errors.New vs fmt.Errorf ?

I have begun learning Go lang recently. I chose Go lang because of performance issues and Its syntax and abilities took my attention. So I read documents and watched videos on YT and Udemy, and I began coding my first project. However, I was disappointed after a while since its performance and ability of handling massive data was not near what I expected, and I started to search the reason of the lack of performance.

Eventually, I found it. There was a lot of malformed data in my test case and my application must have handle them and return an error. The point that I struggled is exactly here because I used Errorf function of fmt package to return errors. Although this function provides us a lot of functionality such as formatting error message with parameters, it costs a lot. Lucky I do not need functionality fmt.Errorf serves because I return just same error string for those malformed data. After a short searching, I found errors package and read comparison blog posts between it and fmt.Errorf function. In the blog posts I read, errors.New is showed as obviously faster than fmt.Errorf, but I still wanted to see how fast it is and I wrote the code below. You can see the result at the end of the code.

errors.New ROCKS!!

Code:

package main

import (
    "errors"
    "fmt"
    "time"
)

func testErrors() error {
    return errors.New("test errors package")
}

func testFmt() error {
    return fmt.Errorf("test fmt package")
}

func main() {
    repeat := 100000000
    start := time.Now()
    for i := 0; i < repeat; i++ {
        testErrors()
    }
    duration := time.Since(start)
    fmt.Println("errors.New", duration)

    start = time.Now()
    for i := 0; i < repeat; i++ {
        testFmt()
    }
    duration = time.Since(start)
    fmt.Println("fmt.Errorf", duration)
}

Results:

errors.New 30.747008ms

fmt.Errorf 9.083900989s

As you can see error.News is almost 300 times faster than fmt.Errorf

CPU and RAM I tested at:

CPU: Intel(R) Core(TM) i7-7600U CPU @ 2.80GHz

RAM: 24GB (16+8) DDR4