C38368 EDIT~ I just checked Intel's site, and confirmed my suspicion: there is no such thing as a 2.26GHz P4 with 256k L2 cache; all have 512k. There is one Prescott variant, but it also has 512k.
Hmm, well I guess I need to pay a little more attention to the Intel processors. Get myself a cheat sheet.
Pipelines... I dropped the ball again. :rolleyes:
Now to find out how many piplines that Celeron has.
What's the price differential between the 2.26 and the 2.4 for you?
About...umm...600Taka. $10.So I plan to get the 2.4A anyway.
Monster_userHmm, well I guess I need to pay a little more attention to the Intel processors. Get myself a cheat sheet. Pipelines... I dropped the ball again. :rolleyes: Now to find out how many piplines that Celeron has.
One ;) AFAIK, all CPUs have one pipeline per core. It's the number of stages in that pipeline that affect performance at any given speed. Northwoods have 20 stages, Prescotts 31. IIRC, AXPs had 12 and A64s have 18. The relationship goes something as follows: each stage of the pipeline performs some amount of work on an instruction (the most common analogy is an assembly line). Shorter pipelines have to do more work to an instruction at each stage, making them "slower". Longer pipelines, conversely, do less work per stage, so instructions move "faster". However, in the event of a misprediction or other error, the pipeline must be flushed and refilled; this takes more time with a longer pipeline. The tradeoff, on the other hand, is that longer pipelines scale to higher effective core speeds much better than shorter ones. This is why all the fastest P4's are Prescotts, but why they're shunned by so many users at lower clock speeds. It's also why AMD's CPUs aren't nearly so fast (in terms of effective clock speed) as Intel's; they can't be as a result of the shorter pipeline (if AMD could produce a 3.6GHz CPU with an on-chip memory controller and nine executable units, they would, and Intel would be in huge trouble for the desktop market).