isqrt_compilerunrolled

inline int isqrt(uint32_t x) {
    const uint32_t *p = squares;
    int i = 32768;
    
    while (i) {
        if (p[i] <= x) p += i;
        i >>= 1;
    }
    
    return p - squares;
}

_isqrt:
	pushl	%ebp
	movl	%esp, %ebp
	movl	8(%ebp), %edx
	cmpl	$1073741824, %edx
	jae	L2
	movl	$_squares, %eax
L3:
	leal	65536(%eax), %ecx
	cmpl	%edx, 65536(%eax)
	ja	L5
	movl	%ecx, %eax
L5:
	leal	32768(%eax), %ecx
	cmpl	%edx, 32768(%eax)
	ja	L7
	movl	%ecx, %eax
L7:
	leal	16384(%eax), %ecx
	cmpl	%edx, 16384(%eax)
	ja	L9
	movl	%ecx, %eax
L9:
	leal	8192(%eax), %ecx
	cmpl	%edx, 8192(%eax)
	ja	L11
	movl	%ecx, %eax
L11:
	leal	4096(%eax), %ecx
	cmpl	%edx, 4096(%eax)
	ja	L13
	movl	%ecx, %eax
L13:
	leal	2048(%eax), %ecx
	cmpl	%edx, 2048(%eax)
	ja	L15
	movl	%ecx, %eax
L15:
	leal	1024(%eax), %ecx
	cmpl	%edx, 1024(%eax)
	ja	L17
	movl	%ecx, %eax
L17:
	leal	512(%eax), %ecx
	cmpl	%edx, 512(%eax)
	ja	L19
	movl	%ecx, %eax
L19:
	leal	256(%eax), %ecx
	cmpl	%edx, 256(%eax)
	ja	L21
	movl	%ecx, %eax
L21:
	leal	128(%eax), %ecx
	cmpl	%edx, 128(%eax)
	ja	L23
	movl	%ecx, %eax
L23:
	leal	64(%eax), %ecx
	cmpl	%edx, 64(%eax)
	ja	L25
	movl	%ecx, %eax
L25:
	leal	32(%eax), %ecx
	cmpl	%edx, 32(%eax)
	ja	L27
	movl	%ecx, %eax
L27:
	leal	16(%eax), %ecx
	cmpl	%edx, 16(%eax)
	ja	L29
	movl	%ecx, %eax
L29:
	leal	8(%eax), %ecx
	cmpl	%edx, 8(%eax)
	ja	L31
	movl	%ecx, %eax
L31:
	leal	4(%eax), %ecx
	cmpl	%edx, 4(%eax)
	ja	L33
	movl	%ecx, %eax
L33:
	subl	$_squares, %eax
	sarl	$2, %eax
	leave
	ret
	.p2align 2,,3
L2:
	movl	$_squares+131072, %eax
	jmp	L3

isqrt_handunrolled

inline int isqrt(uint32_t x) {
    const uint32_t *p = squares;
    
    if (p[32768] <= x) p += 32768;
    if (p[16384] <= x) p += 16384;
    if (p[8192] <= x) p += 8192;
    if (p[4096] <= x) p += 4096;
    if (p[2048] <= x) p += 2048;
    if (p[1024] <= x) p += 1024;
    if (p[512] <= x) p += 512;
    if (p[256] <= x) p += 256;
    if (p[128] <= x) p += 128;
    if (p[ 64] <= x) p +=  64;
    if (p[ 32] <= x) p +=  32;
    if (p[ 16] <= x) p +=  16;
    if (p[  8] <= x) p +=   8;
    if (p[  4] <= x) p +=   4;
    if (p[  2] <= x) p +=   2;
    if (p[  1] <= x) p +=   1;
    
    return p - squares;
}

_isqrt:
	pushl	%ebp
	movl	%esp, %ebp
	pushl	%ebx
	movl	8(%ebp), %ecx
	cmpl	$1073741823, %ecx
	ja	L18
	movl	$_squares+32768, %eax
	movl	$_squares+65536, %ebx
	movl	$_squares, %edx
L2:
	cmpl	(%ebx), %ecx
	jb	L3
	leal	32768(%ebx), %eax
	movl	%ebx, %edx
L3:
	cmpl	(%eax), %ecx
	jb	L4
	movl	%eax, %edx
L4:
	leal	16384(%edx), %eax
	cmpl	16384(%edx), %ecx
	jb	L5
	movl	%eax, %edx
L5:
	leal	8192(%edx), %eax
	cmpl	8192(%edx), %ecx
	jb	L6
	movl	%eax, %edx
L6:
	leal	4096(%edx), %eax
	cmpl	4096(%edx), %ecx
	jb	L7
	movl	%eax, %edx
L7:
	leal	2048(%edx), %eax
	cmpl	2048(%edx), %ecx
	jb	L8
	movl	%eax, %edx
L8:
	leal	1024(%edx), %eax
	cmpl	1024(%edx), %ecx
	jb	L9
	movl	%eax, %edx
L9:
	leal	512(%edx), %eax
	cmpl	512(%edx), %ecx
	jb	L10
	movl	%eax, %edx
L10:
	leal	256(%edx), %eax
	cmpl	256(%edx), %ecx
	jb	L11
	movl	%eax, %edx
L11:
	leal	128(%edx), %eax
	cmpl	128(%edx), %ecx
	jb	L12
	movl	%eax, %edx
L12:
	leal	64(%edx), %eax
	cmpl	64(%edx), %ecx
	jb	L13
	movl	%eax, %edx
L13:
	leal	32(%edx), %eax
	cmpl	32(%edx), %ecx
	jb	L14
	movl	%eax, %edx
L14:
	leal	16(%edx), %eax
	cmpl	16(%edx), %ecx
	jb	L15
	movl	%eax, %edx
L15:
	leal	8(%edx), %eax
	cmpl	8(%edx), %ecx
	jb	L16
	movl	%eax, %edx
L16:
	leal	4(%edx), %eax
	cmpl	4(%edx), %ecx
	jae	L17
	movl	%edx, %eax
L17:
	subl	$_squares, %eax
	sarl	$2, %eax
	popl	%ebx
	leave
	ret
	.p2align 2,,3
L18:
	movl	$_squares+163840, %eax
	movl	$_squares+196608, %ebx
	movl	$_squares+131072, %edx
	jmp	L2