The Arm C1-Ultra scheduling model has been merged into the LLVM/Clang development tree ahead of the LLVM 23 stable release. Derived from the Neoverse V3 scheduling model and tuned using Arm's software optimization guide, this addition enables optimal code generation for the flagship Armv9.3-A C1-Ultra CPU core. The C1-Ultra promises +25% single-thread performance over the Cortex-X925 and +15% average speed-up in real-world applications.
73 Impressions