Re: IBM's a big company (duh), and has a lot of processor lines. An SMP board for Itanium is, well, an SMP board for Itanium
Dan,
I strongly disagree with your point. The current market problem with servers is heavy R & D costs.
Take this simple example. Say it costs $1 billion in costs to get out a new line of servers. You anticipate selling 1,000,000 systems. Thus, your fixed R & D is $1000 per server. This will cover such things as processor design and operating system maintenance. You would still have the manufacturing costs, etc.
Now, say IBM gets an SMP board out for Itanium. This will cut into sales for other server lines, thus, increasing the fixed R & D per system. By making an SMP board to compete with your own products, you are just hurting yourself.
If you look at IA32 servers, you can see the advantages of spreading the R & D around. INTC takes R & D for the processors, MSFT and Linux companies for the OS. Many companies sport IA32 server lines, thus, R & D costs are absorbed by everyone making the systems dirt cheap compared to traditional 64-bit servers.
IBM does sell tons of NetFinity servers based on the Xeon, but 32-bit servers have severe limitations. When you start to get big, you have to move to 64-bits.
Keep us posted, though, maybe there is a big announcement coming up
I highly doubt this. At one point in time, I think this was what IBM wanted. With INTC's inability to deliver, I don't think IA64 is even an option at this point.
The industry wanted a 64-bit solution from INTC. It's too bad they could not deliver. While INTC has been stuck in neutral for 4-5 years, IBM has made huge strides with copper and the POWER3.
I fear that IBM management is too incompetent to realize the monster they have created with the POWER3-II. The Ultra Sparc III will be taped out on the launch (if there ever is one) so SUNW is no threat for big performance gains over the next few years. The copper based POWER3-II is only used in the high end SP 80 at the moment, but by year end will be in all lines of servers. Over the next 2 years the POWER3-II will continue to ramp up and get faster. With no performance competition, IBM's servers will be destroying anything in sight for some time to come. This is already starting to become clear in the market as for the first time ever SGI was overtaken by IBM in Novembers top 500 supercomputer list. In the second week of June when the new list comes out, I expect IBM to build its lead and SUNW to fall even farther than they did in November. By mid next year, I don't think IBM will have any competition in the high end.
I hope IBM management doesn't turn our server line into another OS/2.
chic |
