Dell Is Going to Use Opteron!!

[Markfw]

Long overdue.

 

[Henny]

Originally posted by: Questar

Now here we are supposedly weeks away from Intel's stud cpu and now Dell says, well I guess we'll sell AMD servers now, huh?

Could be a couple of things with the timing:

1. Dell decided they were leaving too much revenue on the table.

2. Notice it's end of year before these servers ship - after fab 36 is on-line and fairly well ramped up. AMD may not be able to supply Dell with enough chips until then.

3. Companies don't decide to enter a new business in a day. Dell and AMD must have been negotiating supply agreements for weeks if not months. An earnings release is just a good time to announce.

4. There's actually a back room deal between Intel and Dell to further weaken AMD's antitrust suit.

1. Likely
2. F36 ramp doesn't start until late this year. AMD would supply anything Dell asks for whenever Dell wants it given the significance of this deal.
3. Months or even quarters.
4. Unlikely


A few more reasons:
5. Apple went with Intel. Dell wants to fire a shot over Intel's bow that the Dell business isn't a given and that Dell better get Intel's most favorable prices/treatment going forward.
6. Intel really did screw up their Xeon offering. Dell can't have this high profile segment tied to any one supplier. I see it as a wakeup call from Dell to Intel that Woodcrest/Cloverton better perform as expected.


The timing is somewhat odd. Xeon is most vulnerable right now. Dell should have done this months ago so they could have an AMD Opteron based server in today's market.

Opteron will be most vulnerable in Q4 when Woodrest/Cloverton/etc are all out in the market.

 

[evenmore1]

Cool! And these are from reliable sources! Unlike the place called the Inquierer

Awesome! I dislike Dell Desktops a little less now...

 

[Viditor]

Originally posted by: Questar

Now here we are supposedly weeks away from Intel's stud cpu and now Dell says, well I guess we'll sell AMD servers now, huh?

Could be a couple of things with the timing:

1. Dell decided they were leaving too much revenue on the table.

2. Notice it's end of year before these servers ship - after fab 36 is on-line and fairly well ramped up. AMD may not be able to supply Dell with enough chips until then.

3. Companies don't decide to enter a new business in a day. Dell and AMD must have been negotiating supply agreements for weeks if not months. An earnings release is just a good time to announce.

4. There's actually a back room deal between Intel and Dell to further weaken AMD's antitrust suit.

1. They've known about the revenue loss for years now...Dell sold fewer than 10 servers last year over the $50k price range.

2. Your point about Fab 36 is spot on! IMHO, Dell has held off signing with AMD because AMD could never guarantee the quantity of chips they require. While the server chips are a low volume/high profit part, Dell needs to cover it's bases across all lines that might be considered...

3. They've both acknowledged these negotiations for years now...

4. This has absolutely no effect on the suit. The suit is based on history and not on actions going forward...

 

[kknd1967]

This is more of a symbolic significance.
If Dell did not have any Opteron servers, people will buy from Sun, IBM and HP.

this is a late but loud&clear wake-up call for intel

 

[Viditor]

Originally posted by: Henny

Originally posted by: Questar

Now here we are supposedly weeks away from Intel's stud cpu and now Dell says, well I guess we'll sell AMD servers now, huh?

Could be a couple of things with the timing:

1. Dell decided they were leaving too much revenue on the table.

2. Notice it's end of year before these servers ship - after fab 36 is on-line and fairly well ramped up. AMD may not be able to supply Dell with enough chips until then.

3. Companies don't decide to enter a new business in a day. Dell and AMD must have been negotiating supply agreements for weeks if not months. An earnings release is just a good time to announce.

4. There's actually a back room deal between Intel and Dell to further weaken AMD's antitrust suit.

1. Likely
2. F36 ramp doesn't start until late this year. AMD would supply anything Dell asks for whenever Dell wants it given the significance of this deal.

F36 has been shipping for revenue all of this quarter...it's the 65nm ramp that starts later this year...

3. Months or even quarters.
4. Unlikely


A few more reasons:
5. Apple went with Intel. Dell wants to fire a shot over Intel's bow that the Dell business isn't a given and that Dell better get Intel's most favorable prices/treatment going forward.

I would think that HPaq's huge recent success is probably more to the point...IBM has been feeling the pain of HP's Opteron offerings as well.

6. Intel really did screw up their Xeon offering. Dell can't have this high profile segment tied to any one supplier. I see it as a wakeup call from Dell to Intel that Woodcrest/Cloverton better perform as expected.

I would be amazed if Dell doesn't already have a very good idea of just how Woodcrest/Cloverton will perform...

The timing is somewhat odd. Xeon is most vulnerable right now. Dell should have done this months ago so they could have an AMD Opteron based server in today's market.

It is indeed, and they should have. It's one of the biggest reasons that I believe that:
1. This is a reaction to HP
2. They required a larger guarantee of supply from AMD

Opteron will be most vulnerable in Q4 when Woodrest/Cloverton/etc are all out in the market.

1. As Woodcrest is a new architecture, it will require a much longer qualification period than a simple Xeon refresh. I expect qualified servers to start appearing near Q2 07

2. Woodcrest will still have problems with the greater than 2P systems...and while I expect that they will be quite competitive at the 2P space, remember that the HT links will reduce the advantage that Conroe has on desktop. I fully expect that Woodcrest will be about the same as Opteron in the 2P space...


BTW, I think that many have forgotten that Dell now also sells AMD desktop systems through their Alienware arm...

 

[Duvie]

Originally posted by: Henny

Originally posted by: Questar

Now here we are supposedly weeks away from Intel's stud cpu and now Dell says, well I guess we'll sell AMD servers now, huh?

Could be a couple of things with the timing:

1. Dell decided they were leaving too much revenue on the table.

2. Notice it's end of year before these servers ship - after fab 36 is on-line and fairly well ramped up. AMD may not be able to supply Dell with enough chips until then.

3. Companies don't decide to enter a new business in a day. Dell and AMD must have been negotiating supply agreements for weeks if not months. An earnings release is just a good time to announce.

4. There's actually a back room deal between Intel and Dell to further weaken AMD's antitrust suit.

1. Likely
2. F36 ramp doesn't start until late this year. AMD would supply anything Dell asks for whenever Dell wants it given the significance of this deal.
3. Months or even quarters.
4. Unlikely


A few more reasons:
5. Apple went with Intel. Dell wants to fire a shot over Intel's bow that the Dell business isn't a given and that Dell better get Intel's most favorable prices/treatment going forward.
6. Intel really did screw up their Xeon offering. Dell can't have this high profile segment tied to any one supplier. I see it as a wakeup call from Dell to Intel that Woodcrest/Cloverton better perform as expected.


The timing is somewhat odd. Xeon is most vulnerable right now. Dell should have done this months ago so they could have an AMD Opteron based server in today's market.

Opteron will be most vulnerable in Q4 when Woodrest/Cloverton/etc are all out in the market.

http://www.amdboard.com/amd_fab36.html

Actually you are wrong...full ramp up to 13k units per month will be achieved at end of year with 20k units per month by 2H 07....The article and bullets on side show production has already commenced in Fab36 as we speak...

http://www.amdboard.com/amd_040406.html

 

[Fox5]

Originally posted by: kknd1967
across the world, only apple is intel only
apple is unique but for how long

I think Sony is still Intel only.

 

[Duvie]

Originally posted by: Fox5

Originally posted by: kknd1967
across the world, only apple is intel only
apple is unique but for how long

I think Sony is still Intel only.

Even after the judgement handed down by the Fair Trade comission of japan???....many believed Sony was behind why the JFTC initially started the inquiry....

I will have to look into it...

 

[Viditor]

Originally posted by: Duvie

Actually you are wrong...full ramp up to 13k units per month will be achieved at end of year with 20k units per month by 2H 07....The article and bullets on side show production has already commenced in Fab36 as we speak...

I beat ya by a few seconds Duvie...
But if he meant the 65nm ramp, then he's correct...

 

[Regs]

I'm actually starting to feel sorry for Intel. Not.

 

[thescreensavers]

woot woo

 

[pkme2]

http://news.zdnet.com/2100-9584_22-6074059.html?tag=nl.e589
Guess it had to happen. It was inevitable because of the recent competition of lowering prices. Sales and profits, that's the bottom line.

 

[BrownTown]

Meh, makes sense to me, we all know that Intel will be better for notebooks and desktops, and 1 or 2 way serves, and that 4+ way servers will be better for AMD. So Dell make a move that any idiot should, and sells the best product it can in each market. I mean the high end servers will still be Netburst till 2007 for Intel, so its a total nobrainer for anybody who knows anytihng about CPUs to choose AMD over Intel in that degment. Conversely, Intel rules the vast majority of the market with its superior processors and process technology. Trust me, Dell isn't even gonna consider for one secodn using AMD in desktops or notebooks, and they shouldn't, Intel is better there anyways, plus Intel gives them lots of money to put little Intel inside stickers on their computers.

 

[kknd1967]

I just recall this
maybe also Fujitsu

Originally posted by: Fox5

Originally posted by: kknd1967
across the world, only apple is intel only
apple is unique but for how long

I think Sony is still Intel only.

 

[mechBgon]

:beer: Congratulations, AMD! :beer:

 

[Henny]

Originally posted by: Viditor

Originally posted by: Duvie

Actually you are wrong...full ramp up to 13k units per month will be achieved at end of year with 20k units per month by 2H 07....The article and bullets on side show production has already commenced in Fab36 as we speak...

I beat ya by a few seconds Duvie...
But if he meant the 65nm ramp, then he's correct...

Yes, I was referring to 65 nm fab capability. That's where it's at in CPU technology.

 

[Duvie]

Originally posted by: Henny

Originally posted by: Viditor

Originally posted by: Duvie

Actually you are wrong...full ramp up to 13k units per month will be achieved at end of year with 20k units per month by 2H 07....The article and bullets on side show production has already commenced in Fab36 as we speak...

I beat ya by a few seconds Duvie...
But if he meant the 65nm ramp, then he's correct...

Yes, I was referring to 65 nm fab capability. That's where it's at in CPU technology.

perhaps, but this Dell deal is to buy 90nm current opterons, which AMD can currently deliver...65nm will be even better in terms of performance per watt....

 

[Viditor]

Originally posted by: Henny

Originally posted by: Viditor

Originally posted by: Duvie

Actually you are wrong...full ramp up to 13k units per month will be achieved at end of year with 20k units per month by 2H 07....The article and bullets on side show production has already commenced in Fab36 as we speak...

I beat ya by a few seconds Duvie...
But if he meant the 65nm ramp, then he's correct...

Yes, I was referring to 65 nm fab capability. That's where it's at in CPU technology.

As Duvie pointed out, the 90nm Opterons are what the deal is for at the moment...and AMD has a huge volume capacity for them already.
FYI, the 65nm ramp actually BEGINS in July/August, at least that's when the first volume production wafer-ins begin. It takes about 3 months to produce chips from the wafer-ins, so AMD will have the first volume 65nm chips in Oct/Nov. However, AMD uses ~10% of their lines for tweaking and new processes, and they have been producing very low volume production ready 65nm chips for a few months now...
IIRC, the first volume shipments of 65nm to OEMs will be in Nov/Dec for a January release.

 

[Amaroque]

Originally posted by: RichUK
No surprise this has been on the cards for ages, and its good news to finally hear something official.

It has all been down to the Intel/Dell politics why Dell had never adopted the AMD approach earlier.

The new Core processor from Intel, be it good in a single socket arrangement, is still undermined by the ever aging FSB subsystem, this bottle necks the I/O between the multiprocessor arrangement seen on large Intel servers.

AMD have a far far superior platform with the use of their on die memory controller and HT links which elevates the bottlenecks , so parallel processing performance is going to be hard to match. The sad thing is the FSB model only allows one I/O operation at one time to one of the multiple processors, where as AMD has two separate buses, one for memory and one for the rest of the subsystem which resides on the HT link, and the HT link is fully duplex.

Intel have now solved the inefficient processors that were used, they now need to sort their subsystem I/O bottleneck. Which I believe should be solved by this rumoured CSI technology.

QFT!

 

[Vee]

Originally posted by: Duvie

Originally posted by: classy
AMD said that they still would be able to beat Woodcrest, this kinda leads me to believe that maybe they are telling the truth. Hmmm updated P4 core and now this, despite the ever looming return to Intel dominace by the new core architecture. Can you say Core Duo 2 yields?

I think their was some talk that the platform was going to be bandwidth limited in multi cpu configs...That may be what they assume...The juggernot of the bandwidth will hold back the 20% clock for clock improvement the core architecture should have over the aging opteron core....I say that with somewhat of a smile..."aging" that is....

Lets hope for faster intro of 65nm parts and quad cores come 1h 07, maybe early launch late 4Q 06.....Who knows...all speculation!!!

There's an interesting psychological phenomenon emerging from Intel's very early showing of Conroe and selected benchmarks. But Intel haven't showed anything that is really competitive.
I fully expect Dell to continue to use Intel on small servers and desktops. The keyword is "highend". Here there is nothing in sight from Intel that will be competitive. If Intel don't have a secret development, that they've told nothing about, unless AMD totally stalls, Intel will likely not be a contender this side of 2009. And, frankly, there's no guarantee that Intel will ever be competitive. It's not the yield that is Intel's problem.

Originally posted by: Duvie

Originally posted by: classy
There is no question that "The Core" is a serious processor, but this coupled with a few weeks back where I seen they updated the P4 core makes me wonder if all is not rosy. Dell has sat and ate at Intel's table their entire existence, despite the fact Intel has trailed AMD for the better part of 3 years. Now here we are supposedly weeks away from Intel's stud cpu and now Dell says, well I guess we'll sell AMD servers now, huh? Something tells me that yields may be a problem, something is not right behind the scenes with Intel. Despite all the pre-launch drama their stock still continues to slide. Just makes me wonder that's all.

Well we are just talking about opterons and server only...I think this is necessarily a response based on "core" architecture...may be reading a bit too much into that....

I think the woodcrest chipset is not as rosy...couple that with server certifcation (which Dell could have been doing with the opteron for sometime now) and I think we see a late introduction which should give AMD an opportunity at Dell for at least a year...if they hit quad cores first and reduce wattage with a 65nm die shrink I think they can hold on even longer...

You may remember that I did my raving about the Conroe two years ago, on this forum. While I have kept my mouth carefully shut here on AT for many months now, the reason I have not joined in the euphoria is not that I've been absent. I haven't. I've been lurking here all the time.
No, I'm not particularly impressed with the Conroe. Satisfied, but not impressed.
First of all, the benchmarks must be viewed in the light that they are carefully selected and controlled by Intel.
Secondly, it must be realized that a good deal of the performance advantage over K8 is due to the full width SIMD execution. It has always been a simple matter to fully stock the K8's three execution pipelines with hardware execution units. The fact that this haven't been done until now (K8L) is simply due to economy and optimizing the transistor real estate on the previously available bandwidth.
Combine this with a new lower power design and that AMD probably now will ramp clockrates, there is no reason to expect that AMD will not be able to close the gap enough to be competitive throughout the mobile/desktop market. On performance per buck.

At the single dualcore CPU - desktop/mobile, Intel may enjoy a marginal performance crown for some months. But that will likely not affect market shares at all. Which poses this interesting question for investors: How is Intel to improve their financial results? What huge, new market shares are ripe for their harvesting? Where is the opportunity to charge premium prices for their transistor swollen chips?

Intel have come up with a massive cache & large execution core with good instruction performance. I believe the real solution to computer performance to be something else. Multi-CPU/core and to get work into, and out of, the CPU. AMD have spent years on memory interfacing and multi-CPU performance scaling. You will see some new such technology emerge on the K8L. But already now, AMD have a firm performance advantage on Intel. Woodcrest will, in my estimate, not be able to change anything about that. It seems Dell shares that estimate. And that would be their reason to go AMD for highend servers. Any illusions that there is something from Intel to wait for, are gone.

For a taste of the matter, consider this article here at AT:

http://www.anandtech.com/IT/showdoc.aspx?i=2745

**********

However, while servers is AMD's best thing currently, it is also my firm belief that concentrating efforts on highend computing, will spell an eventual disaster for AMD, if they would do that. My conviction is that tiers of computing are always conquered from below. AMD have a very good thing going for them in AMD86-64. They would be fools to squander it by leaving mainstream markets to Intel.

DEC and SGI and many others, blinded by attractive margins, have made the mistake of not marketing to the masses. The meagre incomes from small highend markets cannot sustain technical leadership over mainstream products. So it's my hope that AMD's future CPU families will show a continued commitment to budget and mainstream. Spelled out: I expect AMD's multicore technology to be extensively employed also in the consumer sector. And I think Intel will follow.