hipCUB 2.13.1 for ROCm 5.7.0

Changed

• CUB backend references CUB and Thrust version 2.0.1.
• Fixed DeviceSegmentedReduce::ArgMin and DeviceSegmentedReduce::ArgMax by returning the segment-relative index instead of the absolute one.
• Fixed DeviceSegmentedReduce::ArgMin for inputs where the segment minimum is smaller than the value returned for empty segments. An equivalent fix is applied to DeviceSegmentedReduce::ArgMax.

Known Issues

• debug_synchronous no longer works on CUDA platform. CUB_DEBUG_SYNC should be used to enable those checks.
• DeviceReduce::Sum does not compile on CUDA platform for mixed extended-floating-point/floating-point InputT and OutputT types.
• DeviceHistogram::HistogramEven fails on CUDA platform for [LevelT, SampleIteratorT] = [int, int].
• DeviceHistogram::MultiHistogramEven fails on CUDA platform for [LevelT, SampleIteratorT] = [int, int/unsigned short/float/double] and [LevelT, SampleIteratorT] = [float, double].

hipCUB 2.13.1 for ROCm 5.6.1

hipCUB code for ROCm 5.6.1 did not change. The library was rebuilt for the updated ROCm 5.6.1 stack.

hipCUB 2.13.1 for ROCm 5.6.0

hipCUB code for ROCm 5.6.0 did not change. The library was rebuilt for the updated ROCm 5.6.0 stack.

hipCUB 2.13.1 for ROCm 5.5.1

hipCUB code for ROCm 5.5.1 did not change. The library was rebuilt for the updated ROCm 5.5.1 stack.

hipCUB 2.13.1 for ROCm 5.5.0

Added

• Benchmarks for BlockShuffle, BlockLoad, and BlockStore.

Changed

• CUB backend references CUB and Thrust version 1.17.2.
• Improved benchmark coverage of BlockScan by adding ExclusiveScan, benchmark coverage of BlockRadixSort by adding SortBlockedToStriped, and benchmark coverage of WarpScan by adding Broadcast.

Fixed

• Windows HIP SDK support

Known Issues

• BlockRadixRankMatch is currently broken under the rocPRIM backend.
• BlockRadixRankMatch with a warp size that does not exactly divide the block size is broken under the CUB backend.

hipCUB 2.13.1 for ROCm 5.4.4

Fixed

• Fixed compilation and execution issues for benchmarks with HIP on Windows

hipCUB 2.13.0 for ROCm 5.4.3

hipCUB code for ROCm 5.4.3 did not change. The library was rebuilt for the updated ROCm 5.4.3 stack.

hipCUB 2.13.0 for ROCm 5.4.2

hipCUB code for ROCm 5.4.2 did not change. The library was rebuilt for the updated ROCm 5.4.2 stack.

hipCUB 2.13.0 for ROCm 5.4.1

hipCUB code for ROCm 5.4.1 did not change. The library was rebuilt for the updated ROCm 5.4.1 stack.

hipCUB 2.13.0 for ROCm 5.4.0

Added

• CMake functionality to improve build parallelism of the test suite that splits compilation units by
  function or by parameters.
• New overload for BlockAdjacentDifference::SubtractLeftPartialTile that takes a predecessor item.

Changed

• Improved build parallelism of the test suite by splitting up large compilation units for DeviceRadixSort,
  DeviceSegmentedRadixSort and DeviceSegmentedSort.
• CUB backend references CUB and thrust version 1.17.1.