Profile Guided Optimization without Profiles: A Machine Learning Approach.

Warrior1: A Performance Sanitizer for C++.

Deep Learning Inference in Facebook Data Centers: Characterization, Performance Optimizations and Hardware Implications.

Glow: Graph Lowering Compiler Techniques for Neural Networks.

Block Unification IF-conversion for High Performance Architectures.

Optimizing Wait States in the Synthesis of Memory References with Unpredictable Latencies.

The benefits of using variable-length pipelined operations in high-level synthesis.

Using memory profile analysis for automatic synthesis of pointers code.

Hybrid type legalization for a sparse SIMD instruction set.

Combining static and dynamic array detection for binary synthesis with multiple memory ports.

Reducing Memory Constraints in Modulo Scheduling Synthesis for FPGAs.

Finding the best compromise in compiling compound loops to Verilog.