<< So, let me see if I understand your argument: Dell shouldn't have second sourced K6 because its performance was not upto par competitively. To extend that logic, would you argue that Dell shouldn't use Celeron and use Duron exclusively because Celeron gets scorched by Duron? >>
No, I wouldn't for a variety of other reasons. One is an active decision while one is a passive decision. Just as if Dell were to adopt AMD it would become more difficult to drop them, because status quo would be to keep using AMD.
Thus, to make an active decision to add AMD, you need a good reason to break the status quo mold. AMD had a bad manufacturing record, and less competitive product offering. There simply is no compelling reason to change the status quo. (again at the time of the k6-2)
Now in the case of dropping celeron. You have a manufacturer you have a long and productive history with. Who has historically always had the best product and has been a good manufacturer. They supply you're entire business, there is no other supplier who can give you enough volume to meet your needs, and many people prefer their brand to the competition. They currently have a less competitive line up to the competition, and in the recent past you've gotten less supply then you've needed.
There are definitely some bad things in the relationship, but Dell can clearly not go on without Intel (even just the celeron lineup). At this point, I make the active decision to take a change on the second source, but there's no way I go for the gusto and get rid of the celeron.
<< I never saw Celeron and K6 next to each other but based on my recollection of standard non-FPU benchmarks is that they were pretty darn close - assuming both systems are identically configured. >>
If you throw out FPU, then you're throwing out the majority of intense benchmarks right off the bat. Even so, I could tell the difference running an mp3 player and a web browser at the same time. (maybe the mp3 player is fpu intensive, i'm not sure)
<< K6 had over 50% retail laptop market share for several months. Which tells me that any problems you had represent the problems beyond 3rd standard deviation. (i.e, doesn't matter for the mainstream users). >>
The fact that it was sold in a retail store or that it was used in laptops doesn't speak much for compatibility. It says it was very stable under those circumstances (i never argued it wasn't). I said that it had compatibility issues. If the pc doesn't work with a new video board, compaq doesn't have to take it back (neither would dell), but it certainly soils the experience with that particular customer, and I don't think compaq cares, and I think Dell does. (in my experience, dell has much stronger customer loyalty then compaq)
<< The only reason to take "wait and see" attitude, which is what I have been preaching on this thread for a while, is because of new features such as multi-threading which have not seen the light of the day outside of Intel and because Intel has software muscle that can make lower-IPC architecture look better than it actually is. >>
Well there are several more reasons to "wait and see". 1) You don't know how much lower the IPC will be. 2) You don't know how high the processor will scale. 3) If the SSE2 instructions make the chip faster, and they are used in driver software to improve performance of games substantially then they will have a real impact. |
