Abstract
If we look into the scaling law carefully, we find that three crises can be foreseen for realizing LSI's in the year 2010: power crisis, interconnection crisis, and complexity crisis. Lowering supply voltage (V/sub DD/) is very effective in reducing the power but the threshold voltage (V/sub TH/) should be reduced at the same time for high-speed operation. The low V/sub TH/ however, increases the leakage current. To overcome this situation, V/sub TH/ and V/sub DD/ control through the use of multiple V/sub TH/, variable V/sub TH/, multiple V/sub DD/ and variable V/sub DD/ are being pursued. At the system level, a system LSI approach is promising for realizing low power. The new trend is to exploit the cooperation of software and hardware. For sub 1-volt design, abnormal temperature dependence of drain current may be important. The interconnection will determine cost, delay, power, reliability and turn-around time of future LSI's. The RC delay problem can be solved through an LSI architecture realizing "the further, the less communication" with the help of local memories. The complexity issue can only be solved by the sharing and re-use of design data, and the so-called IP-based design will be preferable.

This publication has 0 references indexed in Scilit: