NVIDIA MCP79 for VIA CPUs; Intel P45/G45 delayed(?)

Monday 31st March 2008, 09:30:00 AM, written by Arun

Rumours are swirling around the web that NVIDIA's upcoming MCP79 chipset will be compatible with VIA's Isaiah processor. The deal makes perfect sense, so the real question is: will MCP79 or MCP78 also be compatible with Montalvo's upcoming processor? Meanwhile, Intel's G45/P45 seem to be delayed...

It is extremely difficult to determine when the NVIDIA-VIA agreement was made based on this rumour and MCP79's schedule; based on our information, G98/MCP78/MCP79's schedules have been blown massively off course (the delay's magnitude is NV30-like), as, just like C73 (the 790i SPP), they have required a full silicon respin.  Therefore NVIDIA will be suffering what's effectively a completely new tape-out with all the associated costs.

Given this silicon respin, it is possible that the original tape-out did not support VIA's Isaiah processor and that this was opportunistically added in the new revision. As a reminder, VIA uses their own proprietary V4 Bus which is pin-compatible with the Pentium M's FSB, but the electrical signaling is different. So it really shouldn't be too hard to get this to work in theory (in practice, who knows).

Alternatively, it isn't clear which processes the two MCP79 tapeouts were on; it seems to be 65nm for the first one at the very least. Assuming that the second one is also 65nm, it could be that there will be a third tape-out on 55nm with fundamentally the same RTL which will support VIA's V4 Bus.

Either way, the deal makes perfect sense for both parties: it's a viable combination for the low-end consumer market at HP for example, and it streamlines a potential merger in the future. So as we said in the introductory paragraph, the real question in our mind is whether NVIDIA made a similar deal with Montalvo and our expectation would be that they did.

It is also rumoured that G45/P45 (they are actually the same chip it seems; last-gen, it was P35/G33 which were the same chip, and also P33/G31 apparently) have been delayed to mid-June. And, well, not everyone seems to agree on that estimate. Either way, it looks like G45/P45 will be released in the same rough timeframe as MCP79.

MCP79 is likely a very important component for NVIDIA, as just like the C55 (680i/650i SPP) helped make the G80 launch especially successful, they likely have similar plans for MCP79 and GT200: it is the first NVIDIA chipset for Intel platforms with support for HybridPower, which should be especially appealing (if not necessary in some people's minds) for GT200.


Discuss on the forums

Tagging

nvidia ± mcp79, g98, via, isaiah, intel, p45


Latest Thread Comments (5 total)
Posted by 3dilettante on Monday, 31-Mar-08 21:09:31 UTC
Perhaps B3d has some inside scoop on Montalvo's dealings with Nvidia, but wouldn't a more prudent question be how much--if any--of Montalvo will exist by the time the platform were to become available (edit: if it were)?

What, besides the fact that both Via and Montalvo's chips will be relatively low-power (and that both are miniscule in the grand scheme of x86, Montalvo being to Via what Via is to either AMD or Intel), would lead to any inference that a deal for one would point to a deal for the other?

At least Via exists as a revenue-producing concern.

Posted by Arun on Tuesday, 01-Apr-08 09:50:18 UTC
I think there indeed is a fundamental difference between the two - in the VIA-NVIDIA case, there are advantages for VIA, but most of the benefits are for NVIDIA. In the Montalvo-NVIDIA case, most of the benefits are for Montalvo; i.e. they are a CPU company, not a southbridge/GPU company and they need a partner for that technology or they obviously wouldn't even stand a chance before starting.

The benefit to NVIDIa, besides potential sales down the road, is clear however: realistically, their best bet to integrate a x86 CPU core into their MCPs (which already integrate a southbridge, a northbridge/memory controller and a GPU) is to acquire a startup or a small-cap company developing such a CPU, not by engineering it themselves (which has significant risk as it would take extremely long, would make Intel/AMD even more annoyed, and might not result in a competitive core). By going the acquisition road, they can also clearly know what they are buying and what the core's performance is and how they can combine that into a coherent roadmap.

We don't have any real scoop on Montalvo, but when I first read more about Montalvo, comments made by Jen-Hsun Huang in early 2007 (and a number of other things) suddenly made a lot more sense to me. Seems likely he had already met with Montalvo leaders or those of another startup at that point in time... Also, consider the advantages of providing them southbridge/GPU technology: they could effectively force them not to be acquired by Intel/AMD/... as part of that agreement. NVIDIA needs a backup plan to get a x86 core and/or x86 engineers relatively chea and this would be a pretty good one. Of course, that means NVIDIA's future in a substantial part of the market is dependent on the execution of VIA and/or Montalvo. Stategically, it's not the best position to be in - but it could also be much worse.

_EDIT_: Here's the full question & answer on the subject:
Quote
-----
[...]

Now, the question is below that, is there a single chip architecture PC? You know, something you can put together on one chip, and I think the answer from a Moore's Law perspective is clearly yes. The question, of course, is can you take off enough cost relative from a two-chip perspective to really create a new market, or does this one chip just make the existing PC incrementally cheaper?

You know, what we all hope, and I know that AMD hopes this, and I know that Intel hopes this, and so do I, is that we can create a single-chip architecture that is slower moving, much more related to bringing the last two or three years' technology into the mainstream market, and hopefully make the $200 PC a really, you know, really attractive price point for not just the manufacturer, but also for consumers.

Now, you know, in doing so, can we create the next billion unit PC market place? That's really the open question. I'm a huge fan of it, and if the market wants us to build a single-chip PC, we'll be delighted to do it. But right now, we're leading the pack in integration still, and hmm, as soon as the opportunity comes up to build a single-chip, we'd be delighted to do it.

[...]
-----
I just want to ask you [unhearable]. I think you said that if the market wants a single-chip [unhearable] device, you'd be able to make it. Uh, which if you said that [unhearable] x86 cores. So my question, did I get that right? And if that is what you said, uh, wouldn't you need to have like IP to do that?
---
Uh, I think the answer is yes, I did say that, and hmm, I think you guys, we all just need to collectively internalize what we just said about the x86 core. It's not like going to the moon or something. You know. It's not even like climbing a big tall mountain. It's not even like walking 2 miles.

I don't want you to think that is some incredible, unachievable, some mystical... you don't need any psychic powers to do it. And so I, having said that, the second thing is notice how many people make x86 cores? It's actually a lot. Right?

National sold the company that did that, and I think AMD bought it, and there's a couple of x86 startups, a couple in Taiwan that do it, China just built the next x86 core. There's plenty of x86 cores. So it's possible to do. Now the second question has to do with IP. We've got a lot of IP too! And so, IP is IP, right?

And, you know, the reason why we invent a lot of technology has to do with giving us the necessary protection to enter markets if we need to. So I'm not overly concerned about the IP issue, they'll get sorted by themselves.

I think the most important thing is, I ask myself this question all the time, can we add unique perspective, can we add unique value-added as we're entering that market place, so that we can contribute to the market. I don't ever want to be just another player.

Now, I think that if we were to build a discrete CPU, we're just another player, we're not adding any value [unherable 2-3 words], we're wasting people's money, we're wasting the world's money, and we're wasting our engineers' time.

But if there's a new way to configure a computer, and re-architect the way it's integrated, and we could imagine pretty significant ways to add value, I'll be in there in a second. If the market wants it. Okay?
---
Other questions?
Maybe just as a follow-up to that then, how do you expect the market's needs and desires? Is that something that makes sense for you to take your [unhearable] engineers, and put them on that kind of...
---
At the moment, absolutely not. At the moment, absolutely not. You know, the first single-chip computer we're going to build is for a handheld. You know, the problem with building something for the sake of making it cheaper, is that you can always make something cheaper. And so it takes away all of the value that you bring.

In going from two-chip to one-chip, it isn't such a dramatic architectural change that all of a sudden, you know, going from two-chip to one-chip, we can take a 500 dollars computer and make it 200. Unless we save 300 dollars. But we're not. We're not going to save 300 dollars. It doesn't really change anything.

[...]
---
So the take-away sounds like is, you could clearly do some kind of integrated x86 solution if you wanted to, but it's not...
---
It's kind of a pointless exercise right now.

Posted by INKster on Friday, 06-Jun-08 15:56:02 UTC
Humm (http://www.digitimes.com/mobos/a20080605PD217.html)... (and "Huh (http://my.ocworkbench.com/bbs/showpost.php?p=432248&postcount=1) !?").What's wrong with Drew ? ;)

Posted by Blazkowicz on Thursday, 12-Jun-08 19:31:26 UTC
Quote
Now, the question is below that, is there a single chip architecture PC? You know, something you can put together on one chip, and I think the answer from a Moore's Law perspective is clearly yes. The question, of course, is can you take off enough cost relative from a two-chip perspective to really create a new market, or does this one chip just make the existing PC incrementally cheaper?You know, what we all hope, and I know that AMD hopes this, and I know that Intel hopes this, and so do I, is that we can create a single-chip architecture that is slower moving, much more related to bringing the last two or three years' technology into the mainstream market, and hopefully make the $200 PC a really, you know, really attractive price point for not just the manufacturer, but also for consumers.
actually there are

Posted by Sxotty on Friday, 13-Jun-08 11:12:32 UTC
Quoting INKster
Humm (http://www.digitimes.com/mobos/a20080605PD217.html)... (and "Huh (http://my.ocworkbench.com/bbs/showpost.php?p=432248&postcount=1) !?").

What's wrong with Drew ? ;)
This is really great news for me.

I wanted a VIA system so I could have a tiny system for in my car, but their integrated graphics were terrible. If Nvidia makes something it should work way better for that.

Only problem is how Nvidia's power consumption might compare.

As of now I am just using a micro ATX board with a custom case that just fits under the passenger seat, but it would be handier to have a system that was way smaller which I could stash somewhere else.


Add your comment in the forums

Related nvidia News

So long, Chris, and thanks for all the fish
NVIDIA GF100 graphics architecture details
NVIDIA Fermi: new GPU architecture, starting with GF100
NVIDIA release OpenCL GPU drivers for Linux and Windows
NVIDIA GeForce GTX 275 at $250 to fight HD 4890
A look at NVIDIA's SLI Multi-OS and new Quadros
Ahead Nero gets CUDA support for video encoding
G92b renamed again, this time for notebooks
NVIDIA GeForce GTS 250 announced
New NVIDIA display driver for Windows 7 beta