Blauboad, Re: As far as porting goes, you've hit upon one the biggest selling points of AMD64--the ease of porting to it from x86. It took one programmer one week to port all of the latest Unreal game. One guy, one week. Sure there's testing that needs to be done afterwards, but it just isn't this massive undertaking that you're trying to suggest. Makes great FUD, though.
Unreal is a very easy example, and by far not the common case. Epic Games has been working with AMD for the past couple years on optimizing for their micro-architecture. You can see the results of this in benchmarks. It's no surprise, therefore, that their code was all ready to be ported when AMD decided to make them their poster child. One developer, one week: sounds kind of heroic, doesn't it? It sure does fool enough people, though.
Most applications are not going to port as easily as Unreal did to x86-64. Photoshop, for example, has several hundred DLLs, and all of those require porting to 64-bits. 64-bit Windows will not support running 32-bit DLLs in a 64-bit app, so it's either all or nothing. AMD might be successful in leveraging off of work that Intel already does. Applications that have already been ported to IPF (such as DB2) will have a much easier time being ported to x86-64, since Intel already did all the heavy lifting. AMD might want to use this leverage to go after apps such as SQL Server, SAP, or Oracle, which have all already been ported to IPF. In fact, I expect these to be the next targets.
However, when it comes to out-porting Intel, I think AMD will have a very tough time. It's not FUD; it's the truth. I just wonder how AMD expects to sell a 64-bit story if they are always going to be behind Intel in the porting effort. I also wonder how they are going to sell 32-bit performance compatibility once Intel gets the IA-32 EL in the Windows 2003 Server: Service Pack 1. Quite possibly, the market has room for two 64-bit mainstream architectures, but it should be interesting to see how it plays out.
