power consumption in IO

[eexuke]

Dear all,
Although the power consumption is application-specific,can you give me a rough knowledge about the proportion of core power to IO power consumption in a general microprocessor and a general video decoding processor?

Many thanks!!

 

[lijian_18]

Power consumption in IO can be calculated in the same way as in core.

P = P_dyn + P_lek ;

Dynamic power is related to Pad capacitance and signal switching rate.

Leakage is process relevant.

Hope it helps.

 

[SkyHigh]

eexuke,

can you give me a rough knowledge about the proportion of core power to IO power consumption in a general microprocessor and a general video decoding processor?

I/O power consumption in CMOS microprocessor and CMOS chips = dynamic power consumption + short-circuit or direct path power consumption + static leakage power consumption.

Dynamic power dissipation is due to output load capacitance (interconnect parasitic capacitance + gate capacitance of MOS transistors driven at the receiver end + parasitic overlap capacitance of MOS transistors at the receiver end).

Direct path power consumption is due to short-circuit current from Vdd to Gnd when both PMOS and NMOS are simultaneously turned on for a short time instance when switching logic.

Static power consumption is due to leakage current flowing through the reversed-biased diode junctions of both PMOS and NMOS transistors from Vdd to Gnd and from load to Gnd via NMOS transistor.

 

[eexuke]

Dear all,
Thanks for your answer.However,I am eager to the statistical results that general the proportion of power consumed in core vs. power consumed in IO for applications such as microprocessor or video processor.

 

[hung_wai_ming@hotmail.com]

For a typical 8-bit RISC, typically 0.5ma or 1ma per MHz is easily seen in 0.5um or 0.35um process. If you are talking about IO, u need to know how many IO and freq you are switching. Typically IO consume at least 24mA for high drive and 12mA for low drive and depends on ur output loading, you will know

It's meaningless u just ask the proportion without knowing the core freq, IO freq, output loading, number of IO switching as u know, microprocessor is instruction based and so the power consumption depends on your instructions or programs written. If u want to calculate things without output to IO, you will have 90% of power consumed in core and if you have divider, your IO can be switching at a much lower frequency then core, so IO power is much lower. If u are just toggling ports with simple instruction and toggle all ports (say 50 ports) and each port consume 24mA, u will know the answer IO is much higher

 

[eeeraghu]

In Video processor blocks and other DSP's there r more arithmatic calculations and if we r using 2's complement to implement these calculation there is a power reduction.