Extensive fp cleanup, support Inf/NaN, addition of simple fp multiplier #152
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mkorbel1
reviewed
Jan 8, 2025
mkorbel1
reviewed
Jan 8, 2025
lib/src/arithmetic/floating_point/floating_point_adder_round.dart
Outdated
Show resolved
Hide resolved
lib/src/arithmetic/values/floating_point_values/floating_point_8_value.dart
Outdated
Show resolved
Hide resolved
test/arithmetic/floating_point/floating_point_adder_simple_test.dart
Outdated
Show resolved
Hide resolved
test/arithmetic/floating_point/floating_point_multiplier_test.dart
Outdated
Show resolved
Hide resolved
mkorbel1
reviewed
Jan 8, 2025
lib/src/arithmetic/floating_point/floating_point_adder_simple.dart
Outdated
Show resolved
Hide resolved
mkorbel1
reviewed
Jan 9, 2025
lib/src/arithmetic/floating_point/floating_point_adder_simple.dart
Outdated
Show resolved
Hide resolved
mkorbel1
reviewed
Jan 14, 2025
lib/src/arithmetic/floating_point/floating_point_multiplier_simple.dart
Outdated
Show resolved
Hide resolved
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description & Motivation
Addition of FloatingPointMultiplierSimple class.
Refactor floating point adders to use a base FloatingPointAdder class.
Exhaustive testing expanded to all possible inputs, validating the expected Infinite and NaN cases for FloatingPointAdderSimple,
FloatingPointAdderRound, and FloatingPointMultiplierSimple. Fixed FloatingPointValue to appropriately test Inf and NaN. Fixed the narrow-width FP special cases (e.g. FP8E4M3 does not support NaN).
Pipelined all three fp components.
Added a NativeAdder to use as a functor to use synthesized adders where convenient.
Added a SignedShifter component.
Modified PrefixTreePriorityEncoder to return the position 1 larger than the input width if no '1' exists in the input.
Optionally output a sentinal to indicate this as well. This is used in the fp adders.
Related Issue(s)
#133
#127 This is a proposed solution to use sentinals.
Testing
Exhaustive and random tests for all new components except the SignedShifter (used in fp multiplier).
Backwards-compatibility
Yes: PrefixTreePriorityEncoder used to return position 0 when input was all zeros. Now it returns input.width+1 and sets the sentinal.
Documentation
Yes. The NativeAdder is documented. The new FloatingPointMultiplierSimple needs documentation as well as a mention of pipelining in the FloatingPointAdderSimple. Need to check the PrefixTreePriorityFinder MD file for updates.