Originally Posted by icehotshot
If it was common knowledge then I would think I would know of this.....not everyone was alive/into computers 25+ years ago to know this.
Again, some actual proof would be nice instead of just your words. But whatever it really doesn't matter.
There really isn't any evidence-based proof to present because it's not a point up for debate. If you look at the news articles from the introduction of the Core or Core 2, you will find mentioned that it was heavily based on P6. I'm not saying that Nehalem is a Pentium Pro with an IMC strapped to it, but it is the result of improvement piled on improvement that can be traced back to the P6.
All that matters is AMD needs to completely overhaul the bulldozer architecture if they don't want to collapse.
This is just simply not true. The ideas behind Bulldozer are solid, and it will likely end up being a good bet in the long-term. That the initial execution/implementation isn't good implies AMD's collapse is absurd. NetBurst is the best example of a new-from-the-ground-up architecture like Bulldozer and it was pretty terrible at first too. But it was improved to be pretty competitive in its lifetime.
I don't see why they even came out with a new architecture if it was worse than the previous one. It doesn't make sense at this point.
Most likely, the engineers decided it was better to take a relatively short-term hit to better the product in the long-term. I don't doubt that AMD could have continued to improve the performance of K10.5, but there are limits. The two big ways of improving performance is to improve single threaded performance or to improve scaling. Improving single threaded performance requires logic to reorder instructions to execute in the most efficient way possible and extract instruction-level parallelism. At some point, there is a limit to the amount of ILP you can extract and benefit from and the closer you come to this limit the greater the diminishing return on improving your reordering logic. In other words, the faster your single threaded performance gets, the more expensive / time consuming / difficult it is to continue improving it. AMD was probably getting to the point where K10.5 was beginning to suffer from these diminishing returns.
Originally Posted by Dt_Freak1
this is what i dont get......its been proven that intel keeps recycling OLD tech as demonstrated here by darkcloud89. I remembered reading somewhere that the technology that powered conroe was based in the old p3 era but modernized and reupped with "moar cores." ...
Nobody is just recycling old tech. It's an extremely expensive and time consuming process to create a brand new architecture. It's faster, easier, and more cost efficient to take proven designs and improve them / remove bottlenecks. It's not like Intel just took the Pentium Pro, slapped an IMC to it and dubbed it i7. In fact, you'd be hard pressed to independently look at a p3 and an i7 and determine they have a shared ancestry.
Originally Posted by Clairvoyant129
Intel revolutionized personal computing. It's obvious you clearly don't know your history. Maybe you should stop getting your news from AMDplanet.com?
Except AMD is responsible for introducing 64-bit and multicore to the consumer desktop, things we take for granted today.
Originally Posted by erunion
Huh? Its still an x86-64, with roots back to the 8086.
x86-64 is an instruction set, not an architecture (though it is commonly called that).