testing: Benchmark, impossibly low ns/op

[cblach]

Go version

go version go1.23.1 darwin/arm64

Output of go env in your module/workspace:

GO111MODULE=''
GOARCH='arm64'
GOBIN=''
GOCACHE='/Users/christian.blach/Library/Caches/go-build'
GOENV='/Users/christian.blach/Library/Application Support/go/env'
GOEXE=''
GOEXPERIMENT=''
GOFLAGS=''
GOHOSTARCH='arm64'
GOHOSTOS='darwin'
GOINSECURE=''
GOMODCACHE='/Users/christian.blach/go/pkg/mod'
GONOPROXY=''
GONOSUMDB=''
GOOS='darwin'
GOPATH='/Users/christian.blach/go'
GOPRIVATE=''
GOPROXY='https://proxy.golang.org,direct'
GOROOT='/opt/homebrew/Cellar/go/1.23.1/libexec'
GOSUMDB='sum.golang.org'
GOTMPDIR=''
GOTOOLCHAIN='local'
GOTOOLDIR='/opt/homebrew/Cellar/go/1.23.1/libexec/pkg/tool/darwin_arm64'
GOVCS=''
GOVERSION='go1.23.1'
GODEBUG=''
GOTELEMETRY='local'
GOTELEMETRYDIR='/Users/christian.blach/Library/Application Support/go/telemetry'
GCCGO='gccgo'
GOARM64='v8.0'
AR='ar'
CC='cc'
CXX='c++'
CGO_ENABLED='1'
GOMOD='/Users/christian.blach/repos/exp/benchmarkbug/go.mod'
GOWORK=''
CGO_CFLAGS='-O2 -g'
CGO_CPPFLAGS=''
CGO_CXXFLAGS='-O2 -g'
CGO_FFLAGS='-O2 -g'
CGO_LDFLAGS='-O2 -g'
PKG_CONFIG='pkg-config'
GOGCCFLAGS='-fPIC -arch arm64 -pthread -fno-caret-diagnostics -Qunused-arguments -fmessage-length=0 -ffile-prefix-map=/var/folders/h9/th1sbhm15pjfb3vx7h5kr46c0000gq/T/go-build1899405527=/tmp/go-build -gno-record-gcc-switches -fno-common'

What did you do?

I created a simple benchmark file and I get ns/op in the femtosecond range.

package main

import (
	"math/rand/v2"
	"testing"
)

func BenchmarkRandInt(b *testing.B) {
	rand.Int()
}

What did you see happen?

$ go test -bench='.' -count=1
goos: darwin
goarch: arm64
pkg: benchmarkbug
cpu: Apple M3 Pro
BenchmarkRandInt-11 1000000000 0.0000003 ns/op
PASS
ok benchmarkbug 0.210s

What did you expect to see?

I expected not to see operations completed in sub femtosecond time.

Possible cause: It seems that go test believes that it ran 1000000000 tests while it only ran one and then probably divided the run time with 1000000000.

[gabyhelp]

Related Issues and Documentation

• testing: benchmark iteration reports incorrectly #41637 (closed)
• testing: benchmark performance misaligned for fast tests #30928
• Running testing.Benchmark produces wrong number of iterations #40012 (closed)
• testing: benchmark cleanup causes inconsistent and slower benchmark time #38314 (closed)
• go test is very slow on mac m1 #43370 (closed)
• testing: inconsistent benchmark measurements when interrupts timer #41641 (closed)
• Benchmark results vastly different when Tests are skipped with -run=XXX #42696 (closed)
• runtime: huge performance drops in a timer application #35395 (closed)
• go test -bench #24948 (closed)
• Go bench test never ends #46493 (closed)

_{(Emoji vote if this was helpful or unhelpful; more detailed feedback welcome in this discussion.)}

[cblach]

My error: I saw the issue and realized my mistake

#40012