Tag: Profiling
8 posts
-
C++ Performance Optimization: Measure First, Then Copies, Allocations, Cache and Compiler Flags
A measured approach to C++ performance: profile first, remove hidden copies, reserve allocations, fix memory access order, and use -O2/-O3, -march and LTO without the common traps.
-
Profiling a C++ Game Server: perf, gprof, Valgrind, VTune, Tracy and Flame Graphs
When your multithreaded C++ game server burns 60% CPU and you cannot find the bottleneck: perf, gprof, Valgrind, VTune, Tracy, flame graphs, cache misses.
-
Finding Memory Leaks: Heap Snapshots in Node.js, tracemalloc in Python, Heap Dumps in Java
Tell a leak from high memory use, then find it with heap snapshots and Clinic.js in Node.js, tracemalloc in Python, and heap dumps and jstat on the JVM.
-
Node.js Performance: Clustering, Caching, and Profiling
Node.js performance in practice: measure before tuning, cluster across cores, cache without stampedes, fix N+1 queries, keep the event loop free, and profile with --cpu-prof and clinic.js.
-
Optimizing a Slow Python CSV Job Step by Step, with Measured Timings
A measured walkthrough of a slow Python CSV job: reading cProfile correctly, hoisting loop invariants, NumPy vectorization, and why Pool made it slower.
-
Profiling C++ Before Optimizing: perf, gprof, Flame Graphs and Finding the Real Bottleneck
Measure before you optimize: C++ profiling with Linux perf, gprof, flame graphs, std::chrono, and Valgrind.
-
Why Is My C++ Program Slow? Find Bottlenecks with Profiling
The algorithm is right but C++ code is still slow: find hotspots with perf, flame graphs and Visual Studio, then fix copies, allocations and cache misses.
-
C++ Performance Optimization Case Study
Speeding up a slow C++ REST endpoint step by step: a baseline harness, perf profiling, a hash index, fewer string copies, reserved buffers, and why parallel JSON is the step to doubt.