Sign in with
Sign up | Sign in
Your question

DELL says NO to OPTERON

Last response: in CPUs
Share
October 10, 2003 12:04:17 PM

"""To go with an AMD solution becomes problematic for us from a cost and kind of an ice-breaking standpoint," Rollins said. "We believe 64-bit will be a technology customers will want and will migrate to it. It's a market that's emerging but not quite ripe yet." "

Translated " My daddy Intel will be very cross with us if we use AMD. We do not wanna anger daddy"

More about : dell opteron

Anonymous
a b à CPUs
October 10, 2003 12:08:40 PM

alternate translation: "man we would really love to sell those opteron machines, our customers are asking for them. But we can resist just a bit longer while hoping intel comes out with their own x86-64 chip real soon"

= The views stated herein are my personal views, and not necessarily the views of my wife. =
October 10, 2003 12:21:50 PM

Computers should be custom assembled, not ready-made appliances. Dell should stick to making TVs and toasters.

<font color=blue>
A woman driving is like a <font color=yellow> star </font color=yellow>in the sky, you can see her, but she can't see you.
</font color=blue>
Related resources
October 10, 2003 1:38:00 PM

Another translation: when compaq started making ipaq, michael thought it was a stupid idea, and now we make second rate pdas. This time hpcompaq is doing the same with with 64bit machines. Michael still thinks it is a stupid idea. Maybe if he spat out barretts dick out of his mouth for one second...

I wonder if Michael gives better blojob than carly? I'll have to try carly someday.
October 10, 2003 2:07:01 PM

I've got to hear the reasoning behind this... explain (keeping in mind that most people aren't capable of assembling thier own computers, most people have no desire to, and that oft times its cheaper to buy a prebuilt than it is to buy even the parts at wholesale.)

Shadus<P ID="edit"><FONT SIZE=-1><EM>Edited by shadus on 10/10/03 11:12 AM.</EM></FONT></P>
October 10, 2003 3:05:45 PM

Quote:
"To go with an AMD solution becomes problematic for us from a cost and kind of an ice-breaking standpoint," Rollins said. "We believe 64-bit will be a technology customers will want and will migrate to it. It's a market that's emerging but not quite ripe yet."

Actually, I think that's a pretty reasonable and accurate description of the current situation and personally I am of the opinion that <i>only</i> an AMD fanatic would see it otherwise. Why?

1) The software just isn't there <i>right now</i>.
2) Business (especially server) customers want <i>assurance</i> that the product will stand the tests of time. <i>Right now</i> Opteron and A64 are still too new to give that assurance.

So I see it as perfectly understandable that right now at this very day and time there are major OEMs who see AMD64 as still too much of an emerging product to spend resources into developing systems that use it. It is exactly "<i>a market that's emerging but not quite ripe yet</i>". It was never said that AMD will forever be ignored, just that <i>right now</i> is not the ideal time to jump in.

<pre><A HREF="http://ars.userfriendly.org/cartoons/?id=20030905" target="_new"><font color=black>People don't understand how hard being a dark god can be. - Hastur</font color=black></A></pre><p>
Anonymous
a b à CPUs
October 10, 2003 3:23:24 PM

>1) The software just isn't there right now.

There is plenty of software already. Gazillions of 32 bit x86 programs that perform superbly on opteron; why does everyone always forget Opteron is a stellar performing 32 bit server cpu ? And a decent ammount of 64 bit apps are available where it matters; to name just one, do you have any idea how many DB2 installations are deployed world wide?

Also, if you buy a server now, you will not want to replace it next year. those machines are often bought to last many years, so customers will have plenty of time for 64 bit OS's and apps to arrive (or rewrite their own) while happily running in 32 bit in the mean time. Not buying a 64 bit machine now because these apps arent there yet only applies to IA64, not to opteron. You could claim the Itanium market is emerging, but not ripe quite yet, and it would make much more sense than claiming the same about opteron as a product

>2) Business (especially server) customers want assurance
>that the product will stand the tests of time. Right now
>Opteron and A64 are still too new to give that assurance.

Customers don't give a rat's ass about the cpu manufacturer. They will however, want to buy from a reputable vendor that will still be in business in a few years to support the machine. Also, consider eg 2MB Xeons are newer and less proven than opterons; unless you refer to the ISA, in which case both use the more than proven x86 ISA. One also supports another, which you can ignore if you like, or test it extensively and be ready to deploy if testing proves it to work okay for you. I really see little reason for customers not to buy opterons instead of Xeons, except for the oem support which is still lacking big time.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
October 10, 2003 5:36:29 PM

Quote:
There is plenty of software already. Gazillions of 32 bit x86 programs that perform superbly on opteron; why does everyone always forget Opteron is a stellar performing 32 bit server cpu ?

Who ever said anything about that not being the case? <i>Most</i> businesses looking for top-end boxes and servers however are <i>not</i> going to be thinking about 32-bit software.

Quote:
And a decent ammount of 64 bit apps are available where it matters; to name just one, do you have any idea how many DB2 installations are deployed world wide?

And that would be <i>what</i> percentage of Dell's market exactly?

Quote:
Also, if you buy a server now, you will not want to replace it next year. those machines are often bought to last many years, so customers will have plenty of time for 64 bit OS's and apps to arrive (or rewrite their own) while happily running in 32 bit in the mean time.

That makes absolutely no sense. Why would someone buy an Opteron server for software that will eventually arrive when they could buy a 64-bit machine from IBM or Sun (or dare I suggest even an Itanium) and run 64-bit software <i>now</i>?

Quote:
You could claim the Itanium market is emerging, but not ripe quite yet, and it would make much more sense than claiming the same about opteron as a product

I could and would. As far as the server industry is concerned Itanium <i>is</i> still an emerging product. At least it however has had some time now to gain a little software, trust, and ground. Opteron hasn't even got that much to its name yet. All that it has is good 32-bit performance and a <i>hope</i> for a solid future.

Quote:
Customers don't give a rat's ass about the cpu manufacturer.

What world do you live in? I've personally been a hardware purchase assistant in three different jobs now and I've never once heard anyone <i>not</i> give a rat's ass about the CPU manu. Hell, I've fought for AMD (and lost) more times than I care to even count <i>because</i> business IT departments <i>do</i> care about the CPU manufacturer <i>too much</i>. And don't even get me started on reccomendations to friends and family!

Quote:
I really see little reason for customers not to buy opterons instead of Xeons, except for the oem support which is still lacking big time.

As an enthusiast, I don't argue with that one bit. I <i>completely</i> agree. But we're the rare educated few that <i>aren't</i> biased. Most customers either aren't educated or are biased. Because of that there isn't a strong demand. Because of the lack of a strong demand there isn't a strong reason to be a supplier.

A good 90% of SOHO users are going to see 64-bitness of their CPU and demand a 64-bit version of Windows to run virtually non-existent 64-bit software. Not because they need 64-bit software but because they're undecucated and they paid for a processor that can run it, so in their minds they're being ripped off if they can't.

And a good 99% of businesses looking for 64-bit workstations and servers (and how many actually look for a 32-bit server these days?) are going to want their 64-bit software <i>now</i>. But the software <i>isn't</i> all here <i>yet</i>. Why should they wait for that software when they can go to a <i>different platform</i> and get that software <i>now</i>? To businesses time <i>is</i> money and the few simple bucks that they save on hardware is nothing compared to the money lost from <i>any</i> extra time spent on a solution.

No one is saying that AMD64 is worthless or that it's not a good 32-bit processor. They're saying that the <i>market</i> is not ready <i>yet</i>. So what's the big deal? So Dell waits six months or a year?

<sarcasm>Oh god! They <b>must</b> have Intel's <b>i</b> in their <b>e</b> (it's a visual thing) because they couldn't <i>possibly</i> see that there just aren't enough customers demanding it <i>yet</i> to be profitable to invest in opening up some AMD64 product lines to compete against similar lines from other companies in a market with such a <i>presently</i> weak demand. God no! <i>That</i> might make it a logical business decision!</sarcasm>

Face it folks. It's a perfectly fair <i>business decision</i>. <i>Even</i> if it was wrong (and I don't believe that it is) it's <i>still</i> a <b>business decision</b>.

If you don't like it, then don't give them your business!

<pre><A HREF="http://ars.userfriendly.org/cartoons/?id=20030905" target="_new"><font color=black>People don't understand how hard being a dark god can be. - Hastur</font color=black></A></pre><p>
* Edit - fixed bold close
Anonymous
a b à CPUs
October 10, 2003 6:32:22 PM

>And that would be what percentage of Dell's market exactly?

Fairly big. I am assuming over 90% of those DB2 installations are run on Xeon class machines, and less than 10% on Power machines. So Dell could compete for almost every DB2 server deployed.

>That makes absolutely no sense. Why would someone buy an
>Opteron server for software that will eventually arrive
>when they could buy a 64-bit machine from IBM or Sun (or
>dare I suggest even an Itanium) and run 64-bit software
>now?

Look at it from a Xeon perspective; why would anyone buy a Xeon to run their 32 bit software, when there is no way for them to upscale their OS and app to 64 bit later ? Migrating from 32 to 64 bit isnt done overnight. Sure they could buy Alpha, itanium or power based systems, but if they are migrating from x86 Xeon, their software won't run on those machines. they would need a big bang approach converting everything to another OS, other software, libraries, everything. Opteron allows them to migrate step by step, first the OS, then maybe the DB, then the application server, while still running a few obscure 32 bit binaries at full speed. Those customers may also not be ready to switch to a 64 bit solution just yet, and simply need machines to run their current (x86) business apps, but they may want to prepare the move at the same time. Opteron suits this scenario perfectly. Run production servers on proven 32 bit OS& apps initially, and migrate step by step 64 bits when you feel the time is right, and the requirement there. Switching to IA64, power or alpha does not give you this luxury.

> Opteron hasn't even got that much to its name yet. All
>that it has is good 32-bit performance and a hope for a
>solid future.

Is that better or worse than Xeon having good 32 bit performance and NO hope whatsoever for a 64 bit future ? Also, hope is bit of an understatement. Windows support is on its way, as is Solaris. Linux is here already, as are tons of business apps. Even if your app may not get ported to AMD64, or you do not want to port it, just running the old 32 bit binary under a 64 bit OS will give you tangible benefits, like the ability to run several multigigabyte apps on the same server, or 4 GB per process size instead of ~2 GB. Things like that are perfectly possible and beneficial even now.


>And a good 99% of businesses looking for 64-bit
>workstations and servers (and how many actually look for a
>32-bit server these days?)

I'd say 99% are looking at 32 bit servers. Have you compared Xeon sales with opteron/itanium/alpha/power/parisc lately ?

>are going to want their 64-bit software now.

Like hell. Have you ever witnessed a migration from any server platform to another ? That is extremely expensive, time consuming and painfull. No one wants to migrate if they don't have to. Go talk to PA Risc customers if they like having to move to IA64. Businesses will stick with whatever they are running, as long as humanely possible. Companies running on x86, Alpha, sparc or whatever are still sticking to it, cause they invested huge ammounts of money on the hardware, the software, training of their engineers, and just dont want to give that up. Not until 64 bit becomes unavoidable will they move, and then they will prefer to move to something that is as compatible with their old software as possible. You just don't ditch a few dozen of mission critical x86 servers and server apps and years of tweaking that code to move to Power or Itanium and have to restart from scratch, have to retrain, rewrite, revalidate everything.

Also, 64 bit is rarely an urgent matter of needing it "right f*cking now". Its more a matter that it is on the horizon, business preparing for it. Preparations for such moves can easily takes years.

> To businesses time is money and the few simple bucks that
>they save on hardware is nothing compared to the money lost
>from any extra time spent on a solution.

I agree. price difference between eg an opteron and xeon server is mostly irrelevant. However the cost of moving from x86 to something else is HUGE. Moving it to AMD64 will be far cheaper, as you only need to migrate those elements where 64 bit support is needed (eg backend database), while keeping all the other elements as they where happily running in 32 ibt. That can be an enourmous cost saver.

frankly, the only argument I see you make, is that someone needing 64 bit NOW, Itanium (or power) may make more sense, as a more mature, future proof solution than AMD64. I agree. however,in most situations, they won't need it now, otherwise they wouldnt be buying truckloads of Xeons. Getting opteron's instead and leave the possibility open just makes a lot more sense IMHO.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
October 10, 2003 7:45:04 PM

Quote:
Fairly big. I am assuming over 90% of those DB2 installations are run on Xeon class machines, and less than 10% on Power machines. So Dell could compete for almost every DB2 server deployed.

I ask you how much of it is their market and you come back with how much of their market it <i>could become</i>. Well duh. <i>Any</i> company <i>could try</i> to take a market segment. The likelihood of success on the other hand is why business decisions are made.

As for the rest of your argument it's too contradictory to even be worth debating. If (according to you) 32-bit is what everyone is buying now, and if (according to you) no one wants to move to 64-bit at all, then why even offer a hybrid solution at all? If they're not running 64-bit now and are never going to then what advantage does a hybrid solution even offer? Your view of the situation is rather warped. Almost no one slowly migrates a server. They keep the old server running and drop in a test server until <i>both</i> are running solidly <i>simultaneously</i>. Then, more or less overnight, they switch. It's called minimum downtime. And that's exactly why AMD64 being a hybrid is no more or less advantageous to those customers. All that it does is help is it increases AMD's market spectrum with a single product line (Opteron) instead of with two seperate product lines (Xeon + Itanium). So I highly suggest that you sit down and straighten yourself out before you even try to argue that one any further.

Of course is arguing this even important? As I said, it's a business decision. Dell sees the market as unready for AMD64. They'll act on that. When they see the market as being ready they'll act on that too. Why is it even such a big deal?

<pre><A HREF="http://ars.userfriendly.org/cartoons/?id=20030905" target="_new"><font color=black>People don't understand how hard being a dark god can be. - Hastur</font color=black></A></pre><p>
Anonymous
a b à CPUs
October 10, 2003 8:28:25 PM

let me start by the end:

>Of course is arguing this even important? As I said, it's a
>business decision. Dell sees the market as unready for
>AMD64.

Dell sure has reasons not to sell opteron machines, there are dozens of reasons (like AMD supply issues, intel rebates, r&d costs, etc..). wether or not the market is "unready" is not necessarely the only one, or even the most important consideration to Dell. After all, IBM, Appro or rackserver serve the same market, and do consder it "ready". Bottom line: Dell knows what it is doing, I am not arguing their decission, but I argue their motivation. Intel also says the desktop or Xeon market is not "ready" for 64 bit, and I also understand their motives (Itanium) but hardly anyone believes their official motivation.

>If (according to you) 32-bit is what everyone is buying
>now, and if (according to you) no one wants to move to
>64-bit at all, then why even offer a hybrid solution at
>all?

Either you did not read my post carefully, or I didnt express myself well; I am not saying no one is ever going to move to 64 bit, I am saying most business arent moving yet. Not because they don't see benefits from that move, but because the associated costs outweigh the benefits. If you ignore opteron, the only way to get into 64 bit computing was throwing out everything you have, forgetting everything you know, and restart from scratch. That is one serious show stopper.

Opteron enables a slow migration, and allows customers to capitalize their existing investments in tools, software and people. They can already buy the servers now to replace existing, aging x86 servers (opterons arent more expensive than xeons), and then, when or if they want to, slowly start testing their existing 32 bit x86 software under 64 bit OS's. Once that works, they can try to replace their 32 bit DB with a 64 bit one, then rewrite/recompile/reinstall whatever business app they run, etc, etc. slowly. You can not do that with Itanium or Power. It makes no sense to buy a Power machine and test the x86 app under 64 bit AIX. Guess what, it wont run ! so that means buying a power machine, retraining your staff, learning a new OS (wont run windows either), recompiling rewriting all your apps, tools, and/or basically redoing for example your SAP, Siebel or Oracle implementation on which you just spent millions, probably replace a ton of smaller apps/tools that are not available for power, but that you used on x86, and only then can you test, and maybe you don't like the result ? well congrats, you just wasted a ton of money. If you don't like the 64 bit opteron results, don't have faith in the 64 bit OS, the future of AMD64 or whatever, you wasted close to nothing, you still have capable x86-32 servers you needed anyway, and you could look elsewhere for your next platform.

All this are reasons opteron makes sense to business, therefore however, not necessarely to Dell. I don't see them loosing a lot of revenue to opteron suppliers just yet; they have some time. One or two years from here, situation may be radically different though, and Dell may be forced to adopt some 64 bit x86 platform, either from intel, or AMD if it has to.

Bottom line: Dell's decission isnt final, but for now probably makes a lot of sense. That doesnt mean Opteron doesnt make sense to the market however, or that the market wouldnt be "ready" for a good 32/64 bit hybrid.

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