WIP: Float conv1 #109
WIP: Float conv1 #109
Conversation
lib/src/arithmetic/floating_point/floating_point_converter.dart
Outdated
Show resolved
Hide resolved
lib/src/arithmetic/floating_point/floating_point_converter.dart
Outdated
Show resolved
Hide resolved
lib/src/arithmetic/floating_point/floating_point_converter.dart
Outdated
Show resolved
Hide resolved
| ]), | ||
| Iff(source.isSubnormal() | source.isNormal(), [ | ||
| _result < | ||
| mux( |
There was a problem hiding this comment.
why use a mux here instead of two if statements?
lib/src/arithmetic/floating_point/floating_point_converter.dart
Outdated
Show resolved
Hide resolved
| import 'package:rohd/rohd.dart'; | ||
|
|
||
| void main() { | ||
| const fp32ExponentWidth = 8; |
There was a problem hiding this comment.
this PR should provide these constants for reference:
#110
although, why test these specifically instead of a handful of arbitrary sizes?
| final converter = FloatingPointConverter(fp32, | ||
| destExponentWidth: bf19ExponentWidth, | ||
| destMantissaWidth: bf19MantissaWidth, | ||
| name: 'fp32_to_bf19'); |
There was a problem hiding this comment.
test the whole module works on some things?
|
|
||
| adjustedMantissa <= | ||
| adjustMantissaPrecision(source.mantissa, destMantissaWidth, | ||
| Const(FloatingPointRoundingMode.roundNearestEven.index)); |
There was a problem hiding this comment.
heads up that Const default width is 1
| destMantissaWidth: bf19MantissaWidth, | ||
| name: 'fp32_to_bf19'); | ||
|
|
||
| expect(converter.result.floatingPointValue.toString().contains('x'), false); |
There was a problem hiding this comment.
there's an isValid on LogicValue you can use
| destMantissaWidth: fp16MantissaWidth, | ||
| name: 'fp32_to_bf19'); | ||
|
|
||
| print('Test case:'); |
There was a problem hiding this comment.
you can use the reason field of expect to give more detailed explanations, and '''s to make multi-line strings
| /// | ||
| /// [destExponentWidth] and [destMantissaWidth] are the widths of the exponent | ||
| /// and mantissa of the destination [FloatingPoint] respectively. | ||
| static FloatingPoint handleOverflow( |
There was a problem hiding this comment.
you may want to tag these helper functions with @visibleForTesting unless you think they are useful outside of this converter?
Description & Motivation
Software and Hardware Arbitrary Precision Float Converter