LLVM Clang 16 vs. GCC 13 Compiler Performance On AMD 4th Gen EPYC "Genoa"

Written by Michael Larabel in Software on 30 May 2023 at 11:42 AM EDT. Page 5 of 6. 12 Comments.
oneDNN benchmark with settings of Harness: Recurrent Neural Network Training, Data Type: bf16bf16bf16, Engine: CPU. Clang 16.0.3 was the fastest.
oneDNN benchmark with settings of Harness: Recurrent Neural Network Inference, Data Type: bf16bf16bf16, Engine: CPU. Clang 16.0.3 was the fastest.

The most significant wins for the Clang compiler were when building Intel's oneDNN neural network library.

AOBench benchmark with settings of Size: 2048 x 2048, Total Time. GCC 13.1.1 was the fastest.
SecureMark benchmark with settings of Benchmark: SecureMark-TLS. Clang 16.0.3 was the fastest.
OpenJPEG benchmark with settings of Encode: NASA Curiosity Panorama M34. Clang 16.0.3 was the fastest.
OpenSSL benchmark with settings of Algorithm: SHA512. Clang 16.0.3 was the fastest.
OpenSSL benchmark with settings of Algorithm: RSA4096. GCC 13.1.1 was the fastest.
OpenSSL benchmark with settings of Algorithm: ChaCha20. GCC 13.1.1 was the fastest.
OpenSSL benchmark with settings of Algorithm: AES-128-GCM. GCC 13.1.1 was the fastest.
OpenSSL benchmark with settings of Algorithm: AES-256-GCM. GCC 13.1.1 was the fastest.
OpenSSL benchmark with settings of Algorithm: ChaCha20-Poly1305. GCC 13.1.1 was the fastest.
Liquid-DSP benchmark with settings of Threads: 256, Buffer Length: 256, Filter Length: 57. Clang 16.0.3 was the fastest.
Liquid-DSP benchmark with settings of Threads: 384, Buffer Length: 256, Filter Length: 57. Clang 16.0.3 was the fastest.
ASTC Encoder benchmark with settings of Preset: Fast. Clang 16.0.3 was the fastest.
ASTC Encoder benchmark with settings of Preset: Medium. Clang 16.0.3 was the fastest.

Some of these workloads are more relevant on servers than others, the intent was on simply loading up the AMD EPYC 9654 2P server with a wide range of binaries built under both of these leading open-source compilers.

MariaDB benchmark with settings of Clients: 2048. GCC 13.1.1 was the fastest.
MariaDB benchmark with settings of Clients: 4096. GCC 13.1.1 was the fastest.

The GCC-built MariaDB MySQL server was slightly faster than when built under Clang 16.

PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 800, Mode: Read Only. Clang 16.0.3 was the fastest.
PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 800, Mode: Read Only, Average Latency. Clang 16.0.3 was the fastest.

It was a competitive race with the PostgreSQl server.


Related Articles
Intel Core Ultra "Meteor Lake" Performance Improves With Linux 6.9
GCC 14 vs. LLVM Clang 18 Compiler Performance On Fedora 40
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