https://blog.ast.moe/en/tags/float16/Recent posts: GorimuchuHugohttps://blog.ast.moe/en/blog/2026-09-10/Thu, 10 Sep 2026 19:11:38 +0900https://blog.ast.moe/en/blog/2026-09-10/<p>In the pre‑migration verification of Boa for Omoikane (<a href="https://github.com/ieee0824/omoikane/issues/655">issue #655</a>), we ran the same Boa and Test262 on Linux x86_64 and ARM64, and the results differed in only 10 cases.</p> <p>A total of 50,595 cases were executed.</p> <table> <thead> <tr> <th>Environment</th> <th style="text-align: right">Passed</th> <th style="text-align: right">Failed</th> <th style="text-align: right">Ignored</th> </tr> </thead> <tbody> <tr> <td>Linux x86_64</td> <td style="text-align: right">47,606</td> <td style="text-align: right">933</td> <td style="text-align: right">2,056</td> </tr> <tr> <td>Linux ARM64</td> <td style="text-align: right">47,616</td> <td style="text-align: right">923</td> <td style="text-align: right">2,056</td> </tr> </tbody> </table> <p>The ten cases that differed were all tests related to the conversion to <code>Float16</code>.</p> <p>It’s problematic that JavaScript results change when the CPU changes even though the source code and inputs are identical. We investigated what was happening.</p>