Re:" However, AMD has been able to develop design methodology, design implementation, process technology, etc. which resulted in the K6 (far superior to PPro in my humble opinion) in a fraction of Intel's time..."
"NO! AMD sh*t canned their original K6. The current K6 was designed by NexGEN, which AMD bought in January or February of this year. I would hardly describe this NexGEN/K6 purchase as starting from Square 1."
Paul,
Please read my post more carefully. I said "design methodology, design implementation, process technology." These techniques have to be developed for a process technology. "NexGen" was fabless.
I have read your prior posts and there is something that I do not understand. Your loyalty to Intel is obvious. For this, I understand your desire to discredit AMD. However, you are trying to portray AMD as weak with no technological ability at all. You are implying that AMD shelled out some cash and a production level K6 magically appears. I believe all that was purchased was a conceptual design and some more human resources. Based on your prior posts, I believe that you have some technical knowledge. Because of this, you should know the tremendous effort it takes to go from a design (even a fully simulated one) to actual production. There are a large number of obstacles that AMD had to overcome to make this possible. I can promise you that many design modifications had to be made and the final design differs greatly from the "purchased one" The solutions to these obstacles can only be attained from experience; they cannot be purchased.
I am currently working on a chip in which the design will be finished relatively soon. However, it will not be production ready for another 2 years. Even though we have designed chips similar to this in the past and should have considerable "experience" there is a whole new learning curve to master due to a slight variation. I do not think you understand the daunting complications involved in full custom, deep submicron IC design. Even designing and implementing the simplest circuits are a real pain when you take into account timing, parasitics, etc. Designing in deep submicron is extremely tricky. The purchased K6 was no where near perfection and much effort was required to get the thing to work. Work performed by exempt AMD exployees.
The design is only half the picture. New device technologies have to be "debugged" in order to increase yields. Even if you have debugged your .35 micron fab, it you plan to run a different product through it, you have to debug specifically for that product. If AMD "bought" the K6, why does it take so long from the time of purchase to the time we actually see a K6? It is about the same amount of time it take Intel when they start from scratch.
Most AMD engineers I know (including the considerable number of ex-Intel employees) hold advanced degrees from the MIT, Berkeley, and Stanford. These guys are not the incompetent idiots that you would have people believe. The cause of AMD's product delays are a result of lacking human resources. AMD, as a whole, is only a small fraction of Intel's size. Within that small fraction, an even smaller fraction works on the K-processors due to AMD's diverse product line.
In regards to the K6 being superior, you'll see early next year. The only thing Intel can do to stop it is to bully their customers as in the latest incident of Intel not allowing them to use a "competitors" logo (gee, guess who).
I have great respect for AMD and what they are trying to do. If one of the world's most important global industries becomes completely dependent on a single company located in Santa Clara, the loss will be ours. |
