Does AMD have Intel on the ropes?

[confoundicator]

Here's a very interesting read from <A HREF="http://www.theinquirer.net/?article=16245" target="_new">The Inquirer</A>. (For newbies, if you don't already know, the Inquirer does not confirm their news and they have been known to speculate and do some selective editing. Having said that, they've also reported good information before anyone else.)

I think if AMD can keep the performance lead long enough then they can start cutting into some serious Intel market share, but it won't happen until some high profile customers of Intel start losing money because their computers are slower (brand loyalty is hard to break, but is not always a bad thing either). A big business like a national bank switching over would be huge. I found it interesting that (the article said) the only card Intel has left to play is that the 90nm process will enable them to crank out more chips at lower cost. Hopefully we'll have a little price war and AMD will come down as well.

He that but looketh on a plate of ham and eggs to lust after it, hath already committed breakfast with it in his heart. -C.S. Lewis

 

[HardWareBoss]

Nay

 

[Scout]

It'll be interesting to see if the Optron can make any inroads into the server market, because it sure sounds like the Optron has some performance advantages.

Personally, I think Intel is far from being on the ropes...

Scout
700 Mflops in SETI!

 

[Black_Cat]

Yes and Linux is ready to deal Bill Gates the final blow.

You've tried and failed. The lesson here is, never try again. -- Homer Simpson.

 

[Cybercraig]

Don't fall for it! INTEL is just playing "rope-a-dope" :lol:


"I am become death, the destroyer of worlds. Now, let's eat!

 

[confoundicator]

Ha HAH! But how can you tell if it's "rope-a-dope" or if they're out on their feet.

I'm sure Intel will be fine, but it looks to me like AMD has a performance advantage on the high-end server market. I think it'll be interesting to see if they can hold that advantage (if it is real) long enough to actually gain a big chunk of market share.


He that but looketh on a plate of ham and eggs to lust after it, hath already committed breakfast with it in his heart. -C.S. Lewis

 

[darko21]

AMD does have a performance advantage for the time being.

What AMD don't have is the manufacturing capacity to put intel on the ropes.

Time or an alliance might change this.

If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.

 

[Atolsammeek]

I will keep saying this. From the time Amd and intel war on Speed. They tend to Leap frog each other. Lik Intel faster a few months then Amd. And there was gaps like Amd 2400 to 3400+ chips that Intel Held the crown. Then Amd 64 chip came out and Now Amd holding the crown as the fastest chip.

But the Main question is. How long can Amd hold the Crown? few weeks a few months a year. We dont know. Intel could be in the spot Amd was last year. They want a faster chip but Just bumping the old cpu speed untel they get a better Chip.

This is what fun about Amd And Intel.

 

[TechMan]

The best parts are:
1. they keep on improving their products
2. they keep their prices reasonable

They fight, we win.

 

[P4Man]

It is very real, and its not only a performance advantage, its also a price advantage (and therefore a very significant price/performance lead) and a feature advantage (AMD64, NX, lower power, higher computational density,..) for most parts. In the Athlon MP days they had a tiny performance lead on some benches, and a somewhat bigger price/peformance lead but nothing big, and most of it was offset by lacking RAS features and no tier 1 adoption.

I honestly think AMD has scored a home run with Opteron in the 2/4 way market (and soon 8+), and I just don't see intel matching it until they have a P-M based, 64 bit capable server chip with point to point busses and ODMC (or something like FB DIMM with tons of channels). Won't happen until well into 2005 at the earliest. More likely 2006.

This should buy AMD enough time to eat some significant marketshare from Xeon and stop Itanium from moving down into the more commodity 2/4 way market (perhaps even creap up into Itanium market to some extent).

BTW, significant in my book means something like 20%-25% which would roughly represent a five to ten fold increase in marketshare (depending what market you focus on). I don't think AMD will get much further (nor does it need to), intel just can't afford to lose much more and will undercut opteron as much as it has to to keep the rest. And AMD will happily settle for higher ASPs instead of even bigger marketshare IMHO. Production capacity however, is *not* a concern, AMD as enough capacity to supply the entire workstation/server market with relative ease and if you have the luxury of choosing wether to label a certain cpu as a $900 opteron 848 or a $200 A64 3000+, I think thats an easy call.

In the longer run, I have little doubt intel will fight back and bring out competitive parts, but once AMD is well entrenched in this market, it will be extremely difficult to drive them out again. Which is good for all of us obviously...

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[sonoran]

>>>the only card Intel has left to play is that the 90nm process will enable them to crank out more chips at lower cost

Funny stuff this guy writes. The biggest speed advantage single AMD64 chips have is the on die memory controller. Given how long Intel has been in the chipset business, how long do you really think it would take them to integrate the northbridge on-die with the CPU if they deemed it necessary? And yet they haven't done that. On the ropes? Uh-huh.

On the other hand, hypertransport is a pretty neat trick for multiprocessor configurations. Developing a counter for that might take a little while.

 

[P4Man]

>Given how long Intel has been in the chipset business, how
>long do you really think it would take them to integrate
>the northbridge on-die with the CPU if they deemed it
>necessary?

A year or two ? seriously, there is nothing simple about integrating a memory controller ondie. How much experience do you think intel has with designing 2 or 3 GHz controllers built on 130 or 90nm ?

Besides, integrating the MC is rather pointless for SMP systems if you don't have a fast CPU interconnect like hypertransport. And since intel (like amd) uses the same cores for single cpu as SMP (P4/xeon), there is not much point in integrating a MC only to have to disable it for SMP. So intel either has to swallow its pride and licence HTT, or develop a complete new bus (perhaps based on PCI-E) which will take a considerable effort and ammount of time. There is nothing like that on the roadmaps, so don't hold your breath.

In short, what you are saying is about as pointless as saying "how long you think it would take intel to develop a AMD64 compatible chip, with HTT and ODMC, using SOI, shorter pipeline, better FPU,..." ? Well, a while I would guess. Not much shorter than it would take AMD to clone the P4 netburst design if ever they wanted to. Point is AMD took those design decissions many years ago, and they pay off nicely now. Intel chose a complete different direction, and it payed off until the 3.06 GHz P4, but they crashed into a brick wall after that and there is no quick or easy plan B.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[scamtrOn]

amd will have intel on the rope when it has more market share then intel... until then, no.

</font color=red><b><font color=orange>my sys:
mobo: Abit AN7 @ 442 FSB
CPU: AXPm 2600+ @ 4100+
ram: corsair xms @ 1:1 running 2.5-3-3-6
HDD: two raptors on raid 0

 

[Groveling_Wyrm]

I wouldn't go as far as to say that. Intel still has a very large asset margin over AMD. I would however say that AMD would have Intel on the ropes when it has more market share AND more assets than Intel. But even if that were to come true, it would be years in coming...



<font color=blue> Did you know that 89.72% of all quoted statistics are false? </font color=blue>

 

[scamtrOn]

sure, that works too.

</font color=red><b><font color=orange>my sys:
mobo: Abit AN7 @ 442 FSB
CPU: AXPm 2600+ @ 4100+
ram: corsair xms @ 1:1 running 2.5-3-3-6
HDD: two raptors on raid 0

 

[trooper11]

nah i dont think amd needs to have all that to say it 'has intel on the ropes' ok maybe i should dum it down and say have intel on the defensive. If amd's new pursuits add up to gain in marketshare, lets say up to 25-35%, then id say intel would be in a very defensive mode , since amd has never had that kind of share of the market. I really think its possible within a year to two years that amd could make significant moves in the market if they can stay on track and once they get thier new fab online and running at full steam.

 

[HardWareBoss]

LOL

 

[scamtrOn]

well... amd can do all it can, but there are too many stupid people i'm sure you have come across people who still think amd runs a lot hotter than intel and they burn up... there are still people who think intel is faster than amd in pretty much everything and there are still people who don't understand AMDs pr. the sad part of this whole thing is many of those people are system administrators and such... many have contracts. many of them are bias.

i was helping a friend of a friend to build a pc the other day and i told him to get a AXP since he'll be gaming mostly. he had a real good excuse why he shouldn't; "i just can't rely on a company that uses little rubbery things to cool the cpu" is what he told me. can you guess what he is talking about? i mean how stupid can you be?

i think even if amd does real good, it'll take longer than that for it gain tha much market share.

</font color=red><b><font color=orange>my sys:
mobo: Abit AN7 @ 442 FSB
CPU: AXPm 2600+ @ 4100+
ram: corsair xms @ 1:1 running 2.5-3-3-6
HDD: two raptors on raid 0

 

[P4Man]

I think AMD's slow uptake in the corporate desktop and server market has precious little to do with biased, clueless or fanboy admins/procurement departments, but everything with conservatism (long validation) and even more lackluster oem support.

On the desktop, oem support is still lacking terribly, only HP/Q has a single line of business oriented desktops AFAIK. These machines seem to sell pretty well from reports I read. But neither IBM, Fujitsu-Siemens, Dell or Gateway even offer a *single* corporate desktop oriented machine with AMD chips in them, so its not surprising AMD isnt selling there. You don't need to be a clueless fanboy to choose a P4 (or even celeron) based IBM/Dell/.. desktop over a no name AMD powered device for corporate use.

In the server market, ironically perhaps, AMD is getting better support these days, even though its still trailing Xeon by a significant ammount. For instance, there are still no 2U opterons available from any of the big names. There is only 1U (1/2 way) and 4U (4 way) systems, nothing else. So if you're in the market for a 2U database server (quite typical), there is no AMD option unless you want to settle for lesser names like Appro or Verari. So again, its not surprising AMD doesn't hold a 50% marketshare here

There are of course many reasons those OEM's arent offering more opteron based solutions (yet), but clueless corporations isnt one of the main reasons. Slow product/validation cycles, and high associated costs, lack of support in this area by AMD (very important factor!), intel's comarketing dollars, etc are much more important reasons.

>i think even if amd does real good, it'll take longer than
>that for it gain tha much market share.

Depends what you consider "tha much market share" . I think Q2 or Q3 market reports ought to give us some indications already. But again, as long as IBM only offers a single HPC/cluster oriented 1U opteron, Sun only a single 1U opteron, leaving just HP to offer 4U/4way servers, market penetration just can't go all that high. Once Sun launches its 2U and 4U products later this year, and maybe IBM as well, if/when HP extends its opteron product line, then you might start seeing some uptake real fast. Corporations arent nearly as paranoid about the cpu manufacturer as they are about the oem supporting the system. Otherwise no one would even bother selling VIA or transmeta based blades for instance.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[trooper11]

Well you forgot about Tyan now offering 2U opteron racks, so things are slowly turing and your right, as Sun and IBM introduce thier opteron solutions, marketshare uptake oculd move along nicely.

I think you also forget one thing about these corporate markets. Yes they are worried about the oem supplier, but in the end its thier bottom line, how much does it cost that drives thier purchasing choices. I think once amd gets picked up more, these corporations will see the greater value that opterons can offer, unless intel wants to just give away xeons or itaniums lol

 

[Cybercraig]

A lot of the Corporate market is driven by people who must "watch their azz". A friend of mine is an IT director for an Energy trader. The INTEL vendor came in for a demo a couple of years back. He had two machines, one with an AMD set-up and one with INTEL. After booting them up, he disconnected the CPU fans. The AMD was gone in under a minute, the INTEL kept running for over 30 minutes. Did that have an effect? HELL YES. If those machines go down because of a change that Rod makes, he is FIRED! It's plain survival. Nobody gets blamed for sticking with INTEL.

"I am become death, the destroyer of worlds. Now, let's eat!

 

[P4Man]

>Well you forgot about Tyan now offering 2U opteron racks

Since when is Tyan a tier one server vendor ? or tier 2 for that matter ? Or three ?

>I think you also forget one thing about these corporate
>markets. Yes they are worried about the oem supplier, but
>in the end its thier bottom line, how much does it cost
>that drives thier purchasing choices.

Corporations don't look much at purchasing cost, but TCO. The vendor and its support contracts, guarantees, validation, etc plays a HUGE role. Tyan might be considered for development boxes, HPC, etc, but not to run SAP or something. And even there, most corporations will even prefer to purchase all their kit with the same oem (or as much as possible), remember the story of Dell supplying opteron cpu's upon customer request ? Well guess what, Appro, Verari or Tyan or not likely to win this sort of contracts. And therefore, neither is AMD. The key to this market is not price or performance, its oem support (though obviously, price/performance is a reason for an oem to consider or ship a system based on your cpu). Once you have the oem's onboard, you can win deals on price/performance, not earlier.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[trooper11]

i never said tyan was a tier one or whatever oem, just that they now offer them, as other comapnies are planning to. its just a sign of upturn. be as negative as you want, but to me it looks positive.

it seems your worried about this, i oculd be wrong though. i dont think its time to worry, things liek this take lots of time, more time then has elapsed. These things dont happen overnight, especially when your dealing with marketing and entrenched products that comapnies are use to dealing with.

As far as price goes, I did mean price for the oems, not the end users. If amd can ramp up production and keep pricing agressive tot hem, then they in turn will push them to thier end users.

in the end, things dont look bleak, lord there is more support for amd from oems then ive seen in a long time, and indications are it will be increasing. just give it time.

 

[P4Man]

>it seems your worried about this

LOL, I don't need to worry, I'm not an AMD shareholder anymore No, my comments where just in response to scamtron who blamed the slow uptake of opteron to "fanboys" in the corporate purchasing department, while reality is quite a bit more subtle. So, don't get me wrong, I share your overall optimism about opteron's outlook, but it will take time, and it will require additional products by all the big oems (ignoring Dell) to get anywhere near Xeon in marketshare. FWIW, I find Opterons uptake in just 12 months nothing short of spectacular so far, and it exceeds even my most optimistic projections.

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[darko21]

<A HREF="http://www.theregister.co.uk/2004/06/02/sun_shows_metropolis/" target="_new"> Sun goes back to the future with Metropolis </A>

Sources at Sun Network told The Register the box should be generally available next month. Sun initially placed a limited number of orders for the kit but has already received higher than expected demand, according to the sources. One unnamed customer has pledged to buy more than 5,000 boxes - a figure higher than Sun's original order. When the system starts shipping in July, Sun will offer it with the Opteron 250 processor as well.



If I glanced at a spilt box of tooth picks on the floor, could I tell you how many are in the pile. Not a chance, But then again I don't have to buy my underware at Kmart.