Recursive function data allocation to scratch-pad memory

Abstract

This paper presents the first automatic scheme to allocate local (stack) data in recursive functions to scratch-pad memory (SPM) in embedded systems. A scratch-pad is a fast directly addressed compiler-managed SRAM memory that replaces the hardware-managed cache. It is motivated by its significantly lower access time, energy consumption, real-time bounds, area and overall runtime. Existing compiler methods for allocating data to scratch-pad are able to place only code, global, heap and non-recursive stack data in scratch-pad memory; stack data for recursive functions is allocated entirely in DRAM, resulting in poor performance.In this paper we present a dynamic yet compiler-directed allocation method for recursive function stack data that for the first time, is able to place a portion of recursive stack data in scratch-pad. It has almost no software-caching overhead, and is able to move recursive function data back and forth between scratchpad and DRAM to better track the program's locality characteristics. With our method, all code, global, stack and heap variables can share the same scratch-pad. When compared to placing all recursive function data in DRAM and all other variables in scratch-pad, our results show that our method reduces the average runtime of our benchmarks by 29.3%, and the average power consumption by 31.1%, for the same size of scratch-pad fixed at 5% of total data size. Furthermore, significant savings were observed when comparing our method against cache-based alternatives for SPM allocation. Finally, we show results that analyze the effects of profile variation on our allocation approach and present a modified version of our method which minimizes variation for profile-based allocations.

Stack Caching Using Split Data Caches

“…SSDM showed speedups between 15% and 1% compared to circular stack management. A similar approach optimized for recursive functions has also been explored [4].…”

Optimizing Local Memory Allocation and Assignment through a Decoupled Approach

“…We are interested in data management [18,22,23], and target dynamic methods which are superior to static ones except when code size is extremely constrained [16]. We elaborate on two recent series of results targeting stack and global array management in LMs, embracing the analogies with register allocation.…”

Optimal stack frame placement and transfer for energy reduction targeting embedded processors with scratch-pad memories

“…Approaches supporting recursive functions include [19] and [20]. [19] proposes to use a circular buffer for managing frames within the SPM.…”

“…This technique naturally supports recursive functions, is simple, but is not optimal as the oldest frame eviction is systematic when the SPM is full even if the cost is larger than the gain. The technique of [20] consists of looking for the depths of recursion where there are numerous accesses to the stack using profiling informations. Frames corresponding to such depths are stored into the SPM whereas the others are stored into the MM.…”

“…For the approach where frames corresponding to a given depth of recursion are put into the SPM [20], it is enough to add a grow and a shrink session for the corresponding recursive function per depth of interest to be treated like a different non-recursive function.…”