Ivan Enderlin 🦀<p>Taming Floating-Point Sums, <a href="https://orlp.net/blog/taming-float-sums/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">orlp.net/blog/taming-float-sum</span><span class="invisible">s/</span></a>.</p><p>The article explores how to sum an array of floats with two goals: (i) the lowest imprecision possible, (ii) as fast as possible.</p><p>It goes from 5.5Gb/s with a high imprecision to 112Gb/s with an acceptable imprecision.</p><p>It also presents the new `std::intrinsics::fadd_algebraic` Rust function the author has written.</p><p><a href="https://floss.social/tags/RustLang" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>RustLang</span></a> <a href="https://floss.social/tags/performance" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>performance</span></a> <a href="https://floss.social/tags/float" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>float</span></a></p>