Google Cloud N4 Series Benchmarks: Google Axion vs. Intel Xeon vs. AMD EPYC Performance

Written by Michael Larabel in Cloud on 24 February 2026 at 10:00 AM EST. Page 6 of 7. 13 Comments.

The AMD EPYC VM did offer a slight performance advantage over the Google Axion ARM64 processor for PostgreSQL database performance. The Intel Xeon N4 VM did pick up a rare win when it came to hitting 1000 clients against the PostgreSQL server in read-only mode but trailed the ARM64 and EPYC instance types in value.

👁 nginx benchmark with settings of Connections: 500. N4D EPYC Turin was the fastest.

👁 nginx benchmark with settings of Connections: 1000. N4D EPYC Turin was the fastest.

The AMD EPYC VM was at the forefront for the Nginx HTTPS web server performance in this N4-class 16 vCPU comparison.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: Qwen3-8B-Q8_0, Test: Text Generation 128. N4A Axion was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: Qwen3-8B-Q8_0, Test: Prompt Processing 2048. N4 Xeon EMR was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: gpt-oss-20b-Q8_0, Test: Text Generation 128. N4A Axion was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: gpt-oss-20b-Q8_0, Test: Prompt Processing 2048. N4D EPYC Turin was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Text Generation 128. N4A Axion was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: Llama-3.1-Tulu-3-8B-Q8_0, Test: Prompt Processing 2048. N4 Xeon EMR was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: Mistral-7B-Instruct-v0.3-Q8_0, Test: Text Generation 128. N4A Axion was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: Mistral-7B-Instruct-v0.3-Q8_0, Test: Prompt Processing 2048. N4 Xeon EMR was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: DeepSeek-R1-Distill-Llama-8B-Q8_0, Test: Text Generation 128. N4A Axion was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: granite-3.0-3b-a800m-instruct-Q8_0, Test: Text Generation 128. N4A Axion was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: DeepSeek-R1-Distill-Llama-8B-Q8_0, Test: Prompt Processing 2048. N4 Xeon EMR was the fastest.

👁 Llama.cpp benchmark with settings of Backend: CPU BLAS, Model: granite-3.0-3b-a800m-instruct-Q8_0, Test: Prompt Processing 2048. N4D EPYC Turin was the fastest.

The Llama.cpp performance for CPU-based AI inferencing was quite mixed depending upon the model and TG/PP focus for performance. The Google Axion performance did come out stronger in at least the text generation benchmarks than I was expecting. The Intel Xeon Emerald Rapids picked up some wins here thanks to Llama.cpp supporting Advanced Matrix Extensions (AMX).

13 Comments - Next Page

Page: 1 2 3 4 5 6 7 Next Page

URL: https://www.phoronix.com/review/google-cloud-n4-arm64-epyc-xeon/6

⇱ Google Cloud N4 Series Benchmarks: Google Axion vs. Intel Xeon vs. AMD EPYC Performance - Phoronix

Google Cloud N4 Series Benchmarks: Google Axion vs. Intel Xeon vs. AMD EPYC Performance