Floats

Go has two floating-point types: float32 and float64. They differ in precision and memory usage. Floating-point literals default to float64 when using :=.

Declaring Floats

var f32 float32 = 3.14159265358979
var f64 float64 = 3.14159265358979

fmt.Println("float32:", f32) // 3.1415927
fmt.Println("float64:", f64) // 3.14159265358979

Output:

float32: 3.1415927  | type: float32
float64: 3.14159265358979 | type: float64

Both variables are assigned the same literal, but float32 rounds it — the precision loss is visible in the output.

Default Type with :=

pi := 3.14159265358979323846
// pi is float64 — floating-point literals always infer float64

Precision

Type	Precision	Typical use
float32	~6–7 significant decimal digits	Graphics, game engines, memory-constrained systems
float64	~15–17 significant decimal digits	General use — the default choice

float64 is preferred in most Go code. Use float32 only when you need to reduce memory usage or are working with an API that requires it.

Type Bounds

import "math"

fmt.Println("Max float32:", math.MaxFloat32)               // 3.4028234663852886e+38
fmt.Println("Max float64:", math.MaxFloat64)               // 1.7976931348623157e+308
fmt.Println("Smallest nonzero float32:", math.SmallestNonzeroFloat32) // 1.401298464324817e-45
fmt.Println("Smallest nonzero float64:", math.SmallestNonzeroFloat64) // 5e-324

Key Takeaways

• Default type: floating-point literals infer float64 with :=
• float64 preferred: higher precision and the standard choice for most Go code
• Precision loss is real: assigning the same literal to float32 and float64 produces visibly different values
• Use math constants: math.MaxFloat32, math.MaxFloat64, etc. give exact bounds

Related Topics

• Integers — whole-number types
• Type Conversions — converting between float types and to/from integers
• Go Types — other built-in types