Replies to post #78475 on Intel Corp (INTC)

[Tenchu]

Rudedog, > So the choice for Intel architecture in that scenario is Dunnington, not Nehalem.

WTF? Dunnington is in a VERY different segment of servers than Nehalem, i.e. low volume and high price.

So there are migration issues when going to Nehalem. Doesn't mean customers with legacy Intel servers are going to go to AMD, nor are they going to jump up to the pricier Dunnington servers just to stick with yesterday's virtualization.

That leaves AMD with only those customers who can't easily replace their legacy AMD servers. That's not a formula for growth. That's a formula for inertia, which is exactly what AMD does NOT need if they are to return to profitability.

I respect your opinions, but I really don't think you're thinking through them. Instead, I've seen you move from "AMD has a bright future ahead in virtualization" to "AMD at least can serve its customers who already have AMD servers."

Tenchu

[Saturn V]

Hi Rudedog,
A few months ago you claimed that with the loads you were benchmarking ( VM of 8+ copies of SQL) , there was a much larger access time penalty with Nehalem than with Shanghai. Does that still hold today ? The published VM benchmarks give the clear edge to Intel, so your old statements may not be valid with production Nehalem silicon.

According to you those shops which are AMD shops are going to continue with Shangai, because of greater compatibility with the VMs of vintage AMD servers. This is analogous to the mainframes which have survived because of compatibility issues with vintage software. In this situation does it make sense for AMD to drop the prices of Shanghai to compete with Nehalem, since a lot of its customers are not going to switch anyway ? By mantaining its prices it can milk its customers who cannot afford to do a wholesale migrationn to Intel. Lowering prices on Shangai may not win any revenue on the few greenfield projects, and lower its revenue on sales to legacy sites.

[wbmw]

Rudedog, there is nothing I disagree with in terms of what you said.

Unless I misread the tone, it sounds like you are taking a defensive stance, but there's no reason for that. I am not being accusatory to you, but merely stating a set of facts.

1. Your data is not verifiable to anyone outside of you and the clients you represent. You've pretty much said the same thing.

2. If your data isn't verifiable outside of you and your clients, then no one has any basis to say that your findings will have a larger effect on customer buying patterns over the breadth of industry benchmarks that tell a rather uncontestable story about Nehalem's superiority.

Not that you don't bring up good points, but the summary of your points that I read is that AMD's value prop is mostly relegated to those companies already invested in AMD servers (based on Barcelona and beyond) that can be used as a common installed base for virtualization purposes.

Outside of that good point, I think AMD's market has shrunk to something very small, given Intel's new product, and any claims from people that AMD still has an ace in the hole, are mostly constrained to a small niche of the market.