant to be portable.

I personally have a laptop as my main machine as I need to be mobile, but I find it tiring after a while to have constantly be hunched over while typing because my keyboard is so close to the monitor.

In the end, this is all just speculation anyway. We'll find out what the dominant type of computer will be soon enough.

 

It is apples to oranges. But as I stated, a rough estimate comparison of the current Radeon 3 series as well as 2 series to Nvidia's 8 series is more or less good enough. Factor in 128 bit and 256 bit bus for both cards and you can estimate the performance potential.

Sure, of course price plays into it as that helps to fatten their amenic margins. But consider that these things are selling points to manufacturers. Just like Vista is a selling point and why the big 4 kit makers sell Vista. They could of course offer XP along side but they don't for the most part. In the technology business, as you already know, consumers are conditioned to desire the latest and greatest. No matter how incrementally small it might be.

Just as when Nvidia was first to market with a DX10 card. They used that as a selling point. I only know of one manufacturer who used the Radeon Mobility HD 2600XT. The rest hoped on Nvidia's 8 series DX10 train.

Well, I've got a lot of sales stats and metrics that suggest otherwise. Not that notebooks will obsolete desktops but that desktops will be relagated to a specialized niche and will cease to be the mainstream computing device.

The form factor of choice is increasingly the notebook. Second up, the all-in-one popularized by Apple's iMAC. Then comes the desktop.

Take a look at the Dell, HP, Gateway, etc and compare the specs at the average consumer level and price and you will see that the notebook is fast catching up. Businesses are buying more notebooks than desktops. Unless you categorize thin client systems as true desktops. Then it's about evenly split.

That's my point. That they haven't really considered IGP when coding because of the paltry specs. Older games run fine. But as IGP becomes more viable, which it will begining with Montevina, it will be on the consideration list. I wouldn't be surprised to see on the bottom of a game box's requirements, Intel's IGP listed as a minimum. Perhaps not in the recommend side though. Too early for that.

When most developers (Crytek excluded, ha ha) code games, they look at the installed base of systems going back 1 to 2 years, considering what's available currently, and in many cases considering what'll be available in 6 months, % of demographic that plays games on those systems, and code accordingly. The goal being to attract as wide an audience as possible per genre.

 

While "NEVER" may have been too strong a word, the comparisons you're drawing don't apply. Why doesn't anyone ever think of practicality when they think about these things? Notebooks won't fully supplant desktops because desktops will ALWAYS be more powerful by virtue of the aforementioned power and thermal constraints.

Gaming consoles, for example, are starting to transform into more broad-based entertainment platforms (PS3 as Blu-Ray player, for example). While maybe at some point they could theoretically entirely supplant PCs as a gaming platform, I remain skeptical by virtue of the expandability and versatility of PCs.

LCDs obsoleted CRTs because they're largely a directly superior technology occupying the same market space.

Home consoles are obsoleting arcades because it's more convenient to game at home and because console hardware has largely eclipsed arcade hardware (due to the fact that the home console market is simply more profitable than the dying arcade market).

But notebooks and desktops are unique because they simultaneously do and do not occupy the same market space. Desktops are more ergonomic, more upgradeable, and more powerful than notebooks.

I hate having this stupid argument because it always comes up on these boards and the same conclusion is always drawn: notebooks will not obsolete desktops.

 

GreatG****Ape then take a look at how many cards, i.e high end 8800 cards Nvidia has sold so far and you´ll see how many people high end mainstream gamers that bought those cards. The desktops are not anywhere near their end. It is on the desktop side all the progress is taking place and will continue to do so for years. The notebooks, while efficient, does not have sufficient cooling for a really high end GPU and I don´t see this change anytime soon. That´s why I own both a high end notebook and a high end desktop, since I know I have to buy a new laptop when it´s near it´s end, meanwhile for my desktop I can just switch parts out really easily

Now if the notebooks gets upgradeable easy, just like the desktops, well then yeah they´ll get replaced. Still the thermal solutions isn´t ideal for notebooks, like for the desktops. The development of new hardware will always take place on the desktop side, where heat isn´t such a big issue. Not as closed environment as a laptop chassi has.

 

You're not looking at this on the whole. The 8800 sales is what defines a niche. Do you think that they are selling millions of them? No way.

Compare that to sales of notebooks in the tens of millions per quarter among all manufactures combined. What's more, the average desktop sold today is sub $800US configured. Not only that, but retailers of the brick and mortar variety have steadily diminished floor space dedicated to desktops in favor of notebooks. By the end of this year, you may not even see a tradition bulky box desktop on the floor of any major retailer. You'll see all-in-ones and notebooks. Just go take a peak at your local Best Buy, Circuit City, or whatever local major electronics retailer you have near you. Plus note the price and configuration of teh desktops offered in stores.

Your XPS's with 750watt to 1kilowatt power supplies selling @ over $5000 aren't selling hand over fist.

 

I think you're considering this in terms of a current hardware constraints. My current laptop beats my previous dual AthlonMP editing rig by a very very large margin. Soon you will reach the point at which the average user cannot get diminishing returns on the need for that power. If right now you could put an octo core skulltrail platform with a GF8800Ultra inside it in a mid-level mid-size laptop, what percentage of the market wouldn't have more than enough power that they need?

I'm not saying that wouldn't still be a role for ultra-desktops, the only problem is that niche become so small that it simply isn't focused up by AMD, intel and nVidia. This may not happen until 2012 or later, but it is inevitable.

I don't disagree that to us the preference is PCs, however think of it from the standpoint of both hardware vendors and developers. Once the customer is far enough into the mobile and small/efficient market, they are going to push gamers whether they want it or not.

No they aren't superior, especially not for gaming, you obviously missed that point orignally.

Replace every 'arcade' there with 'desktop' , and voila the future after whichever console update that finally kills desktop gaming. Eventually it will die at the hands of dedicated consoles, and then those who have the more flexible but definitely more portable PCs.

Not quite.
you also miss the idea that an iMac isn't the market you're thinking of either. It a laptop on a stick.

Because that's a fact, and eventually pocket computing will obsolete notebooks and then something will obsolete that. Anyone who thinks otherwise just can't think beyond the current limits of technology and it's aplication.

 

Actually the charts I saw had almost a million high end cards sold by Nvidia. I´ll see if I can find that chart again. True though notebooks sells like ever before, I have ordered one new myself Actually I prefer to play on laptops anyway, I really like them and with the current one on order my XPS M1730 with 8800m GTX SLI I´m good to go for a few years, well maybe 3 at least. S

 

Take a look at the sale of cards that cannot be replaced by desktops, and then think of what fraction of the market that is. The $250 and above market is dwarfed by the other two segments. And as the laptops beef up their lower limit capabilities more people will move there.

You can argue the theory, but you can't argue the numbers;
http://www.businesswire.com/portal/...d=news_view&newsId=20080131005336&newsLang=en

And if AMD takes back the lead in notebooks over nVidia do you think that was with the bes offering of the HD2600XT or more likely the X1300-HD2400 lineup? And was does that say about the nVidia offering which were pretty much the only high end you could buy other than the rare than plutonium Mobility X1900s?

Like I said, your head spaces are still to 'current technology' which limits you to such traditional thinking, which is not going to be what's available when I expect the shift to take place (llike I said 2010-2012 most likely).
Perfect this nice gadget, and your limitation is removed;
http://www.hexus.net/content/item.php?item=7995

Like I said think more modular when it comes to laptops in the future, and don't think about limits so much as problems that haven't been solved yet with current technology.

However you idea of what constitutes a laptop, and a desktop and desktop space may change to something more docking station like. Regardless of the niche markets like gamers, if the large majority of the market goes in a direction that is completely opposite, then the technology will follow the money. It's doesn't take much effort to figure out where that ends up, and it's not back to a traditional desktop.

 

Yeah but Lasso doesn´t make it portable eh? It´s another thing if they can fit four GPU´s into one laptop, but having external GPU´s doesn´t sound like the ideal solution. Now that transition you talk about, maybe you can see into the future, I certainly cannot.

I have a hard time to believe desktops will dissappear, not with the speed they progress now with new GPU´s and CPU´s, Until laptops can catch up performance wise and be equal right off the start when a new GPU is released for the desktop, yeah then I agree desktops are over. But the problem with notebooks is it´s upgradeability.

Once you buy a laptop you are stuck with that technology for a while. In order to replace the desktops I believe a unified form factor for all GPU´s should be developed. Not like it is now. With the same form factor for all notebooks, specifically higher ends like 17" where it´s easy to change GPU and CPU alongside a Bios update and in worst case of course a mobo change. Then I could see myself going straight for only notebooks. But notebooks are today so severly limited and has been so for years. Do you think that will change in just about 2-4 years time?

 

I seriously wish we could build our on laptops like we do with desktops.

 

Vertical scale represents notebook sales as % of total computer sales.

Here's a recent article quote:

"For the year, desktop processors grew by 2 percent, servers were up 5.6 percent, and mobile processors grew a whopping 34.3 percent.

IDC's PC researchers estimate that in 2010, laptop system sales will finally surpass desktop sales -- one year earlier than it had previously projected. Rau said he expects an even earlier transition for CPUs, with laptop chip sales surpassing desktop models in late 2009."

Link: http://www.internetnews.com/ent-news/article.php/3723141

Here's some more in an emerging market (India):

"New Delhi: The sales of personal computers in India registered 11 percent growth in the first half of fiscal 2007-08, with notebook sales logging an impressive 59 percent rise, says a study released Tuesday by an industry lobby.

"Notebooks have emerged as a significant driver for the personal computer market in India in the first half of 2007-08," said Vinnie Mehta, executive director of the Manufacturers' Association for Information Technology (MAIT)."

Link: http://www.siliconindia.com/shownews/39285

A little more:

"IDC estimates that the U.S. PC market grew by only 3.6% year-over-year. According to David Daoud, an analyst with IDC, the negative growth of the desktop market continues to accelerate, with the U.S. shipment decline believed to come in between 6% and 8%. Daoud told TG Daily that year-over-year desktop shipments began to decline in the fourth quarter of 2005, with low single digit decreases until the fourth quarter of 2006. In Q4 2006, the decline peaked at 10.9%.

In contrast, the notebook segment continues to show strength. IDC said that U.S. notebook sales may have grown up to 20% in Q1 year-over-year, which compares to an average growth of 16% for the quarters of the 2006 calendar year.

According to Daoud, the market shift towards notebooks is accelerating as desktop computers are failing to appeal to consumers. While the segment is still growing in emerging markets where low prices are key to success, desktop computers do not represent enough value to consumers in saturated markets such as the U.S. Daoud believes that growth is still possible, but will require products that answer changed “usage models” – such as “true” media center PCs that are easy to use."

Link: http://www.tgdaily.com/content/view/31684/118/

 

Actually it does make it portable, because it's not integral to the system, and like a mouse it would be removable. Sure you don't have gaming on the road to the same level, but I don't think anyone is strictly talking about it being all intergrated into the chassis alone, even though that will still be the majority of PC users and gamers (as lame as the games may be).

Once you move to a more modular internal and external way of thinking things like Larrabee make it that the divide dissapears and you lose that dedicated new high end GPU, then economies of scale kick in and then that niche hgih end stand alone GPU market has no one else to rely on to share the R&D costs because everyone else is using a Fusion product. intel is talking about multi-terraflop power in the first gen Larrabees in 2010, which would give you factors of 4+ times the computing power of a C2D with an HD2-3K or GF8800. Once that happens IMO it's a whole ne ballgame, even for mobo design (single area focus on cooling efficiency instead of 2 differently stressed areas).

Somewhat, but they are getting cheaper too so replacement is easier. It cost me now less to buy this laptop than it cost me to buy the components for my XP2000+ athlon gaming rig at the time. So whereas I alternated before, I upgrade pretty much every year now so I'm not all that 'stuck'. I may chose to stay there of my own accord, but upgrading is much easier than before and is getting ever easier.

Well that wouldn't really be necessary, but they are already kind of doing that with so many laptops just having tweaks to a Clevo chassis. However the future would involve less concern about the mobo as the most important tasks would be done by one part, so replacing just a Fusion/Larrabee would be much easier than our current mess.

Yes, IMO, 2 years is the beginning of the end where we just start seeing the shhift and then the products that will precipitate that ****s and in 4-5 years will likely be the end (by which time a PS4 , Xbox 1080, Nintendo whatever [prince symbol maybe]) will take over and help push desktops off that cliff.

Even if you desktop space is replace by a docking station that the portable snaps onto or slides into (integrated larger monitor), the end result is the same which is a move away from the large format desktop.

Anywhoo, to me it's a given, but obviously for you and others it hard to picture that the same way, and maybe 1 or 2 other things need to happen betwee now and then. I don't look at it as an A vs B thing so much as just evolution of the PC, and what are the logical next steps based on where the market is taking itself.

Maybe I'm wrong, and instead of 1 or 2 events occuring to speed it along there are 1 or 2 events that work aginst that change and shift the future again.

But I've explained my view on it, and sofar it matches alot of the product planning of the big companies. Pulp saying the view is patently false just seems short sighted to me. I don't say this will happen in the next refresh, but with intel the 800lb gorilla's plans being what they are and the market moving to notebooks the way they are this won't take as long as some had thought it would without their entry into the Fusion market. Right now the only thing left to ask is what nVidia's role is in all this, making high end GPUs alone isn't enough to keep that company afloat, so expect them to do something like buy VIA and move toward Fusion, not marginalize themselves further with big single core GPUs. Multi-cores inside multi dies on a single socket doing double duty is the future that makes economic sense, but it doesn't necessarily help gamers.

 

Exactly.

What does that last part sound like if not the same shift consoles are taking?

And those stats you posted are the overall worldwide chip market which they mention lag the shift already taking place in the US. According to this article at Arstechnica, this (2008) could be the year that laptops outsell desktops in the US;
http://arstechnica.com/news.ars/pos...year-laptop-sales-eclipse-desktops-in-us.html

When Larrabee & Fusio become mobilized that's just going to speed things up even more, because their goal is to simplify and reduce cost while increasing power output over existing OEM combinations.