ATI's Intel Chipset Team Reassigned

[ltcommander_data]

http://www.hwupgrade.it/articoli/skvideo/1540/ati-factory-tour_index.html

http://www.google.ca/translate?u=http%3A%2F%2Fwww.hwupgrade.it%2Farticoli%2Fskvideo%2F1540%2Fati-factory-tour_index.html&langpair=it%7Cen&hl=en&ie=UTF8

This Italian site did a recent tour of ATI's facilities and they have a few interesting tidbits to report. Supposedly, most of the design team specializing in Intel chipsets have been redeployed to work on AMD platforms. Obviously the RD600 and probably the RS600 are already in final stages so they will be completed but future chipsets are in doubt. They mention something about expecting just 50% Intel chipset sales levels in 2007 and 0% by 2008. They predict, probably correctly, that Intel is going to move to lock them out anyways so they are acting pre-emptively accordingly.

Giving credit where credit's due I first noticed this article from Beyond3D, which is partnered with Hexus now, but I expect this will be on The Inquirer tomorrow.

 

[bixplus]

Thanks for this information ltcd...it'll be interesting to see how this whole thing shakes out. I guess we'll know in another year or two...either way, very exciting to watch.

 

[Dade_0182]

Thanks for this information ltcd...it'll be interesting to see how this whole thing shakes out. I guess we'll know in another year or two...either way, very exciting to watch.

I second that. Great find.

 

[chuckshissle]

Nice.

 

[Mex]

Do you think that RD600 will have any effect on how long it takes ATi to separate from the Intel platform? Let me also ask how you think it will affect Intel's reputation. I ask this because of an article that I saw testing the RD600 - apparently, it was matching and in some cases outperforming the 975X (That result is from memory, and may not be 100% correct). How would it look if Intel forced out a competitor that put out a higher performing product? (Emphasis on look)

 

[ltcommander_data]

Well in this case it isn't as much Intel forcing ATI out as Intel not wanting ATI in and ATI deciding for themselves to leave quietly. Besides, Intel isn't going to bluntly deny system builders permission to use the RD600 all they are going to do is ensure test systems are built around Intel or nVidia chipsets. What Intel is going to do with the i975X is probably lower its price to make it more competitive and in this case it is justified since it's an old mature chipset so the manufacturing costs are very low allowing Intel to simply pass on the savings. In the case of nVidia, it isn't Intel favouring nVidia chipsets as much as Intel support Quad SLI which just happens to only work on nVidia chipsets. As you say it's all a matter of perception.

The thing that concerns me is also that the RD600 may be a great card, but if they don't launch in the next month it's going to be late to the party. With the i975X it's been around a long time allowing people like ASUS to build a very feature rich board like their Digital Home Edition. This way you may be selling a slightly slower board, but it has lots of things thrown in to make up for it. Feature rich boards take a while to design since you need to layout a more complicated board and ensure all the third party chips work properly and don't interfere with each other. If the RD600 takes too long, the boards that are going to be available in Q4 may be fast, but they may not be a nice Premium edition which enthousiasts like, which limits its momentum.

Besides, the RD600's lifespan only extends to about Q2 2007. Past that point Intel is releasing their new high-end chipset and everything is switching to DDR3 and it looks like ATI isn't going to bother developing any new boards. Between Q4 2006 and Q2 2007, the RD600 should sell very well and in that timeframe Intel probably wouldn't be too concerned anyways because ATI wouldn't be fully merged with AMD yet. Past Q2 Intel will get more aggressive, but by that time ATI looks to have already pulled themselves out anyways so it kind of works out for everyone.

 

[joset]

The notice seems to have spread over the net, already ("Intel pulls the license on ATI").

Intel has a stronghold on its chipset income (about 50% of its chip's); do you think this move could strenghten their investment in more performant & feature-rich chipset solutions or, on the contrary, start or increase its partnership with other contenders, like nVidia?


Cheers!

 

[Zelus]

I think this spells out doom for the "Boundless Gaming" concept. RD600 was going to be THE chipset for that kind of thing. Now, the AMD merger has made it more likely that AMD and ATI are going to get drunk one night and kick out some mutant spawn of a C/GPU nine months later instead.

ATI might still attempt to focus on physics, in fact I'd bet on it, but I don't think they'll persue the 3x PCIe 16x slot idea. From what I remember, several motherboard manufactures have said that they simply won't bother with 3x PCIe 16x slots. They'd rather provide the user with PCI options.

 

[ltcommander_data]

Intel has a stronghold on its chipset income (about 50% of its chip's); do you think this move could strenghten their investment in more performant & feature-rich chipset solutions or, on the contrary, start or increase its partnership with other contenders, like nVidia?

For the time being I think Intel is focused on getting the P965 chipset optimized since it does have quite a bit of potential in it's new memory controller. For the high-end they are probably letting nVidia take the lead with Quad SLI which is what was shown at Conroe's launch and is what Dell's top model is if they ever decide to ship it.

In terms of overclocking potential the i975X definitely still has a lot of potential due to it's advantage of being on the market so long the hardware and BIOS is very heavily optimized.

http://www.xtremesystems.org/forums/showthread.php?t=111490

Here they reached a 100% OC on the E6600 to 4.88GHz and a 2.17GHz FSB. It was on water cooling too, not LN2. I don't think the current nVidia 590SLI is the most overclockable chipset, but nVidia is supposed to release a refresh in Q4 to coincide with Kentsfield, which should improve things. The RD600 will probably still be better, but it probably won't be by much.

In the long term though I can't see Intel sitting idle on high-end chipsets and deferring to nVidia. They just can't do that on their flagship product. The indications are though that with Bearlake-X they are going to come back strong. DDR3 is supposed to launch starting at 800MHz yet Bearlake-X is going to support DDR3 1333, which is very aggressive. Nobodies mentioned anything yet, but I'm hoping it'll have support for a 1600MHz FSB also for single die chips since that's certainly achieveable and would be preferable for a single die 45nm quad core.

The question I have is whether Intel is really going to create their own discrete GPUs. The Inquirer mentioned that Intel acquired 3DLab's design team, the ones who designed OpenGL 2.0, earlier this year so they certainly have the manpower now. With Bearlake-X and it's support for PCIe 2.0 that would be the perfect time to introduce discrete GPUs as the the industry is transitioning to the new bus. At that point, Intel will be beginning to transition to 45nm and the 65nm Fabs would be mature which allows capacity room. That way Bearlake-X doesn't need to support SLI because it can have an Intel in house solution. Having SLI support also would be preferential too of course.

 

[Whizzard9992]

I, for one, am dissapointed. I'm a fan of ATi and Intel. If Intel locks out ATi or vise-versa, then if I buy Intel I'm stuck with AMD (Unless I want a VIA/3rd party chipset).

 

[ltcommander_data]

http://digitimes.com/mobos/a20060821A5023.html

In a related story motherboard makers aren't overly enthousiastic about developing motherboards with ATI chipsets. With motherboard makers unwilling to make Intel motherboards with ATI chipsets, and ATI unwilling to make new Intel chipsets, it looks like the Intel chipset market has been "secured" without any blatant intervention from Intel itself.

 

[mpjesse]

http://www.hwupgrade.it/articoli/skvideo/1540/ati-factory-tour_index.html

http://www.google.ca/translate?u=http%3A%2F%2Fwww.hwupgrade.it%2Farticoli%2Fskvideo%2F1540%2Fati-factory-tour_index.html&langpair=it%7Cen&hl=en&ie=UTF8

This Italian site did a recent tour of ATI's facilities and they have a few interesting tidbits to report. Supposedly, most of the design team specializing in Intel chipsets have been redeployed to work on AMD platforms. Obviously the RD600 and probably the RS600 are already in final stages so they will be completed but future chipsets are in doubt. They mention something about expecting just 50% Intel chipset sales levels in 2007 and 0% by 2008. They predict, probably correctly, that Intel is going to move to lock them out anyways so they are acting pre-emptively accordingly.

Giving credit where credit's due I first noticed this article from Beyond3D, which is partnered with Hexus now, but I expect this will be on The Inquirer tomorrow.

Good find. I don't think it's much of a shock that ATI/AMD will eventually stop making intel chipsets. Not only is it an awkward situation for both companies, Intel could pull the license at any time which would leave AMD in a bad position. And would you blame Intel? Clearly, AMD/ATI would in theory be able to make very competitive intel based chipsets in volume. That would be a huge threat to Intel's existing chipset business.

Look at it this way: do you think AMD would appreciate Intel making competitive Athlon/Operton based chipsets? Hell no.

Right now AMD and ATI are in sort of a anti-trust situation because the deal hasn't gone through. So AMD and ATI still cannot share important information like bus licenses. But when the deal IS final, Intel may very well terminate ATI's chipset license due to a conflict of interest. Again, can you blame them?

;-)

 

[mpjesse]

The notice seems to have spread over the net, already ("Intel pulls the license on ATI").

Intel has a stronghold on its chipset income (about 50% of its chip's); do you think this move could strenghten their investment in more performant & feature-rich chipset solutions or, on the contrary, start or increase its partnership with other contenders, like nVidia?


Cheers!

Undoubtedly, it will create a stronger partnership with companies like nVidia or even VIA. Intel knows it has no competing products in dual graphics- which is a quickly growing platform not only in enthusiast markets, but professional markets as well. Despite what everyone likes to think about SLI, there is a very big market for it. The SLI platform expands way beyond dual video cards. It has a very real use in professional graphics. Anyways, getting back on track- I think Intel NEEDS to have a partner if it wants to get serious about professional graphics platforms. And right now that only partner is nVidia.

So the outcome seems pretty clear to me: Intel will eventually partner with nVidia. And I suspect it'll be sooner rather than later.

 

[ElMoIsEviL]

This may cause a problem for me.

What happens if Intel have the fastest CPU and ATi the fastest Graphics card and I'm wanting to use Crossfire?

What chipset would I be able to use to enable Crossfire? I am worried that Intel will cancel Crossfire on there own chipsets. And with ATi no longer making chipsets for Intel based systems then I'll be left out of the cold.

The only thing that could save me is if nVIDIA end up making the better GPU's or AMD the better CPU's. Other then that.. in the future some users may be royally f*dged!

 

[ElMoIsEviL]

in the future some users may be royally f*dged!

Especially when nVidia currently can't produce an Intel chipset that satisfies my demand and will keep the Intel chipset I need price high.

Agreed... nVIDIA's Intel Chipsets are CRAP...lol. Unless they've improved with the 570 and 590 which I have yet to see a Tomshardware Live stress test to confirm.

 

[joset]

In the long term though I can't see Intel sitting idle on high-end chipsets and deferring to nVidia. They just can't do that on their flagship product. The indications are though that with Bearlake-X they are going to come back strong. DDR3 is supposed to launch starting at 800MHz yet Bearlake-X is going to support DDR3 1333, which is very aggressive. Nobodies mentioned anything yet, but I'm hoping it'll have support for a 1600MHz FSB also for single die chips since that's certainly achieveable and would be preferable for a single die 45nm quad core.

Yes, rumours has it that Intel won't be pushing its 9xx chipset line any further (i.e., a 980, for instance; don't have any link, though); hence, Bearlake (& downgraded versions) will probably be used in the server as well as in the DT spaces.

The question I have is whether Intel is really going to create their own discrete GPUs. The Inquirer mentioned that Intel acquired 3DLab's design team, the ones who designed OpenGL 2.0, earlier this year so they certainly have the manpower now. With Bearlake-X and it's support for PCIe 2.0 that would be the perfect time to introduce discrete GPUs as the the industry is transitioning to the new bus. At that point, Intel will be beginning to transition to 45nm and the 65nm Fabs would be mature which allows capacity room. That way Bearlake-X doesn't need to support SLI because it can have an Intel in house solution. Having SLI support also would be preferential too of course.

I wasn't aware of such a bombastic aquisition (trusting the Inq., that is); bit-tech.net has also mentioned Intel's intentions to move GPU tasks back into the CPU, as quoted:

Intel has already dropped support for CrossFire on its 965-series chipsets and it is unclear whether Intel will continue to design chipsets that support multi-GPU technologies. The company has been hinting at moving GPU tasks back onto the CPU for a few months now.

Not really a very reliable source, so...


Cheers!

 

[joset]

So the outcome seems pretty clear to me: Intel will eventually partner with nVidia. And I suspect it'll be sooner rather than later.

If Intel's aquisition of 3D Labs is confirmed, I find it hard to believe in any soon-to-be partnership, graphics-wise. Intel already has what it takes to be competitive, in the mid-to-high-end space. 3D Labs is a strong & reliable IP company.

That leaves nVidia in a somewhat awkward position; will nVidia 'grow up' into another CPU/GPU/PPU/... manufacturer? A new platform in a 'dull' AMD/ATi vs Intel landscape?


Cheers!

 

[Whizzard9992]

I, for one, am dissapointed. I'm a fan of ATi and Intel. If Intel locks out ATi or vise-versa, then if I buy Intel I'm stuck with AMD (Unless I want a VIA/3rd party chipset).

*correction* "...then if I buy Intel I'm stuck with nVidia..."

Wow that was bad..... I gotta stop posting past my bed-time

 

[Whizzard9992]

So the outcome seems pretty clear to me: Intel will eventually partner with nVidia. And I suspect it'll be sooner rather than later.

If Intel's aquisition of 3D Labs is confirmed, I find it hard to believe in any soon-to-be partnership, graphics-wise. Intel already has what it takes to be competitive, in the mid-to-high-end space. 3D Labs is a strong & reliable IP company.

That leaves nVidia in a somewhat awkward position; will nVidia 'grow up' into another CPU/GPU/PPU/... manufacturer? A new platform in a 'dull' AMD/ATi vs Intel landscape?


Cheers!

Perhaps a close relationship between nVidia and Intel won't happen, but I'd be surprised if they don't have SLI support on their boards to some degree. Intel has seen success with their enthusiest line of products (BadAxe, XE procs), so we know enthusiests are well-within their field-of-view. We also know that Intel highly values its chipset business, and it will lose substancial support, and inevidebly market share, if it discontinues its persuing the enthusiest market.

I think a real concern amongst Intel Chipset fans is that the drivers for Xfire and ATi will start to suck once the merger finalizes. They can still fulfill their obligatory support for their products, but as we all know, the drivers are in an obligatory gray-area. Why would I invest in an ATi card if I know its going to lose support soon on my intel chipset? This is going to hurt Intel's chipset sales for the high-end market.

We also know that the bigger concern with Intel's chipset market is where it currently dominates, and that's the enterprise desktop. It's possible they could move to preserve this market and cut support temporarily, or indefinately, of white-box enthusiest desktops.

I guess its all speculation, but even with the purchase of 3D labs' team, I don't think they're going to snum nVidia. Intel has a discrete graphics team, but it's not dedicated to the same product lines as nVidia and ATi. Even if Intel comes out with a competitive line of enthusiest video cards, they don't have the experience or the reputation to cause any real damage to AMD/ATi, which will give AMD another 2-3 year period of market domination until Intel proves itself as a viable competitor.

So again, unless I'm mistaken, I don't really think Intel will snub nVidia. I think that they need each other in the interim, at least until some new business strategies solidify.

 

[sviola]

that translation was pretty rocky its amazing how those romantic languages use words,,,

Yes, that's why we get all the chicks... È solo l'amore!!

 

[ltcommander_data]

Further confirmation that ATI is discontinuing development of Intel chipsets.

http://www.hkepc.com/bbs/itnews.php?tid=656138&starttime=0&endtime=0

ATI has removed the 2007 timeline chipsets the RD700, RS700, and RC710 from their roadmaps. All that remain are the already confirmed RD600 for high-end, RS600 for mid-range, and RC610 for low-end.

I wasn't aware of such a bombastic aquisition (trusting the Inq., that is);

The original article is here:

http://www.theinquirer.net/default.aspx?article=33812

Essentially, 3DLabs was downsizing their graphics division so they laid off people who Intel picked up. I don't think it was an official corporate deal just that it worked out that way. Intel and 3DLabs has worked together on graphics chips before so it's likely the engineers that were hired were known to Intel so it was basically Intel helping "friends" in a tight spot.

X-Bit Labs also confirms this story and they reported on it before The Inquirer picked up on it, which is probably a good thing.

http://www.xbitlabs.com/news/video/display/20060816030933.html

While the possibility for Intel to develop a graphics core is fairly high, as the company is continuously hiring new staff members (for instance, the chipmaker now employs several people from the former 3Dlabs workstation graphics team), another project that involves experience of CPU and GPU development could be a processor that combines general-purpose processing and graphics processing capability. Earlier this year Intel already confirmed intention to build such a chip.

the pic of the ati integrated video mb was cool,it has the equivalent of an x700 gpu;I think thats what i read.

The RS600 and the AMD equivalent the RS690 use a graphics core based on the X700 although there are quite a few differences. For one, those chipsets look to incorporate some form of AVIVO which is a X1xxx generation technology. However, the performance of the IGP will not be at X700 level. The target performance is X600 level and it'll have 4 pixel shaders and 2 vertex shaders which is doubled what the Xpress 200 has. The RS600 will also only be DirectX 9.0b level not DX9.0c or DX10. The DX10 IGP was supposed to be the RS700 in 2007 (Q2?), but that looks to be cancelled now. In contrast to the RS600, the Intel GMA X3000 has 16 ALUs which seems to agree with an 8 unified shader configuration so in theory it should be faster than the RS600. However, Intel can't seem to get their drivers in line since hardware PS, VS, and T&L aren't activated yet. That lack of driver support is why people are reporting poor GMA X3000 performance up to this point.

 

[joset]

So again, unless I'm mistaken, I don't really think Intel will snub nVidia. I think that they need each other in the interim, at least until some new business strategies solidify.

While I do agree that, overall, every company needs the other, in more ways than one can think of, this particular conjuncture has already sprung a fair amount of surprises; I'd dare to say that Intel doesn't play dice ; if this 3D Labs aquisition is confirmed, Intel will not just be 'upgrading' its GPU division; it will be creating an entirely new one (that's my conviction, anyway). Otherwise, why should they need such a high-end [graphics] expertise?
More than ATi, 3D Labs' main competitor has been nVidia's Quadro line, in the high-end workstation graphics; now, nVidia will have to deal with two immensely strong contenders: AMD/ATi & Intel, graphics-wise. In my opinion, the 'enthusiast' platform will be strenghtened; there's a non-zero probability that the 'enthusiasm' will be shifted... Again, where will nVidia stand?
On the other hand, nVidia 'only' lacks its own CPU for a complete "n-platform"...
Pure speculation, of course.

Cheers!

 

[ltcommander_data]

On the other hand, nVidia 'only' lacks its own CPU for a complete "n-platform"...
Pure speculation, of course.

I believe it was The Inquirer that also pointed out that CPU defiency for nVidia. They suggested that nVidia might acquire VIA. It's pretty farfetched right now, but it is logical and it would give nVidia very strong mobile expertise. With low power consumption being a primary reason why AMD bought ATI, it would make a lot of sense for nVidia to bulk up on that area in addition to acquiring VIA's C7 processors. VIA's processors are also ideal for integrated GPU solutions since they use very little die space allowing GPU room and are great low-end solutions which would be the perfect target market for CPUs with built-in GPUs.

 

[joset]

Essentially, 3DLabs was downsizing their graphics division so they laid off people who Intel picked up. I don't think it was an official corporate deal just that it worked out that way. Intel and 3DLabs has worked together on graphics chips before so it's likely the engineers that were hired were known to Intel so it was basically Intel helping "friends" in a tight spot.

(...)

While the possibility for Intel to develop a graphics core is fairly high, as the company is continuously hiring new staff members (for instance, the chipmaker now employs several people from the former 3Dlabs workstation graphics team), another project that involves experience of CPU and GPU development could be a processor that combines general-purpose processing and graphics processing capability. Earlier this year Intel already confirmed intention to build such a chip.

Seems a causal reaction to the AMD/ATi merger; in some sort, that also implies that AMD/ATi might have some [obvious?] plans of going a similar line...


Cheers!

 

[joset]

On the other hand, nVidia 'only' lacks its own CPU for a complete "n-platform"...
Pure speculation, of course.

I believe it was The Inquirer that also pointed out that CPU defiency for nVidia. They suggested that nVidia might acquire VIA. It's pretty farfetched right now, but it is logical and it would give nVidia very strong mobile expertise. With low power consumption being a primary reason why AMD bought ATI, it would make a lot of sense for nVidia to bulk up on that area in addition to acquiring VIA's C7 processors. VIA's processors are also ideal for integrated GPU solutions since they use very little die space allowing GPU room and are great low-end solutions which would be the perfect target market for CPUs with built-in GPUs.

Yes, that seemed to me a way of nVidia reacting to this platform polarization (the 'growing-up' of a proprietary CPU); although the aquisition of VIA seems reasonable, it also sounds like a downgrade in nVidia's current offerings; excellent GPU/Chipsets and a so-so CPU (for the low-end to mid-range); unless both (nVidia/VIA) could come up with a competitive mainstream-to-high-end CPU (which I find unlikely, in the mid-term). And, they'd be left with Intel, for the mainstream/high-end...

Well, seems that speculation can be very stimulating...


Cheers!