[webkit-reviews] review granted: [Bug 234408] [JSC] OpPow should have a "small int exponent" fast path at lower tiers : [Attachment 447425] Patch

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Sat Dec 18 22:14:26 PST 2021


Yusuke Suzuki <ysuzuki at apple.com> has granted Ross Kirsling
<ross.kirsling at sony.com>'s request for review:
Bug 234408: [JSC] OpPow should have a "small int exponent" fast path at lower
tiers
https://bugs.webkit.org/show_bug.cgi?id=234408

Attachment 447425: Patch

https://bugs.webkit.org/attachment.cgi?id=447425&action=review




--- Comment #8 from Yusuke Suzuki <ysuzuki at apple.com> ---
Comment on attachment 447425
  --> https://bugs.webkit.org/attachment.cgi?id=447425
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=447425&action=review

r=me

> Source/JavaScriptCore/llint/LowLevelInterpreter64.asm:1341
> +    addq numberTag, t0

What is the purpose of this addition? t0 is already 1, so just ci2ds is fine
(it is not boxed double etc.).

> JSTests/microbenchmarks/pow-double-int.js:7
> +for (let i = -50; i <= 50; i += 0.1)
> +    for (let j = 0; j <= 1000; j++) 
> +	   exponentiate(i, j);

Can you add a test case getting Infinity, -Infinity, and NaN for double input?


More information about the webkit-reviews mailing list