[H]ardocp (Kyle Bennett) thoughts on ATI/AMD, Intel, and Nvidia.

Well, it's not like they have burned the bridges, killed the hostages and left flaming bags of poop on the doorstep of AMD HQ. However, if K10 is all that and a bag of cocaine-laced chips, Nvidia can only do so much considering that AMD would much rather you bought their GPU and motherboard to go along with their CPU. (Of course, so would Intel). AMD no longer need Nvidia to carry their banner to enthusiast community.

Basically, the long-term strategy for Nvidia to survive and thrive is predicated upon their graphics technology being so much better then the alternative that the people would not only flock to it but also buy their chipsets. If G80 wasn't what it is, how many 6xx chipsets would Nvidia have sold? My bet, a fraction of what they ended up moving.
 
AMD no longer need Nvidia to carry their banner to enthusiast community.

Mmm, not sure about that yet, actually. At least when you throw "enthusiast" in front of it. I think it's not clear yet that there isn't plenty of room for NV in high-end value-add enthusiast type AMD mobos with everything but the kitchen sink in them. That segment of the mobo chipset market wasn't generally ATI's strong point anyway. Right? And surely it's going to be hard for AMD to claim enthusiast supremacy if they don't have robust SLI support mobos available, no matter how "Oh no, there goes Tokyo!" their CPUs are.

Tho it might help everyone find a relationship they can live with on that front if NV agreed to allow CrossFire to work on their mobos. Not so much do the work for AMD/ATI in that respect, but share information in a timely manner and promise not to do anything to try to screw it up.
 
I think Nvidia will cling to SLI to the end. The situation in high-end market right now is such that you first pick a video card, weight the possibilities of getting a second one at some point down the road and buy the motherboard based on that - never mind that this approach has proven a failure in the last few generations*. If people didn't have that annoying "but what about SLI?" voice in the back of their heads, I think Nvidia's motherboard sales would not be what they are.
I know this is as anecdotic as it gets, but I have seen far too many people building new systems buy G80 because it was the best card on the market AND 680 motherboard with SLI possibility being the key deciding factor over 965/975. Nvidia's position is such that by keeping SLI closed they will sell more chipsets if they are leading the GPU race but lose sales if they are behind. Thus, in order for them to open it up they would need to be behind long enough for decline in the chipset sales to become unacceptable, as some sort of last resort move. That, or a large enough moneyhat from an interested party.






*From G70 and on, it has been a much better proposition to sell your current GPU and buy the next generation card than to add a second one.
 
Well, I'm not talking about SLI on AMD chipsets. I'm talking about CrossFire on SLI chipsets, and possibly AMD noting that they'd have no real need to compete in that gold-plated end of the chipset market (Hardware firewall! Zomg! Giant northbridge coolers, ZOMG++++!) so long as CrossFire support was taken care of without their participation.

Edit: Tho GPGPU might screw up such a proposed detente anyway, come to think of it. . .
 
if Intel makes them a good offer - let's say $42 a share - Nvidia will vanish in a snap of the fingures.
That's not exactly true. You are thinking of it in a "friendly takeover" kind of way; but if Intel proposed that, you'd expect the board of directors to disagree and it'd degenerate into a hostile takeover with poison pill & friends.

Intel might afford NV at $10B, but if they realize they would have to pay $20B+ due to the various provisions and that they'd pissoff the entire workforce, which is really what they are acquiring, I don't see any reason why they would want to do this. Unless you are presuming the board of directors would agree to an offer at $42... Which is a tad ridiculous, to say the least.
Basically, the long-term strategy for Nvidia to survive and thrive is predicated upon their graphics technology being so much better then the alternative that the people would not only flock to it but also buy their chipsets. If G80 wasn't what it is, how many 6xx chipsets would Nvidia have sold? My bet, a fraction of what they ended up moving.
I agree with this completely, on the other hand. If NVIDIA had graphics solutions that barely matched AMD or Intel's, they're as good as death. Clearly, their strategy now most likely is to increase their per-architectural-generation R&D budget to ridiculous levels and hope AMD/Intel can't match them. It also has the benefit that they have one efficient architecture to create derivatives out of in a variety of market segments (IGPs, Handhelds, GPGPU, etc.)

Of course, as NV30 proved, there is more to a good architecture than a large budget. On the other hand, I don't think anyone ever had a competitive disadvantage from having more well-designed semi-custom or full-custom logic blocks.
trinibwoy said:
And that can change in the blink of an eye - Nvidia is in a sticky position here. There is no guarantee that AMD won't rise again as the darling of the enthusiast community and if they do Nvidia will want to be there with competitive products supporting AMD CPU's.
I would personally expect Intel and NVIDIA to have a much better idea of AMD's prospects than we do (via confidential information they were, or were not, supposed to have), but maybe I'm wrong...
 
Arun, why do you think Nvidia's board would oppose a buyout at mid-40s level? I know there has been a lot of rhetoric, but it tends to subside when the cash is on the table.
 
Arun, why do you think Nvidia's board would oppose a buyout at mid-40s level? I know there has been a lot of rhetoric, but it tends to subside when the cash is on the table.
Well, if you think that Jen-Hsun is sincere when he thinks he can make NVIDIA a $10B revenue company within 5 years -- and he has said or implied that several times -- it is beyond me why he (or the board of directors in general) would be in favour of a takeover with such a small premium.

Companies that firmly believe they have significant growth potential, all the capital they need to achieve it, and have nothing to gain from being part of a larger family in terms of synergies or IP... Well, those companies don't really like to be acquired in general. And unless you think Jen-Hsun is bluffing (I think he's being slightly overoptimistic myself, but nothing drastic)... :) Of course, they could lose faith in a few months/years if they no longer believe they can achieve their growth targets.

P.S.: I guess Digi's message is another way to say the same thing, heh :p
 
No reason or evidence I have seen out there may suggest possible buy out of Nvidia by Intel. No reason to what so ever. Now I can understand possible partnerships for a temporary point in time, for trading tech, or buying licenses or what ever may be the case. Frankly, Intel does not value Nvidia at 10 billion dollars.

Intel has a partnership with Imagination technologies for their GPU effort as well as 3Dlabs who most of their engineers have jumped to Intel.

It would seem Intel is looking to buy some IP from Nvidia. Don't know exactly what Nvidia is looking for in return, but I would not put high hopes for x86. If that is the case, Nvidia must be giving them hell of alot of IP and that is something I don't see Nvidia doing. More than likely this is a small fry business transaction.


It's hard to predict what is going to happen. Their is intresting times ahead for sure. One firm seems to have already played their cards, another holding on to their cards waiting for the right time, and the the other seems to be putting on a poker face.
 
To avoid any confusion: I am not saying that Intel is buying Nvidia or anything of that sort, just that I think they COULD be bought if the offer is good.
 
Well, if you think that Jen-Hsun is sincere when he thinks he can make NVIDIA a $10B revenue company within 5 years -- and he has said or implied that several times -- it is beyond me why he (or the board of directors in general) would be in favour of a takeover with such a small premium.

Companies that firmly believe they have significant growth potential, all the capital they need to achieve it, and have nothing to gain from being part of a larger family in terms of synergies or IP... Well, those companies don't really like to be acquired in general. And unless you think Jen-Hsun is bluffing (I think he's being slightly overoptimistic myself, but nothing drastic)... :) Of course, they could lose faith in a few months/years if they no longer believe they can achieve their growth targets.

P.S.: I guess Digi's message is another way to say the same thing, heh :p


JHH does have ego, but he also is not shy on common sense. And while I am not accusing him of lying, I have yet to see any tech company come out and say "Our growth potential is limited and in the mid-to-long term the prospect for our continuing progress is dicey. For what it's worth, Orton was publicly bullish on ATI's "leadership" and "near-unlimited growth potential" while getting backrubs from Hector.
 
I don't know what to make of that, frankly.

Is it:

1). ZOMGering by Inq
2). FUD by AMD
3). Something real that means launch frequences for Phenom are going to be higher than they thought a little while ago.

At the moment, I can't rule out any of the three (tho obviously I hope it isn't #2, as that kind of thing bites you in the butt later, almost always)
 
I think about as much as the "R600 is confirmed 65nm" news from the same source.

Some of the questions that need to be answered before any strategic implications can be ascertained are in regard to price/performance (at whatever frequency it launches)/power consumption of K10 Vs C2D/Q as well as availability, volume and Penryn specks and release date.

Let's take the best-case scenario for AMD: top of the line Phenom is on average 30% faster than Intel's top-of-the-line, draws less power, is priced competitively and available in July, with extremely rapid ramp providing mass availability. Let's also assume that AMD retains sizable perform ace advantage into 08 AND Intel will not throw more fuel into price war by chopping prices. Yes, that's a LOT of "If"s.

Financially, this would of course be a boon for AMD form Q3 and on. Nvidia would defiantly want to be in AMD chipset market. And as always, they would need to have a stand-out 3D product to make investment in their "semi-platform" seem attractive.
 
JHH does have ego, but he also is not shy on common sense. And while I am not accusing him of lying, I have yet to see any tech company come out and say "Our growth potential is limited and in the mid-to-long term the prospect for our continuing progress is dicey. For what it's worth, Orton was publicly bullish on ATI's "leadership" and "near-unlimited growth potential" while getting backrubs from Hector.
Obviously. I'm sure they would consider it very seriously if Intel gave a very large premium (think $60+) and a guarantee of sufficient independence so that the engineers don't flee to other companies...

That's more than $20B I think. I don't believe Intel is ready to cough up that kind of cash. I mean, if Sound_Card doesn't think NVIDIA is worth $10B, surely they would never consider that much money! ;)

As for Barcelona... If the clocking bottleneck was the memory controller, why were all leaks indicating that quad-cores would be clocked lower than dual-cores? They're based on the same chip with redundancy afaik, ffs. This would clearly indicate a TDP limitation, imo. But it's not impossible that a respin is helping the power dissipation too, I'd imagine.
 

http://www.theinquirer.net/default.aspx?article=39492

Charlie has some "facts" of his own:


- 1600 MHz FSB license (pretty predictable).
- Intel branded motherboards with Nvidia Chipsets (like the old deal with ATI, another of his extrapolations).
- Nvidia GPU's bundled with Intel motherboards (again, not new, since Intel did it by bundling Geforce 6200's with certain i925 and i915 products).
- Certain joint marketing operations and one-time promotions.
- CSI license (this means access to "Nehalem", also not out of this world).
- Access to a Xeon 2S license.


Of all the above, the Xeon 2S license is probably the most significant (not to mention profitable), since i can't recall any 2S server board for Intel processors with anything other than an Intel chipset.
This could mean big bucks for Nvidia, if they play their cards right.

Still, if any proof was needed to take this article with a pinch of salt, look for the following contradiction:

Look for low end and specialty boards, not mainstream business grade boards here, NV chipsets are too buggy for that kind of application.

Later, in the same article...
Intel is letting NV into the one place it slammed the door on to everyone, 2S servers. Yes, NV is getting a Xeon license.

See anything out of place ? :D
 
Last edited by a moderator:
Ok, I'll admit to being stupid...why is it such a big deal that nVidia gets a Xeon license? :oops:

Much greater profit margins, big Intel-only businesses now know "Nvidia" (brand recognition) and start to trust them, etc.
The benefits are undeniable. Look what it did for nForce Professional in the AMD Opteron Server/Workstation market.
 
Back
Top