Archived tweet #10777 from the @leastfixedpoint Twitter Archive

@fanf Yes -- you're right -- it was just a thought triggered by the ultimate paragraphs in the article, on potential (actually nonexistent, throughput/latencywise) cost differences between xor and sub