Decimal casting - #8562
Merged
Merged
Conversation
Signed-off-by: "Nicholas Gates" <nick@nickgates.com>
Signed-off-by: "Nicholas Gates" <nick@nickgates.com>
Use array_session in decimal cast tests so the PR merge ref matches develop's test-session import pattern. I, Nicholas Gates <nick@nickgates.com>, hereby add my Signed-off-by to this commit: 25713a5 I, Nicholas Gates <nick@nickgates.com>, hereby add my Signed-off-by to this commit: 8948181 Signed-off-by: Nicholas Gates <nick@nickgates.com>
Merging this PR will not alter performance
|
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | Simulation | chunked_varbinview_canonical_into[(1000, 10)] |
154.8 µs | 191 µs | -18.99% |
| ❌ | Simulation | slice_empty_vortex |
310 ns | 368.3 ns | -15.84% |
| ⚡ | Simulation | chunked_bool_canonical_into[(1000, 10)] |
26.7 µs | 16.2 µs | +64.55% |
| ⚡ | Simulation | chunked_varbinview_canonical_into[(100, 100)] |
259.1 µs | 224.4 µs | +15.49% |
| ⚡ | Simulation | chunked_varbinview_into_canonical[(100, 100)] |
305.7 µs | 271.4 µs | +12.63% |
| ⚡ | Simulation | bitwise_not_vortex_buffer_mut[128] |
273.6 ns | 244.4 ns | +11.93% |
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing ngates/fix-decimal-casting (e2178ef) with develop (4668b6e)
Footnotes
-
4 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩
Generalize the decimal CastKernel buffer-reuse fast path: when the scale is unchanged, the target precision only widens, and the current physical value type already fits the target precision, reuse the values buffer instead of allocating and re-scanning every lane. This restores the zero-copy behavior the vectorized cast regressed for the common widening case (e.g. casting a Decimal(10,2) i64 array to Decimal(18,2)). Also assert (debug-only) that the vectorized fast path and the scalar slow path agree: decimal_cast_error's success arm is unreachable unless they drift. Add a regression test asserting the widening cast shares the source buffer. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Signed-off-by: Nicholas Gates <nick@nickgates.com>
gatesn
enabled auto-merge (squash)
June 23, 2026 23:33
lwwmanning
approved these changes
Jun 24, 2026
robert3005
reviewed
Jun 24, 2026
|
|
||
| assert_eq!(casted.dtype(), &target_dtype); | ||
| let scalar = casted | ||
| .execute_scalar(0, &mut LEGACY_SESSION.create_execution_ctx()) |
Contributor
There was a problem hiding this comment.
NOOOO, I have just fixed all the tests to not use it.
robert3005
reviewed
Jun 24, 2026
Comment on lines
+146
to
+195
| fn cast_decimal_values<F, T>( | ||
| array: ArrayView<'_, Decimal>, | ||
| from_decimal_dtype: DecimalDType, | ||
| to_decimal_dtype: DecimalDType, | ||
| validity: Validity, | ||
| valid_values: &Mask, | ||
| ) -> VortexResult<ArrayRef> | ||
| where | ||
| F: NativeDecimalType, | ||
| T: NativeDecimalType + CheckedMul, | ||
| DecimalValue: From<F>, | ||
| { | ||
| let values = array.buffer::<F>(); | ||
| let values = values.as_slice(); | ||
| let cast_plan = DecimalCastPlan::<T>::new(from_decimal_dtype, to_decimal_dtype); | ||
|
|
||
| let buffer = match valid_values { | ||
| Mask::AllTrue(_) => { | ||
| let mut buffer = BufferMut::<T>::with_capacity(values.len()); | ||
| values | ||
| .try_map_into(&mut buffer.spare_capacity_mut()[..values.len()], |value| { | ||
| cast_plan.cast(value) | ||
| }) | ||
| .map_err(|idx| { | ||
| decimal_cast_error::<F, T>(values[idx], from_decimal_dtype, to_decimal_dtype) | ||
| })?; | ||
| // SAFETY: try_map_into initializes every lane before returning Ok. | ||
| unsafe { buffer.set_len(values.len()) }; | ||
| buffer.freeze() | ||
| } | ||
| Mask::AllFalse(_) => BufferMut::<T>::zeroed(values.len()).freeze(), | ||
| Mask::Values(mask) => { | ||
| let mut buffer = BufferMut::<T>::with_capacity(values.len()); | ||
| values | ||
| .try_map_masked_into( | ||
| mask.bit_buffer(), | ||
| &mut buffer.spare_capacity_mut()[..values.len()], | ||
| |value| cast_plan.cast(value), | ||
| ) | ||
| .map_err(|idx| { | ||
| decimal_cast_error::<F, T>(values[idx], from_decimal_dtype, to_decimal_dtype) | ||
| })?; | ||
| // SAFETY: try_map_masked_into initializes every lane before returning Ok. | ||
| unsafe { buffer.set_len(values.len()) }; | ||
| buffer.freeze() | ||
| } | ||
| }; | ||
|
|
||
| Ok(DecimalArray::new(buffer, to_decimal_dtype, validity).into_array()) | ||
| } |
Contributor
There was a problem hiding this comment.
This should use the same logic that PrimitiveArray cast uses
robert3005
reviewed
Jun 24, 2026
Comment on lines
+164
to
+169
| let rescaled = Self::rescale_i256( | ||
| self.as_i256(), | ||
| from_decimal_dtype.scale(), | ||
| to_decimal_dtype.scale(), | ||
| )?; | ||
| Self::try_from_i256(rescaled, to_decimal_dtype) |
Contributor
There was a problem hiding this comment.
Why would this need to widen to i256?
Contributor
There was a problem hiding this comment.
I see, it's a bit less code to do it in i256
robert3005
reviewed
Jun 24, 2026
|
|
||
| let scaled = DecimalValue::rescale_i256(value, 0, decimal_dtype.scale())?; | ||
| DecimalValue::try_from_i256(scaled, decimal_dtype) | ||
| } |
Contributor
There was a problem hiding this comment.
This seems lazy? There's 3 conversions here which seems excessive
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
No description provided.