IntroPower0 - Power Consumption Introduction Today more and...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
Power Consumption Introduction Today more and more embedded applications Targeted towards Small hand held or other types of portable devices Common thread through all such applications Need to have long battery life Translates to low power consumption Power consumption can be attacked in several ways Certainly hardware solution Low power devices Turning portions of system off ACPI – Advanced Configuration and Power Interface Surprisingly have software contribution as well Let’s look at each and begin with a view into the software Software There are a number of places that we can attack From software point of view Initial places to look The algorithms that we use Location of code Memory accesses can have significant impact on power Using software to control subsystems As we have been stating To analyze then control particular aspect of performance Must be able to measure that aspect Both before and after modification Measuring Power Consumption For the moment Will assume goal is to reduce power consumed by processor To such an end Measuring power consumption is two step process 1. Identify the portion of code to be analyzed Typically this will be a loop Doesn’t need to be - 1 of 9 -
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Measure the current consumed by the processor While the code is being exercised 2. Modify the loop such that the code comprising the loop is disabled Ensure that the compiler hasn’t optimized loop out Measure the current consumed by the processor Once we have identified power consumed Next step is to try to reduce if appropriate Studies have identified several software factors That contribute to processor power consumption Among the contributors we find The kind of instruction The collection or sequence of instructions executed The locations of the instructions and their operands Memory system and transfers in and out Have been shown to be most expensive operation Performed by processor Here memory is referring to main memory not cache This is the DRAM in our system Using simple addition operation as reference we find Operation Relative Power Consumption 16 Bit Add 1 16 Bit Multiply 3.6 8x128x16 SRAM Read 4.4 8x128x16 SRAM Write 9 I/O Access 10 16 Bit DRAM Memory Transfer 33 Evident from table Using cache can have significant affect on power consumption Assumes a cache hit Cache miss requires main memory access SRAM generally consumes more power than DRAM On per cell basis Cache is generally SRAM Want to optimize size of cache\ - 2 of 9 -
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 9

IntroPower0 - Power Consumption Introduction Today more and...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online