storage vs. processing
Ups ] [ Post
Followup ] [ CPU-Central Message Board ] [ Home ]
Subject: storage vs. processing
Date: 11/3/2001 10:25:56 PM (GMT-7)
IP Address: 126.96.36.199
In Reply to: Nice link posted by
CPU -- In a system processor, you have logic based on transistors and each one dissipates a certain amount of current, in addition to faint amounts of loss that occur in the traces between those transistors.
Smaller die and improvements in design process lead to lower power draw per transistor, but the quantity of those has been increasing dramatically to the point where we now have several million or more on each die. Also current draw will depend on how often you switch the transistor, switch it more often and current draw goes up.
Demands for speed lead to designs that push the limits of currently available technology, hence the majority of modern processors run very hot.
DRAM -- Although they've gotten much faster, a DRAM cell is still a DRAM cell -- each bit stored requires two parts, a controlling transistor and a capacitor to store the charge, IIRC.
Faster operation means more frequent access and more frequent refreshes (the charge will bleed off that capacitor, so you have to re-initialize it periodically), but even so memory speeds have not increased at the same rate that CPU speeds have risen. Furthermore, although memory chips have gotten denser over time, they are nowhere close to the kind of situation presented in a system processor.
Keep in mind that current memory is implemented on 0.18 and 0.15µ CMOS technology, if I am not mistaken the early Pentiums were something like 1.0 or 0.7µ and the overall process wasn't as refined as it is now. I cannot say for certain but I'll bet something like the P233MMX could be re-designed for, and implemented on, the newest 0.13µ copper technology and -- with no heatsink at all -- would run only warm to the touch.
FollowUps | Post
E-Mail Sender |
Maximum of 100 messages displayed.
Post a Followup
Ups ] [ Post
Followup ] [ CPU-Central Message Board ]