Fix output of warp-level kernel in reduction user guide
All threads resolved!
All threads resolved!
- Fixes a critical TypeError in the (now called)
_block_local_thread_index_per_dim
function occurring for optimized GPU reductions with dimensionality > 1 - Improves variable names and formatting in
reduction.md
user guide. Optimized example actually uses warp-level reductions now
Edited by Richard Angersbach
Merge request reports
Activity
requested review from @zy69guqi
assigned to @zy69guqi
- Resolved by Frederik Hennig
If there was an error occuring for kernels with
rank > 1
, I think we should extendtest_reductions.py
to also test multidimensional kernels.
added 1 commit
- e597c4c6 - Fix 3d np array creation for min/max case by using random numbers
added 1 commit
- 89942f1c - Adapt get_expected_solution to work with random numbers
mentioned in commit 6c635e09
Please register or sign in to reply