Researchers including Yan Solihin from the North Carolina State University in the US developed a technique called Dense Footprint Cache, in which the cache learns over time which data the processor needs from each macroblock.
Computer processors have to retrieve data from memory to perform operations. All data is stored in off-chip 'main' memory.
However, data that is used a lot is temporarily stored in a die-stacked dynamic random access memory (DRAM) cache located closer to the processor, where it can be retrieved more quickly.
However, for any given operation, the processor does not need all of the data in a macroblock - and retrieving the unnecessary data takes time and energy.
More From This Section
To make the process more efficient, researchers have developed a technique in which the cache learns over time which data the processor needs from each macroblock.
This allows the cache to compress the macroblock, retrieving only the relevant data. It also enables the cache to send data to the processor more efficiently.
The researchers tested this approach, called Dense Footprint Cache, in a processor and memory simulator.
After running three billion instructions for each application tested through the simulator, researchers found that the Dense Footprint Cache sped up applications by 9.5 per cent compared to state-of-the-art competing methods for managing die-stacked DRAM.
Dense Footprint Cache also used 4.3 per cent less energy. The researchers also found that Dense Footprint Cache led to a significant improvement in 'last-level cache miss ratios'.
These cache misses make operations much less efficient - and Dense Footprint Cache reduced last-level cache miss ratios by 43 per cent.