Last reply for my free account today and then back to lurking, so I'll hit more than one message:
Message 22530960
I am not a IT xpert nor do I play one on TV but doesn't this presume a large enough output..seems like it will be only 25%
at 1/1/2007
Implied in "higher end" was "higher priced," which is alligned with lower supply.
Message 22531206
That will be the case with at least HP.
We really don't like HP's A64 offering today but will be checking to see what the next product cycle brings. We would rather see a smaller form factor case, and that's why we made the mistake and went with Dell last time around -- better package.
Message 22531375
AMD's stock price will be in single digits if in q2 they show a small YOY profit increase.
YOY for Q2 will be large, but discounted by the street due to the hammer that everyone expects to be falling. I was really speaking about Q3 and even more about Q4. I expect smaller increases here due to last year's strong performance, particularly in Q4.
wow, most of the IT people would not even consider anything that is even perceived as bleeding edge. Out of curiosity, if you are after bleeding edge, why did not you eliminate netburst long time ago?
I was trying to categorize (generally) types of IT guys. The first group are followers, they just figured out AMD is good and they won't be fast to figure out that Conroe is good, therefore they will be pushing AMD blindly. The next group are smart and know they want the bleeding edge for their frag box at home, and it may well be Conroe, but they will just buy cheap for the office. The final group are the most savy. They will end up with Conroe in higher end applications like workstations, but will end up with A64 on the desktop.
As for why we didn't eliminate netburst long ago, 2 reasons:
1) Commonality. We didn't love the decision but it wasn't a bad enough one to change our path in the middle. Of our 100 machines we will end up with about 30 netburst boxes at the end of the cycle.
2) We're not after bleeding edge for this application -- on the desktop it's bang for the buck. For workstations we actually build our own because we can build a very nice machine for about 50% of what anyone wants to sell a comparable workstation for. The OEM's view that workstation == high margin. We've also built 4 of our own Opteron servers.
Where are we going for the next 12-18 months?
Desktop
I would guess switch A64 desktop this fall if we can get boxes in a form factor we like. If there's a reasonably cheap Conroe box from Dell in a great form factor we would certainly consider it, knowing that prices will come down early next year as volume ramps and we're making a decision that will last through our next upgrade cycle. Also, Dell has REALLY been willing to give away this stuff to us to keep a foot in the door because we remind them that we've not bought a server from them in 5 years even though we've added a new server every 8 months for the last 3. Their problem? No Opteron. Also we've never bought a workstation from them even though we've bought 7 in the last 12 months. Again, they don't have the product. This will be at least a 12 month decision and will involve 20 machines or so, assuming we don't go through another big growth spurt.
Worsktation
A64 will remain our choice for engineering workstations for the near future because of commonality. We view that the boxes we're building with 4800+'s work great for our 3D design and CFD applications, and the boxes we're building are nice and quiet (Asus A8N-SLI Premium for no chipset fan, for instance). We will probably bump up clock speed on future boxes so long as we can do it on the Socket 939 architecture. So in this case, the logical Conroe application, we probably won't bother because we view it's not time to change our platform for at least 6 months. We will probably wait for K8L to make our next platform change.
Servers we don't have any large purchase plans on the near horizon. We still have an old PIII-1000 box handling file and print and an even older box working as a gateway server.
Low Level Server
We might add another 1U single socket dual core Opteron box this fall to replace our ancient gateway server. Configuration would be identical to an existing box.
Mid-range Server (for us)
No plans in the works, we have a dual Opteron 250 and dual Opteron 242 box running applications and exchange. Both are adequate at this point in the game.
High end server (for us)
Our MRP database server is currently a dual Opteron 252 box with plan to upgrade processors to either dual core (285/290 range of product) or maybe simply to the 3GHz 256 product depending on whether we decide more frequency or more cores is most useful.
We don't play in the 4 socket world yet, but Opteron continue to own that market. For us Woodcrest might be a logical choice except that we don't have any needs that would drive us to a new platform. The database server is the most critical but likely we'll just drop $1000-$1500 into processor upgrades to get a 10% improvement there rather than buy a whole new box to try for 25%...
So, all of this said, it's likely that many of our decisions will mirror other companies. Some companies may be ready to make a platform switch at the server or workstation level, but many will be like us and with no compelling need to change, especially since it will be cheap to stay with AMD due to pricing pressure, momentum will carry things forward and they will wait for K8L to compare to. |
