Intel Sapphire Rapids versus AMD Zen 4 – Daniel Lemire’s weblog

Intel has launch a brand new technology of server processors (Sapphire Rapids) whereas the most recent AMD know-how (Zen 4) is now broadly out there. There are intensive comparisons out there. Of explicit curiosity is the open benchmark outcomes which assess varied facets of processor speeds, together with JSON parsing performance. In these benchmarks, AMD methods seem to dominate.
I made a decision to run my very own benchmarks utilizing JSON parsing as a reference and generally out there Amazon large nodes. For these assessments, I take advantage of Amazon Linux 2023 with GCC 11. I take advantage of two situations that value about 5 {dollars} per hour. Amazon prices me about the identical quantity for each the AMD and Intel methods.
The AMD occasion is of kind c7a.24xlarge with an AMD EPYC 9R14 processor (Zen 4 microarchitecture). The Intel occasion is of kind c7i.metal-24xl with an Intel XeonPlatinum 8488C (Sapphire Rapids microarchitecture). I take advantage of methods with a number of cores however my benchmark is totally single threaded. I may have optimized both system by going with methods which have fewer cores working hotter. In my case, each processors run in follow at a comparable frequency, with a slight benefit for AMD (3.5 GHz vs 3.4 GHz).
The gist of the result’s that neither system dominates the opposite one. In some benchmarks, Intel wins, in others AMD wins. It is rather intently matched.
Intel outcomes:
simdjson On-Demand | simdjson DOM | yyjson | rapidjson | nlohmann/json | Enhance JSON | |
json2msgpack | 3.68 GB/s | 2.67 GB/s | 1.72 GB/s | 0.71 GB/s | 0.03 GB/s | 0.42 GB/s |
partial_tweets | 6.83 GB/s | 4.77 GB/s | 2.41 GB/s | 0.77 GB/s | 0.13 GB/s | 0.50 GB/s |
distinct_user_id | 6.99 GB/s | 4.90 GB/s | 2.52 GB/s | 0.67 GB/s | 0.14 GB/s | 0.49 GB/s |
kostya | 2.92 GB/s | 2.03 GB/s | 0.83 GB/s | 0.80 GB/s | 0.12 GB/s | 0.47 GB/s |
AMD outcomes:
simdjson On-Demand | simdjson DOM | yyjson | rapidjson | nlohmann/json | Enhance JSON | |
json2msgpack | 3.09 GB/s | 2.45 GB/s | 1.93 GB/s | 0.68 GB/s | 0.03 GB/s | 0.38 GB/s |
partial_tweets | 6.84 GB/s | 4.22 GB/s | 2.64 GB/s | 0.77 GB/s | 0.12 GB/s | 0.46 GB/s |
distinct_user_id | 6.94 GB/s | 4.26 GB/s | 2.58 GB/s | 0.77 GB/s | 0.13 GB/s | 0.47 GB/s |
kostya | 4.03 GB/s | 2.71 GB/s | 1.00 GB/s | 0.78 GB/s | 0.12 GB/s | 0.52 GB/s |
You may reproduce my outcomes by grabbing simdjson and working bench_ondemand.
I don’t fake that this single knowledge level is ample to make buying selections or to evaluate the Intel and AMD know-how. Take it as a knowledge level.
Additional studying. On-demand JSON: A better way to parse documents?, Software program: Apply and Expertise (to look)
Printed by
