Skip to content

Latest commit

 

History

History
36 lines (29 loc) · 1.05 KB

File metadata and controls

36 lines (29 loc) · 1.05 KB

Platform Specification

Target Machine (Linux)

Spec Value
Machine Lenovo LOQ 15
CPU AMD Ryzen 5 7235HS (4C/8T, Zen 3+, 3.2-4.2 GHz)
L2 Cache 2 MB
L3 Cache 8 MB
RAM 12 GB DDR5-4800 (1x SO-DIMM)
Storage 512 GB NVMe PCIe 4.0x4
GPU NVIDIA RTX 3050A 4GB (not used)
OS Ubuntu 22.04 LTS
Compiler g++ 11.x, -std=c++17 -O3 -march=native
Profiling perf stat + AMD uProf

Previous Machine (Mac — reference only)

Spec Value
CPU Apple M1 (4 Firestorm, 3.2 GHz)
L1D Cache 128 KB
L2 Cache 12 MB shared
RAM 8 GB LPDDR4X (68.25 GB/s)
OS macOS 26.2
Compiler Apple Clang 17, -std=c++17 -O3
Profiling xctrace CPU Counters

Key Architecture Differences

The SPPS cache-miss advantage is algorithmic (sequential access pattern), not architecture-specific. Expected to replicate on AMD Zen 3+.

AMD Zen 3+ has different L2/L3 cache sizes and prefetcher heuristics than Apple M1, which may affect absolute timings but not relative speedups.