GCC 14 vs. LLVM Clang 18 Compiler Performance On Fedora 40

Written by Michael Larabel in Software on 24 April 2024 at 11:02 AM EDT. Page 2 of 5. 34 Comments.
QuantLib benchmark with settings of Configuration: Multi-Threaded. Clang 18.1.1 was the fastest.
QuantLib benchmark with settings of Configuration: Single-Threaded. Clang 18.1.1 was the fastest.

Right out of the gate we see the competitiveness of today's LLVM Clang compiler against GCC. Going back a decade or so GCC tended to consistently lead while these days is a very nice and competitive race on x86_64 (and AArch64).

Quicksilver benchmark with settings of Input: CTS2. GCC 14.0.1 20240411 was the fastest.
Quicksilver benchmark with settings of Input: CORAL2 P1. GCC 14.0.1 20240411 was the fastest.
Quicksilver benchmark with settings of Input: CORAL2 P2. Clang 18.1.1 was the fastest.

There still are some workloads where GCC was yielding faster binaries than LLVM Clang on this AMD Zen 4 workstation.

miniBUDE benchmark with settings of Implementation: OpenMP, Input Deck: BM1. GCC 14.0.1 20240411 was the fastest.
miniBUDE benchmark with settings of Implementation: OpenMP, Input Deck: BM1. GCC 14.0.1 20240411 was the fastest.
miniBUDE benchmark with settings of Implementation: OpenMP, Input Deck: BM2. Clang 18.1.1 was the fastest.
miniBUDE benchmark with settings of Implementation: OpenMP, Input Deck: BM2. Clang 18.1.1 was the fastest.

But most commonly it was a neck-and-neck race and in some cases Clang outperforming GCC.

SMHasher benchmark with settings of Hash: wyhash. Clang 18.1.1 was the fastest.
SMHasher benchmark with settings of Hash: SHA3-256. Clang 18.1.1 was the fastest.
SMHasher benchmark with settings of Hash: Spooky32. Clang 18.1.1 was the fastest.
SMHasher benchmark with settings of Hash: fasthash32. GCC 14.0.1 20240411 was the fastest.
SMHasher benchmark with settings of Hash: FarmHash128. GCC 14.0.1 20240411 was the fastest.
SMHasher benchmark with settings of Hash: FarmHash32 x86_64 AVX. GCC 14.0.1 20240411 was the fastest.
SMHasher benchmark with settings of Hash: t1ha0_aes_avx2 x86_64. Clang 18.1.1 was the fastest.

In some of the SMHasher benchmarks (note the compiler flag differences here are set by its build system), the Clang-built binaries had nice gains over GCC while for some crypto algorithms the GCC binaries were faster.


Related Articles
Intel Core Ultra "Meteor Lake" Performance Improves With Linux 6.9
Intel Xeon Max Sees Some Performance Gains For OpenVINO & ONNX With Linux 6.9
Linux 6.9 Drives AMD 4th Gen EPYC Performance Even Higher For Some Workloads
Intel Core Ultra "Meteor Lake" Yields Faster Performance With Linux 6.9
Google Cloud's C3D Instances Provide Strong Performance Value For PingCAP's TiDB
Linux 6.9 Features: DM VDO, AMD Preferred Core, Intel FRED & Larger Console Fonts