> This was not intended as a Doctoral Thesis. It was a
> simple comparison and I think the majority of the
> improvement was due to the increase of L2, but probably
> not all.
The thing is that you were using speculation to disprove something that Dan said. Dan stated that the difference in performance between a processor with X L2 cach and a processor with 4X L2 cache is minimal. As a counterargument, you submitted a comparison between a PIII and a P3XP. The problem is that the P3XP not only had better compilers, but it had a better defined and probably faster hard drive, sixteen times as much system memory, and possibly a better motherboard. I don't think a doctoral thesis is needed, but at the very least you should not be submitting evidence for your reasoning that is completely full of potential holes. Saying that Dan is wrong because "I think the majority of the improvement was due to the increase in L2" is not much of a rebuttal, and you're basically suggesting that you have no counterevidence to offer Dan's allegations.
> Ali was the first person to grace my ignore list so
> I don't see his posts.
I'll keep that in mind. I can understand your reason for doing that, though. I was going to respond to your message, but I stopped when I saw Ali's.
> What I tried to prevent was any suggestion that the
> extreme lithography techniques used by AMD could be seen
> as superior to those used by Intel. The question seems to
> be, if AMD can produce 70-80nm gates, why can't Intel if
> Intel's process is equal or better? This question needs to
> be put into perspective.
The thing is: You are taking as given some things which to me seem like speculation. You make more than one reference in the above about AMD's process techniques being "extreme" and Intel's process techniques being "conservative". Could you elaborate on that? From what I can see, each process has different advantages over the other. I have been told by those more intelligent than I that Intel's current 180nm process has more aggressive transistor's than AMD's Dresden process, while AMD's process has more aggressive interconnects.
> Intel is now at 2GHz with more conservative techniques, by
> reasonable standards Intel holds the performance as well
> as marketing highground while AMD is gasping for breath at
> 1.4GHz with extreme techniques. Pipeline length aside, the
> answer seems pretty obvious.
Your comment here is exactly about what we were complaining earlier. You just completely throw aside the core issue of why Intel and AMD have their respective operating frequencies. First of all, you imply that Intel is comfortably at 2.00GHz while AMD is "gasping for breath" at 1.40GHz. In truth, it is likely from currently known volumes that AMD is binning much better at 1.40GHz than Intel is at 2.00GHz. I will grant you that from a marketing standpoint, AMD is in trouble. But that'd be even the case if AMD had a 3.00GHz processor right now. AMD has some sort of marketing curse that says "we must make ourselves look bad at every opportunity!"
Anyway, I would like you to really back up this whole "extreme techniques" allegation. Continuing to use it is like hearing the AMD fans here constantly bringing up the P4 throttling issue, except that your suggestion has even less evidence supporting it! :O
> Desperate times call for desperate measures and Intel
> isn't the one facing desperate times.
Everybody in the sector is facing desperate times. AMD is obviously posting a huge loss this quarter (I might even go as far to say that Paul might have been conservative in his guesstimate). But [imho] they would likely have posted breakeven or profit if the semi market were not in its current state. And Intel is obviously not doing nearly as well as it was a year ago, though they are profiting. We're just not in a good time. :/ |
