4 - Bufferizing Block-Scaled Types: The 1:N Split Aug 31, 2026 3 - Lowering MX Block-Matmul to linalg.generic with Block-Scale Affine Maps Aug 26, 2026 2 - Canonicalization and the E8M0 Power-of-Two Trick Aug 17, 2026 1 - A Dialect Is a Set of Co-Designed Decisions: Designing for MX Quantization in MLIR Aug 12, 2026 Compiler Strategies for Cache- and Memory-Friendly Design Across CPUs and GPUs Oct 18, 2025 Lowering as Graph-of-Graphs: A Structural View of MLIR and LLVM Pipelines Oct 14, 2025 Dynamic, Hybrid, and Static Graphs in Deep Learning Compilers Oct 8, 2025 Speculative Duplication and Cost Modeling in MLIR Oct 6, 2025 From Use–Def to Preimage: Fusing Structured Ops in MLIR Oct 1, 2025