bpo-29882: _Py_popcount32() doesn't need 64x64 multiply (pythonGH-30774)

vstinner · web-flow · commit cd8de40b3b10 · 2022-01-22T00:54:42.000+01:00
32x32 bits multiply is enough for _Py_popcount32().
diff --git a/Include/internal/pycore_bitutils.h b/Include/internal/pycore_bitutils.h
@@ -125,7 +125,7 @@ _Py_popcount32(uint32_t x)
     // Put count of each 8 bits into those 8 bits
     x = (x + (x >> 4)) & M4;
     // Sum of the 4 byte counts
-    return (uint32_t)((uint64_t)x * (uint64_t)SUM) >> 24;
+    return (x * SUM) >> 24;
 #endif
 }
 

Original file line number	Diff line number	Diff line change
`@@ -125,7 +125,7 @@ _Py_popcount32(uint32_t x)`
`125`	`125`	`// Put count of each 8 bits into those 8 bits`
`126`	`126`	`x = (x + (x >> 4)) & M4;`
`127`	`127`	`// Sum of the 4 byte counts`
`128`		`- return (uint32_t)((uint64_t)x * (uint64_t)SUM) >> 24;`
	`128`	`+ return (x * SUM) >> 24;`
`129`	`129`	`#endif`
`130`	`130`	`}`
`131`	`131`