Diagnose and optimize Go CPU, memory, GC, and concurrency performance bottlenecks.
npx skills add https://github.com/smallnest/chao-go-perf --skill chao-go-perf-smallnest
Please help me install this Agent Skill. Skill: chao-go-perf Source: https://github.com/smallnest/chao-go-perf Command: npx skills add https://github.com/smallnest/chao-go-perf --skill chao-go-perf-smallnest
go test -bench=. -cpuprofile=cpu.out, identifying the hottest function in a flame graph, checking escape analysis with -gcflags="-m", applying a fix, and validating the improvement statistically with benchstat. ## Quick Start Ask the assistant to analyze your Go benchmark or pprof profile and suggest performance optimizations for the hottest code paths.High-intent search queries and answers about installing and using this skill.
Generate a CPU profile with go test -bench=. -cpuprofile=cpu.out, then open it with go tool pprof -http=:8080 cpu.out. Use the flame graph view to find the widest functions, which consume the most CPU time.
Run go build -gcflags="-m" to inspect escape analysis, preallocate slices and maps with known capacity, use strings.Builder instead of + concatenation, and reuse temporary objects with sync.Pool. Verify improvements with -benchmem and benchstat.
Use sync.Pool for high-frequency temporary objects like bytes.Buffer that are cheap to reuse. Always call Reset before putting objects back, and avoid it for stateful or long-lived objects since GC may clear pooled items.
Use atomic operations for simple counters and flags, sync.Mutex for balanced read/write workloads, and sync.RWMutex when reads exceed roughly 90 percent. For high write contention, consider sharded locks to spread contention.
False sharing occurs when goroutines on different cores write variables sharing one 64-byte cache line, causing severe slowdowns. Fix it by padding structs with 56 bytes between hot fields so each lands on its own cache line.
PGO typically improves CPU-bound programs by 2-7 percent. Collect a representative profile, save it as default.pgo in the main package directory, and run go build; the compiler automatically applies profile-guided inlining and layout decisions.
Harmeet10000Measure Go performance with benchmarks, pprof profiles, and benchstat comparisons.
Bebang-Enterprise-IncDeploy production branch changes to AWS EC2 running Frappe Docker.
monforjeAnalyze Go code for concurrency issues like goroutine leaks and race conditions.
belokonmProfile CPU hot paths with eBPF, perf, and Flamegraphs.