Skip to content

feat: add float16 dtype support to array/min-dtype #14018

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Erennn7 wants to merge 3 commits into
base: develop
Choose a base branch
from
Open

feat: add float16 dtype support to array/min-dtype #14018

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b0d5141
feat: add float16 dtype support to array/min-dtype
Erennn7 Aug 7, 2026
bad55e0
Merge branch 'develop' into float16-support/array/min-dtype
gururaj1512 Aug 9, 2026
dd2677f
fix: update comments and JsDOC
gururaj1512 Aug 9, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion lib/node_modules/@stdlib/array/min-dtype/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ Returns the minimum array [data type][@stdlib/array/dtypes] of the closest "kind

```javascript
var dt = minDataType( 3.141592653589793 );
// returns 'float32'
// returns 'float16'

dt = minDataType( -3 );
// returns 'int8'
Expand Down
2 changes: 1 addition & 1 deletion lib/node_modules/@stdlib/array/min-dtype/docs/repl.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@
Examples
--------
> var dt = {{alias}}( 3.141592653589793 )
'float32'
'float16'
> dt = {{alias}}( 3 )
'uint8'
> dt = {{alias}}( -3 )
Expand Down
2 changes: 1 addition & 1 deletion lib/node_modules/@stdlib/array/min-dtype/docs/types/index.d.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,7 @@
*
* @example
* var dt = minDataType( 3.141592653589793 );
* // returns 'float32'
* // returns 'float16'
*
* @example
* var dt = minDataType( 3 );
Expand Down Expand Up @@ -89,7 +89,7 @@
* var dt = minDataType( 'beep' );
* // returns 'generic'
*/
declare function minDataType( value: any ): 'generic';

Check warning on line 92 in lib/node_modules/@stdlib/array/min-dtype/docs/types/index.d.ts

View workflow job for this annotation

GitHub Actions / Lint Changed Files 

Unexpected any. Specify a different type


// EXPORTS //
Expand Down
2 changes: 1 addition & 1 deletion lib/node_modules/@stdlib/array/min-dtype/lib/index.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,7 +27,7 @@
* var minDataType = require( '@stdlib/array/min-dtype' );
*
* var dt = minDataType( 3.141592653589793 );
* // returns 'float32'
* // returns 'float16'
*
* dt = minDataType( 3 );
* // returns 'uint8'
Expand Down
61 changes: 46 additions & 15 deletions lib/node_modules/@stdlib/array/min-dtype/lib/main.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,10 @@ var isNegativeZero = require( '@stdlib/math/base/assert/is-negative-zero' );
var isComplexLike = require( '@stdlib/assert/is-complex-like' );
var PINF = require( '@stdlib/constants/float64/pinf' );
var NINF = require( '@stdlib/constants/float64/ninf' );
var FLOAT16_SMALLEST_SUBNORMAL = require( '@stdlib/constants/float16/smallest-subnormal' ); // eslint-disable-line id-length
var FLOAT16_MAX_SAFE_INTEGER = require( '@stdlib/constants/float16/max-safe-integer' );
var FLOAT16_MIN_SAFE_INTEGER = require( '@stdlib/constants/float16/min-safe-integer' );
var FLOAT16_MAX = require( '@stdlib/constants/float16/max' );
var FLOAT32_SMALLEST_SUBNORMAL = require( '@stdlib/constants/float32/smallest-subnormal' ); // eslint-disable-line id-length
var FLOAT32_MAX_SAFE_INTEGER = require( '@stdlib/constants/float32/max-safe-integer' );
var FLOAT32_MIN_SAFE_INTEGER = require( '@stdlib/constants/float32/min-safe-integer' );
Expand All @@ -49,22 +53,40 @@ var UINT32_MAX = require( '@stdlib/constants/uint32/max' );
*/
function minFloatDataType( value ) {
if ( value !== value || value === PINF || value === NINF ) {
return 'float32';
return 'float16';
}
if ( isInteger( value ) ) {
if ( value >= FLOAT32_MIN_SAFE_INTEGER && value <= FLOAT32_MAX_SAFE_INTEGER ) { // eslint-disable-line max-len
if (
value >= FLOAT16_MIN_SAFE_INTEGER &&
value <= FLOAT16_MAX_SAFE_INTEGER
) {
return 'float16';
}
if (
value >= FLOAT32_MIN_SAFE_INTEGER &&
value <= FLOAT32_MAX_SAFE_INTEGER
) {
return 'float32';
}
return 'float64';
}
// Assume that if we are provided a tiny value, we don't want to underflow to zero by storing as `float32`...
// Assume that if we are provided a tiny value, we don't want to underflow to zero by storing as `float16`...

@kgryte kgryte Aug 9, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
// Assume that if we are provided a tiny value, we don't want to underflow to zero by storing as `float16`...
// Assume that if we are provided a tiny value, we don't want to underflow to zero by storing in a smaller precision...

if (
value > -FLOAT32_SMALLEST_SUBNORMAL &&
value < FLOAT32_SMALLEST_SUBNORMAL
value > -FLOAT16_SMALLEST_SUBNORMAL &&
value < FLOAT16_SMALLEST_SUBNORMAL
) {
return 'float64';
if (
value > -FLOAT32_SMALLEST_SUBNORMAL &&
value < FLOAT32_SMALLEST_SUBNORMAL
) {
return 'float64';
}
return 'float32';
}
// Any number which reaches this point is checked against the maximum half-precision floating-point number, as floating-point format supports a limited number of decimals (e.g., (1.0+EPS)*10**15 => 1000000000000000.2, which is less than ~65504)...
if ( value >= -FLOAT16_MAX && value <= FLOAT16_MAX ) {
return 'float16';
}
// Any number which reaches this point is less than the maximum single-precision floating-point number, as floating-point format supports a limited number of decimals (e.g., (1.0+EPS)*10**15 => 1000000000000000.2, which is less than ~3.4e38)...
return 'float32';
}

Expand All @@ -79,7 +101,7 @@ function minFloatDataType( value ) {
*
* @example
* var dt = minDataType( 3.141592653589793 );
* // returns 'float32'
* // returns 'float16'
*
* @example
* var dt = minDataType( 3 );
Expand All @@ -99,11 +121,11 @@ function minDataType( value ) {
return 'generic';
}
if ( value !== value || value === PINF || value === NINF ) {
return 'float32';
return 'float16';
}
if ( isInteger( value ) ) {
if ( value === 0 && isNegativeZero( value ) ) {
return 'float32';
return 'float16';
}
if ( value < 0 ) {
if ( value >= INT8_MIN ) {
Expand All @@ -128,14 +150,23 @@ function minDataType( value ) {
}
return 'float64';
}
// Assume that if we are provided a tiny value, we don't want to underflow to zero by storing as `float32`...
// Assume that if we are provided a tiny value, we don't want to underflow to zero by storing as `float16`...
if (
value > -FLOAT32_SMALLEST_SUBNORMAL &&
value < FLOAT32_SMALLEST_SUBNORMAL
value > -FLOAT16_SMALLEST_SUBNORMAL &&
value < FLOAT16_SMALLEST_SUBNORMAL
) {
return 'float64';
if (
value > -FLOAT32_SMALLEST_SUBNORMAL &&
value < FLOAT32_SMALLEST_SUBNORMAL
) {
return 'float64';
}
return 'float32';
}
// Any number which reaches this point is checked against the maximum half-precision floating-point number, as floating-point format supports a limited number of decimals (e.g., (1.0+EPS)*10**15 => 1000000000000000.2, which is less than ~65504)...
if ( value >= -FLOAT16_MAX && value <= FLOAT16_MAX ) {
return 'float16';
}
// Any number which reaches this point is less than the maximum single-precision floating-point number, as floating-point format supports a limited number of decimals (e.g., (1.0+EPS)*10**15 => 1000000000000000.2, which is less than ~3.4e38)...
return 'float32';
}

Expand Down
3 changes: 3 additions & 0 deletions lib/node_modules/@stdlib/array/min-dtype/package.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,6 +58,9 @@
"memory",
"minimum",
"dtype",
"float16",
"half",
"half-precision",
"utilities",
"utility",
"utils",
Expand Down
12 changes: 6 additions & 6 deletions lib/node_modules/@stdlib/array/min-dtype/test/test.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -157,10 +157,10 @@ tape( 'the function returns the minimum array data type of the closest "kind" ne
}
];
expected = [
'float32',
'float16',
'uint8',
'uint8',
'float32',
'float16',
'uint8',
'int8',
'uint16',
Expand All @@ -169,14 +169,14 @@ tape( 'the function returns the minimum array data type of the closest "kind" ne
'int32',
'float64',
'float64',
'float32',
'float32',
'float16',
'float16',
'float64',
'float64',
'float64',
'float64',
'float32',
'float32',
'float16',
'float16',
'generic',
'generic',
'bool',
Expand Down