Scalar Replacement in the Presence of Multiple Write Accesses with Non-constant Reuse Distances
High-level synthesis (HLS) reduces design time of domain-specific accelerators from loop nests. Usually, naive usage of HLS leads to accelerators with insufficient performance, so very time-consuming manual optimizations of input programs are necessary in such cases. Scalar replacement is a promisin...
Saved in:
| Published in | IPSJ Transactions on System and LSI Design Methodology Vol. 18; pp. 10 - 18 |
|---|---|
| Main Author | |
| Format | Journal Article |
| Language | English |
| Published |
Tokyo
Information Processing Society of Japan
01.01.2025
Japan Science and Technology Agency |
| Subjects | |
| Online Access | Get full text |
| ISSN | 1882-6687 1882-6687 |
| DOI | 10.2197/ipsjtsldm.18.10 |
Cover
| Summary: | High-level synthesis (HLS) reduces design time of domain-specific accelerators from loop nests. Usually, naive usage of HLS leads to accelerators with insufficient performance, so very time-consuming manual optimizations of input programs are necessary in such cases. Scalar replacement is a promising automatic memory access optimization that removes redundant memory accesses. However, it cannot handle loops with multiple write accesses to the same array, which poses a severe limitation of its applicability. In addition, it is difficult to automatically apply scalar replacement to memory accesses with non-constant reuse distances. In this paper, we propose a novel memory access optimization technique that overcomes these existing limitations. Experimental results show that the proposed method achieves 2.14x performance gain on average with decreased total gate count of 5% for the benchmark programs which the state-of-the-art memory optimization techniques cannot optimize. |
|---|---|
| Bibliography: | SourceType-Scholarly Journals-1 ObjectType-General Information-1 content type line 14 |
| ISSN: | 1882-6687 1882-6687 |
| DOI: | 10.2197/ipsjtsldm.18.10 |