This repo contains the LLVM transforms developed for Gem Forge framework, including the tracing instrumentation, the trace processing and stream specialization. It is used in these work:
Z. Wang and T. Nowatzki
"Stream-based Memory Access Specialization for General Purpose Processors,"
2019 ACM/IEEE 46th Annual International Symposium on Computer Architecture (ISCA)
Z. Wang, J. Weng, J. Lowe-Power, J. Gaur and T. Nowatzki
"Stream Floating: Enabling Proactive and Decentralized Cache Optimizations,"
2021 IEEE 27th International Symposium on High-Performance Computer Architecture (HPCA)
To use this, checkout the top repo gem-forge-framework.