Refactor __get_sycl_range to align with sycl

[danhoeflinger]

Align __get_sycl_range with SYCL runtime behavior for write access mode

Fixes #1272

Summary

This PR aligns __get_sycl_range's handling of the write access mode with SYCL runtime semantics by:

1. Adding support for the no_init property to control copy-in behavior
2. Making write mode perform copy-in by default (SYCL-compliant)
3. Fixing transform_if patterns to use proper access modes instead of workarounds
4. Removing vestigial _Iterator template parameter from __get_sycl_range

Changes

Core Implementation

• Added bool _NoInit = false template parameter to __get_sycl_range
• Updated __is_copy_direct_v to make write mode copy-in by default unless no_init is specified
• Now properly aligns with SYCL specification: write implies copy-in unless suppressed with no_init

Preserved Existing Behavior (No Functional Changes)

Updated all existing write mode callsites to use /*_NoInit=*/true, preserving their current no-copy-in behavior:

• algorithm_impl_hetero.h: 8 callsites (copy_if, partition_copy, unique_copy, merge, reverse_copy, rotate_copy, set
  operations)
• numeric_impl_hetero.h: 3 callsites (transform_scan variants, adjacent_difference)
• async_impl_hetero.h: 2 callsites (async operations)
• parallel_backend_sycl.h: 3 callsites (set operation temporary buffers)
• single_pass_scan.h: 1 callsite (kernel template)

Fixed transform_if Workarounds

Changed __pattern_walk2_transform_if and __pattern_walk3_transform_if from using read_write (workaround) to write
(without no_init):

• Uses proper semantics: copy-in to preserve non-transformed elements
• Eliminates confusing access mode misuse that could cause issues with vectorized paths
• Updated comments to reflect the proper solution

Fixed Histogram Pattern

Changed histogram implementation from write workaround to proper read_write + no_init:

• Resolves TODO comment requesting this functionality
• Correctly expresses intent: kernel needs to read bins for atomic updates, but doesn't need initial copy-in

Cleanup

• Removed unused _Iterator template parameter from __get_sycl_range (noted in TODO)

Bonus Bug Fix

• Fixed issue in esimd_radix_sort, where the wrong __get_sycl_range result was used for output keys and values. Caught in review.

[Copilot]

Pull Request Overview

This PR refactors __get_sycl_range to align with SYCL runtime semantics for the write access mode. The primary change introduces a _NoInit template parameter to control copy-in behavior, making write mode perform copy-in by default (SYCL-compliant) unless explicitly suppressed.

Key Changes:

• Added _NoInit template parameter to __get_sycl_range to control copy-in behavior for write access mode
• Updated transform_if patterns to use proper write access mode instead of read_write workaround
• Fixed histogram pattern to use read_write + no_init instead of write workaround

Reviewed Changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
utils_ranges_sycl.h	Core implementation: added _NoInit parameter, removed unused _Iterator parameter, updated __is_copy_direct_v logic
algorithm_impl_hetero.h	Updated all callsites to remove _Iterator parameter; added /*_NoInit=*/true to preserve existing behavior for write mode; fixed transform_if patterns
numeric_impl_hetero.h	Updated callsites to remove _Iterator parameter and add /*_NoInit=*/true for write mode
histogram_impl_hetero.h	Fixed histogram to use read_write + no_init instead of write workaround; removed _Iterator parameter from callsites
parallel_backend_sycl.h	Updated set operation temporary buffers with /*_NoInit=*/true; removed _Iterator parameter
binary_search_impl.h	Removed unused _Iterator template parameter from all __get_sycl_range calls
async_impl_hetero.h	Updated async operations with /*_NoInit=*/true for write mode
glue_async_impl.h	Removed _Iterator parameter from sort_async
single_pass_scan.h	Updated scan kernel template with /*_NoInit=*/true
esimd_radix_sort_dispatchers.h	Removed _Iterator parameter from radix sort dispatcher
esimd_radix_sort.h	Removed _Iterator parameter from all radix sort variants

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.