Fix error in nonzero

c1051d51 · Ted Themistokleous · 7433c2d3 · c1051d51
Commit c1051d51 authored May 12, 2023 by Ted Themistokleous
Hide whitespace changes
Inline Side-by-side

Showing with 1 addition and 1 deletion

src/targets/gpu/device/nonzero.cpp src/targets/gpu/device/nonzero.cpp +1 -1

No files found.
--- a/src/targets/gpu/device/nonzero.cpp
+++ b/src/targets/gpu/device/nonzero.cpp
@@ -46,7 +46,7 @@ argument nonzero(hipStream_t stream, const argument& result, const argument& arg
        auto* ptr          = result.cast<int64_t>();
        gs_launch(stream, block_size, block_size)([=](auto, auto idx) __device__ {
            // fill all output to 0 first
-            idx.local_stride(out_elem_num, [&](auto j) { ptr[j] = 0; });
+            idx.local_stride(out_elem_num-1, [&](auto j) { ptr[j] = 0; });

            block_scan<block_size>(
                idx,