> I'm sure that the board is growing weary with me about now, and,
> my intention is not to drag the board down, so, I'll allow some
> posters that are more knowledgeable than I to post for a while.
> Thanks all for your input and I'll keep reading even while not
> writing...
There are some very smart posters here that are very knowledgeable
in their areas of expertise and this place is quite valuable because
of that expertise. The thing is that it can be very, very hard to
translate twenty to thirty years of expertise and knowledge into
a form that a non-technical person can understand. And some folks
aren't inclined to try too hard.
My area is software and I know some of the hardware stuff but I
will almost always defer to chipguy and others on hardware and
architectural stuff. They know the stuff in far greater detail,
often at a level that I wouldn't be able to understand, even with
a moderate amount of studying. They also know manufacturing, costs,
markets, etc. Stuff that you can't just read off a Yahoo Finance
page.
That said, we can be wrong. Someone could be developing a product that we don't know about. But a lot of the big picture stuff isn't that hard to figure out if you've been in the industry for a while.
As far as your processor scenario goes: I have a web browser, email client, some widgets and a Tarantella session running on my quad-core Sandy Bridge system. These are all integer applications. These programs really don't benefit from GPU acceleration, at least in the way that I use them. How do you convince software writers to port their software platforms to your processor architecture? You're talking about hundreds of millions of dollars over several years of development. Microsoft apparently thinks that it is worth it. Some other software vendors may ask you for money to port to your platform if it isn't already a standard platform. Will nVidia pay? Will nVidia start up their own compiler group to build optimized compilers? Intel has such a group - they're so good that AMD runs their benchmarks using Intel's compilers. AMD could just go with GNU on the cheap but a good compiler is a good compiler.
I worked at DEC for a long time and did porting work from VAX to Alpha, Unix and NT. We got our product working on NT after a long effort only to see the platform (Alpha/NT) scrapped. It's hard if you're not the standard platform.
So don't take many of personally. Engineers often have abrasive personalities - perhaps because they know so much and don't always like to explain the basics, especially if it is likely that the other person won't understand them.