Benchmark: Math.imul vs. polyfills 2

Script Preparation code:

var x = 0x12345678;
var y;

Math.imula = function(opA, opB) {
  opB |= 0; // ensure that opB is an integer. opA will automatically be coerced.
  // floating points give us 53 bits of precision to work with plus 1 sign bit
  // automatically handled for our convienence:
  // 1. 0x003fffff /*opA & 0x000fffff*/ * 0x7fffffff /*opB*/ = 0x1fffff7fc00001
  //    0x1fffff7fc00001 < Number.MAX_SAFE_INTEGER /*0x1fffffffffffff*/
  var result = (opA & 0x003fffff) * opB;
  // 2. We can remove an integer coersion from the statement above because:
  //    0x1fffff7fc00001 + 0xffc00000 = 0x1fffffff800001
  //    0x1fffffff800001 < Number.MAX_SAFE_INTEGER /*0x1fffffffffffff*/
  if (opA & 0xffc00000 /*!== 0*/) result += (opA & 0xffc00000) * opB |0;
  return result |0;
};




Math.imulb = function(a, b) {
  var aHi = (a >>> 16) & 0xffff;
  var aLo = a & 0xffff;
  var bHi = (b >>> 16) & 0xffff;
  var bLo = b & 0xffff;
  // the shift by 0 fixes the sign on the high part
  // the final |0 converts the unsigned value into a signed value
  return ((aLo * bLo) + (((aHi * bLo + aLo * bHi) << 16) >>> 0) | 0);
};



Math.imulc = function(x, y) {
   var UINT16 = 0xffff;
 var xn = +x;
 var yn = +y;
 var xl = UINT16 & xn;
  var yl = UINT16 & yn;
 return 0 | xl * yl + ((UINT16 & xn >>> 16) * yl + xl * (UINT16 & yn >>> 16) << 16 >>> 0);
  }

Tests:

Math.imul
for (var i = 0; i < 10000; ++i) { y = Math.imul(x, i); }
Polyfill a
for (var i = 0; i < 10000; ++i) { y = Math.imula(x, i); }
Polyfill b
for (var i = 0; i < 10000; ++i) { y = Math.imulb(x, i); }
Polyfill c
for (var i = 0; i < 10000; ++i) { y = Math.imulc(x, i); }

Rendered benchmark preparation results:

Suite status: <idle, ready to run>

Previous results

Test case name	Result
Math.imul
Polyfill a
Polyfill b
Polyfill c

Fastest: N/A

Slowest: N/A

ⓘ Get performance results from real users on your website with Request Metrics.

Website Performance Monitoring by Request Metrics

Latest run results:

Run details: (Test run date: one year ago)

User agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/105.0.0.0 Safari/537.36 Edg/105.0.1343.42

Browser/OS: Chrome 105 on Windows

View result in a separate tab

Test name	Executions per second
Math.imul	462.7 Ops/sec
Polyfill a	415.6 Ops/sec
Polyfill b	427.7 Ops/sec
Polyfill c	424.9 Ops/sec

Math.imul vs. polyfills 2 (version: 0)

Comparing performance of: Math.imul vs Polyfill a vs Polyfill b vs Polyfill c

Created: one year ago by: Guest

Jump to the latest result

Math.imul

Polyfill a

Polyfill b

Polyfill c

Suite status: <idle, ready to run>

Experimental features:

Fastest: N/A

Slowest: N/A