Regarding my #2 being wrong, there is nothing in your reply that rebuts it. Intel needs to be able to continue providing excuses for continuing computer/chip sales, that is capitalism and there's nothing wrong with it (except from an environmental view!) One needs to upgrade to run the latest version of the bloated and inefficient MS Windows.
"Even lowly i5 CPU can boot Win7 is under 10 seconds with an SSD and can run any game and decode any video with ease."
Right, if it's a clean install. So you agree with me. Without hardware compensating for the additional bloat we have a turkey besides the number of cores on a CPU does not make any great affect on boot time. As to running any game or encode/decode video, these are task best left to your video card not the CPU -- and the last time I looked, gamers were still overclocking because they had to for improved gaming.
"Most people may not need a quad core, but even a dual core makes your OS experience seem much more responsive."
Very few, if any people need a quad core; absolutely agree. And for those that do, what software out there efficiently utilizes any SMP based machine? Nothing I know of for the typical user.
As to dual core, as per your previous statement, an SSD is what really would make a difference -- not dual core. Besides, again, what software make efficient use of an SMP dual core?
"AMD and Intel do not "convert" their chips into single cores." True, but it's a pretty good layman's example of what is happening; some cores are shutdown with power boosts to the remaining core(s) increasing their speed. Any way you look at it, you just bought cores that are not needed and can slow down your system.
"It's just logical that if you're loading the whole CPU, why not make one part faster." Actually that is exactly what Intel used to do till some thermal barriers were hit forcing Intel to move to multi-core. Logically, and in the hardware/software world I grew up in -- Intel and embedded -- the first thing you do is get rid of the inefficiencies so the system can handle the load (as in proper design) Logically, as well, if by speeding up CPU you can make the system more efficient why have more cores in the first place? Doesn't it strike you as odd that after all the hype about multi-core, the latest and greatest is the ability to not use multi-core? There's not much software out there that uses multi-core, efficiently or otherwise. Again, as you stated, "Most people may not need a quad core"
All the discussion about how to increase throughput for a single thread is pretty much not relevant. I will state that the best ways to increase throughput for a single thread are by optimizing architecture, design (eg algorithm chosen) and implementation. Unfortunately in the PC world (thank you MS) engineering is more often hacking/programming than real engineering. (See TGMLC.)
"In the end you have two things. 1) single threaded performance about hit a brick wall and was going almost no where. 2) If you don't go SMP, someone else will and you'll die a corprate death for not advancing."
1) single threaded performance did not hit a brick wall nor was it any where near hitting a brick wall -- I/O bound was/is a much greater problem. The CPU thermal barriers to increased speed have nothing to do with the level of parallelism of any program, process, or task.
2) Maybe. In the server world the quest is to find the best all around solution which may or may not be SMP on a single core. In the end user world, it a question of marketing and, right now, marketing is pushing SMP on a single chip. (See TGMLC)
Finally, why add complexity where none is needed. Many of the difficulties of parallel programming are quite well documented. Here's yet another: http://arstechnica.com/security/news/2010/05/multicore-cpus-move-attack-from-theoretical-to-practical.ars