dwchang wrote: Uhm...no...model numbers were adopted to counter this and it has only mildly suceeded. The conumer is programed that GHZ = Performance...that's how it is...putting instructions into the equation is (I know how insulting this sounds)...too much for the consumer.
But the consumer doesn't automatically know that gigacraps would have anything to do with instructions. All they'd know is that your numbers are bigger than Intel's, and both start with "giga". So they can either choose the bigger number, or they can decide to educate themselves. Either way, you win.
dwchang wrote:Up until the P4, the trend WAS that speed = performance because the architectures were efficient enough..it's not the case anymore. However, since consumers have been used to this...they still believe it.
Thank you for reiterating a history lesson that we've gone over a few dozen times already.
the Black Monarch wrote:And on top of that, I'm pretty sure I saw some benchmark tests wherein a 1.4 gig PIII (the fastest PIII ever) marginally outperformed the 1.4 gig P4.
I'm sorry, that was supposed to be "was marginally outperformed by"
dwchang wrote:Well again...it's not an inherent flaw and you haven't disproved anything. Centrino Architectures ~ PIII. This is a well-known fact in the architecture field and if you choose to ignore it fine...I don't really care. As for benchmarks, I disagree. I am an architect and KNOW (because I have seen the designs of these architectures)...speed for speed, the PIII is A LOT better and more elegant. Think what you want though. Again you haven't really rebutted the main point and well...not that anyone could since it's been well-accepted...this is a pretty pointless argument....i don't know why I replied.
I don't either, since I was never disagreeing with you in the first place. I'm just passing along to you what I've seen and heard and waiting for your explanations.
dwchang wrote:Actually, I already knew that...the K5 and all that.
You said you didn't...
dwchang wrote:Well considering I'm not on the Hammer team or even if I was..I didn't design it...I wouldn't know. We work on a "need-to-know" basis...I don't need to know it...so I don't. I'll probably find out the details when you guys do...like when I read a site

. The most private thing I have is the Hammer debug stuff and Hammer Instruction set.
Oh. Damn.
dwchang wrote:And who said the clock speeds were lower than the Centrino (1.6 Ghz)? You must've read something odd.
Opterons go as low as 1.4 gig; Centrinos go as high as 1.8 gig with 2.0 in the very near future.
dwchang wrote:Uhm...I mentioned it quite a few times...the whole performance = speed * instructon/clock and thus..if they get a REALLY fast speed (like 3 Ghz), they can be comprable or beat a slower frequency with more efficient core....it's just simple mathematics. I was only getting at (the ENTIRE time) that it's an inefficient design and the frequency is not the only indicator...as demonstrated by 2.25 Ghz parts being comprable....
Really? What a shock! I must have missed that during my web research and the first fifty dozen times you explained it!
Seriously, though, I KNOW you "mentioned it quite a few times." I'm not disagreeing with it. I'm just wondering why you like to praise efficiency and criticise inefficiency so much more than clock speed limits when you openly admit that both are valid ways of achieving performance.
dwchang wrote:As for this discussion...as I have stated already...we've gotten QUITE off-topic and I know this'll sound mean, but I don't really see it going anywhere...it's been quite circular...if you reallyi wanna discuss it just create a new thread or something...I am kind of getting sick of answering the same questions...take that as you may (like if you're gonna say "ooh I win" or something)....
Yes, it is becoming vey circular indeed. A brief summary of the loop:
you: performance = clock speed * instructions per clock cycle
me: Yes, you said that already. So you openly admit that clock speed is just as important as efficiency. So if both are equally important, and the Athlon XP performs about as well as the Pentium 4, why do you like the slower yet more efficent core over the faster but less efficient core?
you: I don't. Like I said before, performance = clock speed * instructions per clock cycle. Consumers equate Ghz with performance, so they stupidly think Intel's chips are better. But the Pentium 4 is so inefficient, I hate it and it sucks!
me: Even though its higher clock speeds allow it to perform on par with the Athlon XP?
you: Yeah, like I said before, performance = clock speed * instructions per clock cycle. Why are you disagreeing with me?
me: I'm not! I just want to know why you like the slower yet more efficent core over the faster but less efficient core!
you: I don't. Like I said before, performance = clock speed * instructions per clock cycle. Consumers equate Ghz with performance, so they stupidly think Intel's chips are better. But the Pentium 4 is so inefficient, I hate it and it sucks!
(repeat)
you: this argument isn't going anywhere. I'm leaving. you win.
me: I do? But I wasn't arguing!
Consider this my revenge for your impersonation of my attitude toward the RCB. Not meant to be a personal attack, just expressing how I see our conversation going.
Ask me about my secret stash of videos that can't be found anywhere anymore.