0 views collected around this technical thread.
The article explains how performance can be dramatically improved by keeping frequently used data in CPU registers instead of memory, understanding basic assembly syntax and instruction types, using branch‑prediction hints, and exploiting the CPU pipeline to reduce stalls and wasted cycles.
Profile‑Guided Optimization (PGO) collects runtime profiling data to recompile programs for higher performance, reducing branch mispredictions and improving code layout; Go gained built‑in PGO in 1.21 with typical 5 % gains, while C++ sees 15‑18 % QPS improvements and devirtualization benefits, and future work aims at deeper block ordering and register allocation.
Modern mobile CPUs are superscalar, using deep pipelining, branch prediction, register renaming, out‑of‑order issue, execution, write‑back, and commit stages to boost instruction‑level parallelism, while performance modeling via CPI and hardware counters helps engineers overcome power, memory, and compiler limitations for efficient code.
This article explains how the Linux kernel’s likely and unlikely macros, which wrap GCC’s __builtin_expect, guide branch prediction to improve cache utilization and pipeline efficiency, and demonstrates their impact with sample code and assembly analysis.
This article explains how the Linux kernel's likely and unlikely macros, which wrap __builtin_expect, guide the compiler's branch prediction to improve cache usage and pipeline efficiency, and demonstrates the effect with sample C code and assembly output.
The article examines a puzzling StackOverflow case where sorting a random array before summation yields a six‑fold speedup, explains the phenomenon through CPU five‑stage pipeline fundamentals, structural, data, and control hazards, and shows how branch prediction and operand forwarding mitigate the performance loss.
The article demonstrates how Top‑down Micro‑architecture Analysis Methodology (TMAM) can quickly pinpoint CPU bottlenecks—such as front‑end, back‑end, and bad speculation stalls—in a simple C++ accumulation loop, and shows that applying targeted compiler, alignment, and branch‑prediction optimizations reduces runtime by roughly 34 % while increasing retiring slots.