>1) I believe that it is of highest importance for AMD to effectively establish X86-64. The Hammer should not be branded as "just" the best IA32 processor (P4 will always be a tough competitor). The Hammer should be branded as a true revolution in the PC industry bringing the ULTIMATE POWER OF 64BIT SUPERCOMPUTERS to the masses.<
This is a two edged sword, just as trying to sell V8 engines to the masses would be a two edged sword. The best approach is not to promote the 64-bitness of Hammer, but to acquire SGI and launch Silicon Graphics as the ultimate in consumer PCs using Hammer. Silicon Graphics branded Hammer based systems should be presented as a step up from Compaq and HP P4 based systems in the retail market.
>2) AMD should create a "LINUX-64 INITIATIVE". They should seriously promote and fund Linux developers. I suppose that they already do so to a certain extent but they should do it much more openly. AMD should talk about it frequently. Everybody should know that AMD is dead serious with Linux for X86-64. *At least* AMD should buy or license the existing, powerful X86-64 simulator for Linux and distribute it freely to the developers (NOT the lame version that AMD has developed by itself).
Of course that will anger Microsoft. Well, that may be the best way to get Microsoft to support X86-64 as well...<
No, no, no! AMD should, after acquiring SGI, reposition Irix as a consumer OS. Silicon Graphics branded consumer PCs should ship in a dual boot configuration, with Irix being the default bootup. Users should be required to actively change their bootup OS to bootup in Windows XP if so desired. This position should be maintained until Microsoft ships a 64-bit MP version of Windows as a consumer OS.
>3) Probably the most important point: There is a very, very good argument put forward by Dean Kent. See realworldtech.com.
Here is the decisive quote: "I am talking about 'visibility'. What I mean is, Intel has a two or three year roadmap they can show to businesses and partners. I assume IBM has one at least that far ahead. AMD only has one that is a year out at best. Whether this is intentional or because they just don't know, it is a problem. This has been mentioned to me by several companies dealing with large OEMs and Fortune 500 companies. No visibility to the future, no deal. If I am going to commit to tens or hundreds of millions of dollars over the next 5 years, I want to know what to expect in 5 years.
If AMD wants to be accepted as a big player, they have to act like one."
I think that there may be a lot of truth in these words. We all know how unreliable roadmaps are. Short-term roadmaps (Palomino...) and even worse long-term roadmaps (Merced...). Nevertheless, having a long-term roadmap tells a lot about the self-confidence of a company. If it is that what the big customers want to see, why not give it to them??
So AMD should make a LONG-TERM ROADMAP. AMD should start to talk about the future of X86-64. About plans for the K9. How totally superior this chip will be. That it will make X86-64 so good that any competition will pale in comparison.
It is almost irrelevant whether there are already any concrete design studies for a Hammer successor or not. Just talk about it. Tell the big companies that throughout the next 3-4 years the Hammer will be by far the highest performer in the market and that in five years from now on the K9 will blow the competition out of the water. That X86-64 is the only viable long-term solution.
Is that unethical? Maybe, but that's the way it goes. Look what Intel has done all the time with IA64 (and also with the P4; I suspect that the recent drop of AMD by certain big OEMs was due to Intel's claims that soon the P4 will totally dominate the Athlon). The actual performance of Itanium is a joke. But Intel talked about it for so long and so convincingly that almost everybody believes that IA64 is the future. WHAT A JOKE!
Again: If companies want to see a long-term roadmap, just give it to them. At least behind closed doors. Talk about some ultra-parallel RISC/CISC ultra-scalar X86-64 architecture, or whatever. No matter what, it just has to show that you are CONVINCED about your plans. That is the point! Show that you are CONVINCED to drive the industry with X86-64. Show that AMD's way into the future of computing is by far the most promising one. And do this with a long-term roadmap (which you can always change if necessary) because only that shows real self-confidence.
By that, X86-64 might become a self-fulfilling prophecy - in just the same way as Intel tries with IA64. But this time AMD may be in the better position due to the evolutionary approach of X86-64. And then the proposed marketing campaign isn't even unethical any more... ;-)<
Unveiling new processors would undermine AMD's ability to sell into the business sector. Businesses like stability. Businesses want a platform that will be around as long as possible. AMD should be mapping out long term roadmaps for existing processors, not next generation. AMD would be much more likely to have success with the announcement of their plans for a 0.10-micron Athlon in 1995 than the migration to K9 in 1995. |
