The min numeric instruction is used for getting the lower of two numbers.
(module
(import "console" "log" (func $log (param f32)))
(func $main
;; load 10 and 2 onto the stack
f32.const 10
f32.const 2
f32.min ;; calculate the lower number
call $log ;; log the result
)
(start $main)
)const url = "{%wasm-url%}";
await WebAssembly.instantiateStreaming(fetch(url), { console });value_type.min
value_typeThe type of value the instruction is being run on. The following types support min:
f32f64v128 interpretations:f32x4f64x2minThe min instruction. Must always be included after the value_type and a period (.).
[input1, input2] -> [output]
input1The first input value.
input2The second input value.
outputThe output value.
For a non-SIMD min, the inputs will be basic numeric values such as 3.0 or 3.5, and the output will be the smaller of input1 and input2.
For a SIMD min, the inputs will be v128 value interpretations, for example f32x4 2.0 30 86.9 120. Each lane of the output pushed to the stack is the smaller of the corresponding lanes in the input values.
| Instruction | Binary format | Example text => binary |
|---|---|---|
f32.min | 0x96 | f32.min => 0x96 |
f64.min | 0xa4 | f64.min => 0xa4 |
f32x4.min | 0xfd 232:u32 | f32x4.min => 0xfd 0xe8 0x01 |
f64x2.min | 0xfd 244:u32 | f64x2.min => 0xfd 0xf4 0x01 |
min exampleIn this example, we demonstrate using min to return the smaller value of the same lane index from two SIMD values.
In our script, we grab a reference to a <p> element that we will output our result to, then define an object for import into Wasm containing a single function that writes a value to the output <p>. We then compile and instantiate our Wasm module using the WebAssembly.instantiateStreaming() method, importing the object in the process.
const outputElem = document.querySelector("p");
const obj = {
output(val) {
outputElem.textContent += val;
},
};
WebAssembly.instantiateStreaming(fetch("{%wasm-url%}"), {
obj,
});In our Wasm module, we first import the JavaScript output() function, making sure to declare that it has an f32 parameter. We then declare two SIMD f32x4 values, then use f32x4.min to return a new f32x4 value that contains the lower lane value out of the two inputs in each case. Finally we extract the value stored in lane 3 of the output value using the extract_lane instruction, and output it to the DOM by calling the imported output() function.
(module
;; Import output function
(import "obj" "output" (func $output (param f32)))
(func $main
;; load two SIMD values onto the stack
v128.const f32x4 20 0 1015 1000
v128.const f32x4 4 38 15 108
;; Return a new f32x4 containing the highest lane value in each case
f32x4.min
f32x4.extract_lane 3 ;; Extract a value from the result
call $output
)
(start $main)
)The output is as follows:
The result is 108. This is because the value stored in lane 3 of the first input value is 1000, and the value stored in lane 3 of the second input value is 108. Since 108 is less than 1000, the new f32x4 value outputted by the f32x4.min instruction has 108 set in lane 3, which we then extract and output to the DOM.