🚀 onika – Release Notes v1.0.4

✨ What's New

🧵 Parallel For 3D Iteration Initial implementation of 3D parallel iteration logic. #6 – @carrardt Reverted due to issues: #8 – @rprat-pro Refined and reintroduced: #9 – @lafourcadep
⚙️ GPU Atomic Min/Max Fix Resolved incorrect behavior in atomic min/max operations on GPU. #7 – @carrardt
🧩 Block-Level Voting Functions Added new CUDA block-level vote macros: ONIKACUBLOCKSYNCOR and ONIKACUBLOCKSYNCAND. #10 – @carrardt
Debug Compatibility Enforcement Ensured compatibility between debug builds of Onika and non-debug builds of exaNBody. #11 – @carrardt
🪲 Fix: SIMD Loop with 2D/3D Blocks Corrected an incompatibility in ONIKACUSIMDFORLOOP for multi-dimensional thread blocks. #12 – @carrardt
📦 CUDA-Compatible span Type Introduced a span implementation compatible with CUDA environments. #14 – @carrardt
🧠 Runtime-Named Field Combiners Enabled dynamic naming of field combiners at runtime, improving flexibility for user-defined workflows. #15 – @carrardt 👥 New Contributors
🎉 @carrardt made their first contribution in #6
🎉 @lafourcadep made their first contribution in #9

- C++
Published by rprat-pro 11 months ago

🚀 Parallel for 3D
- implementation by @carrardt (#6)
🧮 GPU Atomic Min/Max Fix
Fixed incorrect behavior in atomic min/max operations.
@carrardt (#7)
🧩 New Block-Level Vote Functions
Introduced ONIKA_CU_BLOCK_SYNC_OR and ONIKA_CU_BLOCK_SYNC_AND for improved thread block coordination.
@carrardt (#10)
🛠️ Build Compatibility Improvements
Ensured debug builds of Onika are compatible with non-debug builds of exaNBody.
@carrardt (#11)
🐞 SIMD Loop Bug Fix
Fixed a compatibility issue in ONIKA_CU_SIMD_FOR_LOOP with 2D/3D block configurations.
@carrardt (#12)
🧵 CUDA-Compatible span
Added a CUDA-compatible version of span.
@carrardt (#14)

📘 Full Changelog: Compare v1.0.0...v1.0.2

- C++
Published by rprat-pro about 1 year ago

First version of Onika:

- C++
Published by rprat-pro about 1 year ago