-fifcvt-allow-complicated-cmps

Description

If-conversion is enhanced to use more registers to reduce conflicts.

Usage

This optimization is part of the if-conversion optimization of the register transfer language (RTL). To control its enabling status, use the following option:

-param=ifcvt-allow-register-renaming=[0,1,2]: The default value is 0. The value indicates the optimization range.

Note: This optimization requires the -O2 optimization level and must be used together with --param=max-rtl-if-conversion-unpredictable-cost=48 and --param=max-rtl-if-conversion-predictable-cost=48.

Result

The test case is as follows:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24

typedef unsigned int uint16_t;

uint16_t
foo (uint16_t x, uint16_t y, uint16_t z, uint16_t a,
     uint16_t b, uint16_t c, uint16_t d)
{
    int i = 1;
    int j = 1;
    if (a > b)
    {
        j = x;
        if (b > c)
	    i = y;
        else
	    i = z;
    }
    else
    {
        j = y;
        if (c > d)
	    i = z;
    }
    return i * j;
}

Test command:

1

gcc -O2 -fifcvt-allow-complicated-cmps --param max-rtl-if-conversion-unpredictable-cost=100 --param max-rtl-if-conversion-predictable-cost=100 --param=ifcvt-allow-register-renaming=2 -S test.c -o test.s

Figure 1 Option disabled

Figure 2 Option enabled

When this option is enabled, the generated assembly code eliminates branch jumps and uses more registers to reduce conflicts.