
We present EXTREMIS, a compile-time pipeline that improves energy consumption of battery-less devices by ensuring that memory operations occur at the most efficient device frequency setting. Different memory operations incur different energy consumption depending on a device's current operating frequency. Volatile memory operations, for example, are generally most efficient at the highest frequency, whereas non-volatile memory operations may require wait cycles that make lower frequency setting more energy savvy. EXTREMIS reorders the instructions without violating data dependencies and inserts instructions to change the operating frequency depending on program flow and memory access patterns, reconciling their energy overhead with the gains they possibly enable. This is achieved by solving a series of optimization problems at compile-time. Our evaluation shows that, compared to a static frequency setting, EXTREMIS reduces a program's energy consumption by up to 11%, without incurring in any extra cost.
non-volatile memory, Datorsystem, Computer Systems, intermittent computing, Frequency scaling
non-volatile memory, Datorsystem, Computer Systems, intermittent computing, Frequency scaling
| selected citations These citations are derived from selected sources. This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | 0 | |
| popularity This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network. | Average | |
| influence This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | Average | |
| impulse This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network. | Average |
