DAE4HLS: Exposing Memory-Level Parallelism for High-Level Synthesis using Explicit Decoupling
For HLS users, this work enables acceleration of applications with complex memory access patterns that previously could not benefit from HLS, reducing design effort.
DAE4HLS introduces a decoupled access-execute paradigm for high-level synthesis that explicitly separates memory requests and responses, unlocking memory-level parallelism for complex access patterns on large datasets. Applied to AMD Vitis HLS, it achieves a total speedup of 10-79×.
High-level synthesis (HLS) performs well for simple memory access patterns, such as for sequential accesses that can be turned into bursts, or for memory accesses into small datasets that can be stored in scratchpads. This limits HLS to accelerating only the low-hanging fruit, where memory-level parallelism is either trivially abundant, due to simple access patterns, or latency is low, due to the small dataset. Applications with more complex access patterns on large datasets would also benefit from acceleration, and would especially benefit from the reduction in design and verification effort that HLS promises. In this paper, we present DAE4HLS, a decoupled access-execute (DAE) paradigm for HLS. We propose a new programming model for explicitly decoupling requests and responses, which unlocks memory-level parallelism that otherwise cannot be automatically provided by a compiler. We apply the DAE4HLS paradigm to the commercial AMD Vitis HLS toolchain and show that the existing AXI stream and AXI burst interfaces can be repurposed for explicit decoupling. We further apply the paradigm to a dynamic-HLS framework, which is better suited for handling irregular workloads as compared to statically scheduled HLS. We show that support for explicit decoupling improves the performance and achieves a total speedup of 10-79$\times$.