> Given the miniscule percentage of the market that upgrades
> hardware, I'd have to say the lack of large memory options in
> configurable PCs today points to a lack of demand, and that
> pretty much renders the 64-bit option irrelevant for consumer
> PCs for the time being. Even if someone were to buy a 64-bit PC
> today, they couldn't get the memory to future proof it in the
> way you are arguing that a 64-bit CPU can future proof it.
This isn't true. There are systems being sold today with 256 MB of memory on the cheap. Do they page and swap? Sure they do. Is this ideal? No. But it works for some customers. And you could extrapolate that to 64-bits as well.
As far as the upgrade considerations go, consider that the new
WMF hole in Windows 98/ME is a killer problem and that Microsoft plans no fix for it. ars technica is recommending that users upgrade. An XP upgrade is $100 to $150 retail to
just home. In most cases, users would be better off upgrading
their hardware. If XP Home loses security updates in 2007,
that's another reason to upgrade hardware. Especially if that
forces you into Windows Vista. Windows XP Pro is an option
but you're paying $150 to $200 of something that's going EOL
two years later.
> My argument is to go the dual core route, because at least
> then you have applications and an OS that already supports
> it, with more to come in the future without needing
> expensive hardware upgrades.
If Microsoft is serious about dumping XP Home in 2007, then you
don't have the OS.
> I shutter at the thought of memory bandwidth being limited
> by an I/O pipe the width of USB. USB can only sustain about
> 40MB/s of bandwidth, and you want to use it as a 64-bit
> memory expansion? Ugh.
Do you do performance enginering? I do and I'd be happy to have
USB 2.0 performance for paging.
AI
Market Data 
Markets 
