ReadMeFeed / Community knowledge
Real questions. Useful conversations. Find the people who know your stack.
Frameworks & web · #llvm
Stop letting the linker stall your CI. Learn how LLVM ThinLTO parallelizes the link step to reduce build times from 15 minutes to under 3 minutes for large C++ projects.
When LLVM’s LTO produces link failures, the culprit is often a mismatch in flags, libraries, or linker support. This guide walks through a concise cause‑effect table, ordered checks, and targeted fixes so you can restore a successful LTO build.
ThinLTO balances compile‑time speed and runtime performance by generating compact summaries during compilation and performing global optimizations in parallel, reducing link time by up to 70% while still delivering most of the benefits of full LTO.
A concise guide to creating a minimal LLVM New Pass Manager pass, covering requirements, the smallest workable design, trust boundaries, operational checks, failure modes, and verification steps.
A concise guide to building a minimal ORC JIT plugin system for servers, covering design, security, verification, and when to redesign.
PassManagerAdapter and Analysis Result Persistence The transition to the New Pass Manager (NPM) in LLVM introduces a functional approach to pass orchestration, utilizing a nested pipeline structure to optimize IR. To maintain compatibility during migration, the PassManagerAdapter allows legacy passes to execute within the NPM framework. A critical design con
Analysis Result Persistence in the New Pass Manager The LLVM New Pass Manager (NPM) utilizes a caching mechanism for analysis results to optimize the pipeline. To maintain correctness, passes must explicitly specify which analyses remain valid after an IR transformation by returning a PreservedAnalyses object. When a pass modifies the IR but fails to correct
LLVM’s LoopVectorize pass decides whether to vectorize a loop based on a cost model that evaluates trip count, memory stride, alignment, and target‑specific vector width. The model is consulted during the pass’s execution, which typically runs before profile‑guided optimization (PGO) marks functions as hot or cold. After PGO, the optimizer may reorder or re‑
LLVM IR Optimization Limits The interaction between aggressive loop unrolling and vectorization within complex loops presents an unresolved behavior. The optimization pipeline is designed for transformation, not imperative retry logic. Verification Context Considering LLVM's established IR data-flow analysis capabilities, what is the documented boundary for