We've had major announcements from AMD and Intel in the server and workstation space, but across most of the vendors there doesn't seem to be a lot of differentiation between products that utilize the two technologies. However, there is actually a great deal of difference between how Intel and AMD approach an OEM partnership, which should result in more highly differentiated products than we are generally seeing. IBM stands out with its offerings in that it seems to be exploring these differences and optimizing unique products. The result should be offerings that are better optimized against the unique differences that exist between the two vendors. Let's talk about what makes Intel and AMD different and then move into how IBM is leveraging that difference with System x and BladeCenter.
Intel: Power, Volume, Stabilityhttps://o1.qnsr.com/log/p.gif?;n=203;c=204663295;s=11915;x=7936;f=201904081034270;u=j;z=TIMESTAMP;a=20410779;e=i
Intel is the big kid on the processor block; it has the largest market share, the most R&D money, and arguably the most technically advanced offering. It builds products that are highly standardized and developed around Intel policies and guidelines. It also has the largest manufacturing capacity, which gives it volume advantages. This means products that leverage Intel's strengths will be very similar across vendors, because Intel keeps relatively tight control over its offerings, and will typically be differentiated by software, services, and the containers (cases) the technology resides in.
AMD's strength really resides in its willingness to let the OEMs drive the development cycle; it is willing to work more collaboratively on solutions. I've even had one OEM tell me that they can actually influence the instructions that go into the processor. This is the way most markets actually work, where the company building the solution dictates what the supplier will provide and both work together on the result. You get less consistency between vendors, but products that better target unique and specific needs. In short, AMD-based offerings, done properly, are best when used to target the fringes, or unique needs that a more generic offering can't effectively meet. You can still wrap them with similar software, containers, and services, but you trade off absolute performance on any vector against a more customized solution.
AMD vs. Intel
In my view, done right, the two solutions (at least in the server space) shouldn't compete as much as they generally do. If the specification is set properly, the more generic desired result should naturally favor Intel, and the more unique should naturally favor AMD (and the related vendor).
If you look at the IBM iDataPlex dx360 M2 server, you see it designing around Intel while taking advantage of the core Intel benefits. It is designed as a performance powerhouse and it is cased in a way that allows for maximum densities while not degrading the performance you are paying for. Differentiation is outside of the Intel specification and deals with storage, case size and memory configurations. It's generic at the Intel core but configurable outside of that core to meet unique needs and wrapped with IMM, ToolsCenter, and Director 6.1 tools.
Now if you look at the 2P IBM BladeCenter LS42, you see how it designed with AMD. Similar wrappings but the core of the offering is unique to IBM. This product, which ships with two processors, can be upgraded to four processors with no penalty. This is unheard of in this space, as you generally have to pay a large premium to take a two-processor server to four processors. In most competitive offerings, you'd likely be better off just swapping out the 2P server for a 4P. In this instance, you can buy what you need today and expand as you need to without replacing the server or incurring any financial hardware penalty. To do this, AMD and IBM had to work closely together and here the core technology was modified to differentiate the offering and make it unique to IBM.
With the Intel-based product, it focuses largely on performance per watt, which is where the core benefit of the new Nehalem-based Intel offering is. With the AMD part, it focuses on providing the best value for a growing company. The Intel-based server would be too expensive for the buyer who didn't yet need the performance it provided and the AMD-based server's flexibility would be worthless to someone who needed maximum performance on day one.
Wrapping Up: IBM's Clearly Serious About X86
In looking at whether a vendor is serious about a segment, you not only want to see how they spend financial capital but how they spend intellectual capital. IBM is clearly thinking through how to best use both Intel and AMD in these examples more than most other vendors, and the result is not only better differentiated offerings but a greater assurance that IBM places a high priority on this market and is solidly committed to it. In the end, however, it is the example it sets in terms of thinking through how to best use these two chip suppliers that I'd like you to remember. All suppliers have advantages and disadvantages; knowing when and where to make use of them generally results in better decisions and, in this case, better products.