asm (
// Convention is low_reg:high_reg
// FIXME: lea for x48 = x11 * 0x13
"mov $0x13, %rdx\t\n"
"mov %[mx11], %[rx86]\t\n"
"mulx %[rx86], %rdx, %[rx86]\t\n" // x48 = x11 * 0x13
"mov %[rx86], %rdx\t\n"
"mov %[mx17], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x55_tmp = x48 * x17
"mov %r13, %r12\t\n" // bucket: x50_high + x52_high + x54_high + x56_high
"mov %r14, %[rx90]\t\n" // bucket: x50_low + x52_low + x54_low + x56_low
// FIXME: lea for x47 = x9 * 0x13
"mov $0x13, %rdx\t\n"
"mov %[mx9], %[rx82]\t\n"
"mulx %[rx82], %rdx, %[rx82]\t\n" // x47 = x9 * 0x13
"mov %[rx82], %rdx\t\n"
"mov %[mx19], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x53_tmp = x47 * x19
"adx %r14, %r12\t\n" // bucket: x50_high + x52_high + x54_high + x56_high
"add %r13, %[rx90]\t\n" // bucket: x50_low + x52_low + x54_low + x56_low
// FIXME: lea for x46 = x7 * 0x13
"mov $0x13, %rdx\t\n"
"mov %[mx7], %r9\t\n"
"mulx %r9, %rdx, %r9\t\n" // x46 = x7 * 0x13
"mov %r9, %rdx\t\n"
"mov %[mx18], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x51_tmp = x46 * x18
"adx %r13, %r12\t\n" // bucket: x50_high + x52_high + x54_high + x56_high
"adc %r14, %[rx90]\t\n" // bucket: x50_low + x52_low + x54_low + x56_low
"mov %[mx5], %rdx\t\n"
"mov %[mx13], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x20_tmp = x5 * x13
"adx %r14, %r12\t\n" // bucket: x50_high + x52_high + x54_high + x56_high
"adc %r13, %[rx90]\t\n" // bucket: x50_low + x52_low + x54_low + x56_low
// FIXME: lea for x45 = x10 * 0x13
"mov $0x13, %rdx\t\n"
"mov %[mx10], %r9\t\n"
"mulx %r9, %rdx, %r9\t\n" // x45 = x10 * 0x13
"mov %r9, %rdx\t\n"
"mov %[mx15], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x49_tmp = x45 * x15
"adx %r13, %r12\t\n" // bucket: x50_high + x52_high + x54_high + x56_high
"adc %r14, %[rx90]\t\n" // bucket: x50_low + x52_low + x54_low + x56_low
"adcx $0, %r12\t\n" // bucket: x50_high + x52_high + x54_high + x56_high
"and $0x7ffffffffffff, %[rx90]\t\n" // x70 = x56_low & 0x7ffffffffffff
"mov %[rx86], %rdx\t\n"
"mov %[mx19], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x61_tmp = x48 * x19
"mov %[rx90], %rdx\t\n"
"mov %[rx90], %r12\t\n"
"mov %rdx, %[rx90]\t\n"
"shrd $0x33, %r12, %[rx90]\t\n" // x69 = x56_low:x56_high >> 0x33
"mov %r12, %[rx89]\t\n" // bucket: x23_low + x58_low + x60_low + x62_low + x71_low
"mov %r14, %[rx79]\t\n" // bucket: x23_high + x58_high + x60_high + x62_high + x71_high
"add %r13, %[rx89]\t\n" // bucket: x23_low + x58_low + x60_low + x62_low + x71_low
"mov %[rx82], %rdx\t\n"
"mov %[mx18], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x59_tmp = x47 * x18
"adx %r13, %[rx79]\t\n" // bucket: x23_high + x58_high + x60_high + x62_high + x71_high
"adc %r14, %[rx89]\t\n" // bucket: x23_low + x58_low + x60_low + x62_low + x71_low
"mov %r9, %rdx\t\n"
"mov %[mx17], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x57_tmp = x45 * x17
"adx %r14, %[rx79]\t\n" // bucket: x23_high + x58_high + x60_high + x62_high + x71_high
"adc %r13, %[rx89]\t\n" // bucket: x23_low + x58_low + x60_low + x62_low + x71_low
"mov %[mx5], %rdx\t\n"
"mov %[mx15], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x21_tmp = x5 * x15
"adx %r13, %[rx79]\t\n" // bucket: x23_high + x58_high + x60_high + x62_high + x71_high
"adc %r14, %[rx89]\t\n" // bucket: x23_low + x58_low + x60_low + x62_low + x71_low
"mov %[mx7], %rdx\t\n"
"mov %[mx13], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x22_tmp = x7 * x13
"adx %r14, %[rx79]\t\n" // bucket: x23_high + x58_high + x60_high + x62_high + x71_high
"adc %r13, %[rx89]\t\n" // bucket: x23_low + x58_low + x60_low + x62_low + x71_low
"adcx $0, %[rx79]\t\n" // bucket: x23_high + x58_high + x60_high + x62_high + x71_high
"and $0x7ffffffffffff, %[rx89]\t\n" // x73 = x71_low & 0x7ffffffffffff
"mov %[rx86], %rdx\t\n"
"mov %[mx18], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x65_tmp = x48 * x18
"mov %[rx89], %rdx\t\n"
"mov %[rx89], %[rx79]\t\n"
"mov %rdx, %[rx89]\t\n"
"shrd $0x33, %[rx79], %[rx89]\t\n" // x72 = x71_low:x71_high >> 0x33
"mov %[rx79], %r12\t\n" // bucket: x26_low + x28_low + x64_low + x66_low + x74_low
"mov %r13, %[rx86]\t\n" // bucket: x26_high + x28_high + x64_high + x66_high + x74_high
"add %r14, %r12\t\n" // bucket: x26_low + x28_low + x64_low + x66_low + x74_low
"mov %r9, %rdx\t\n"
"mov %[mx19], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x63_tmp = x45 * x19
"adx %r14, %[rx86]\t\n" // bucket: x26_high + x28_high + x64_high + x66_high + x74_high
"adc %r13, %r12\t\n" // bucket: x26_low + x28_low + x64_low + x66_low + x74_low
"mov %[mx7], %rdx\t\n"
"mov %[mx15], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x27_tmp = x7 * x15
"adx %r13, %[rx86]\t\n" // bucket: x26_high + x28_high + x64_high + x66_high + x74_high
"adc %r14, %r12\t\n" // bucket: x26_low + x28_low + x64_low + x66_low + x74_low
"mov %[mx5], %rdx\t\n"
"mov %[mx17], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x24_tmp = x5 * x17
"adx %r14, %[rx86]\t\n" // bucket: x26_high + x28_high + x64_high + x66_high + x74_high
"adc %r13, %r12\t\n" // bucket: x26_low + x28_low + x64_low + x66_low + x74_low
"mov %[mx9], %rdx\t\n"
"mov %[mx13], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x25_tmp = x9 * x13
"adx %r13, %[rx86]\t\n" // bucket: x26_high + x28_high + x64_high + x66_high + x74_high
"adc %r14, %r12\t\n" // bucket: x26_low + x28_low + x64_low + x66_low + x74_low
"adcx $0, %[rx86]\t\n" // bucket: x26_high + x28_high + x64_high + x66_high + x74_high
"and $0x7ffffffffffff, %r12\t\n" // x76 = x74_low & 0x7ffffffffffff
"mov %r9, %rdx\t\n"
"mov %[mx18], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x67_tmp = x45 * x18
"mov %r12, %rdx\t\n"
"mov %r12, %[rx86]\t\n"
"mov %rdx, %r12\t\n"
"shrd $0x33, %[rx86], %r12\t\n" // x75 = x74_low:x74_high >> 0x33
"mov %[rx86], %[rx79]\t\n" // bucket: x31_low + x33_low + x35_low + x68_low + x77_low
"mov %r14, %r9\t\n" // bucket: x31_high + x33_high + x35_high + x68_high + x77_high
"add %r13, %[rx79]\t\n" // bucket: x31_low + x33_low + x35_low + x68_low + x77_low
"mov %[mx9], %rdx\t\n"
"mov %[mx15], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x34_tmp = x9 * x15
"adx %r13, %r9\t\n" // bucket: x31_high + x33_high + x35_high + x68_high + x77_high
"adc %r14, %[rx79]\t\n" // bucket: x31_low + x33_low + x35_low + x68_low + x77_low
"mov %[mx7], %rdx\t\n"
"mov %[mx17], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x32_tmp = x7 * x17
"adx %r14, %r9\t\n" // bucket: x31_high + x33_high + x35_high + x68_high + x77_high
"adc %r13, %[rx79]\t\n" // bucket: x31_low + x33_low + x35_low + x68_low + x77_low
"mov %[mx5], %rdx\t\n"
"mov %[mx19], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x29_tmp = x5 * x19
"adx %r13, %r9\t\n" // bucket: x31_high + x33_high + x35_high + x68_high + x77_high
"adc %r14, %[rx79]\t\n" // bucket: x31_low + x33_low + x35_low + x68_low + x77_low
"mov %[mx11], %rdx\t\n"
"mov %[mx13], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x30_tmp = x11 * x13
"adx %r14, %r9\t\n" // bucket: x31_high + x33_high + x35_high + x68_high + x77_high
"adc %r13, %[rx79]\t\n" // bucket: x31_low + x33_low + x35_low + x68_low + x77_low
"adcx $0, %r9\t\n" // bucket: x31_high + x33_high + x35_high + x68_high + x77_high
"and $0x7ffffffffffff, %[rx79]\t\n" // x79 = x77_low & 0x7ffffffffffff
"mov %[mx9], %rdx\t\n"
"mov %[mx17], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x43_tmp = x9 * x17
"mov %[rx79], %rdx\t\n"
"mov %[rx79], %r9\t\n"
"mov %rdx, %[rx79]\t\n"
"shrd $0x33, %r9, %[rx79]\t\n" // x78 = x77_low:x77_high >> 0x33
"mov %r9, %[rx82]\t\n" // bucket: x38_low + x40_low + x42_low + x44_low + x80_low
"mov %r13, %[rx86]\t\n" // bucket: x38_high + x40_high + x42_high + x44_high + x80_high
"add %r14, %[rx82]\t\n" // bucket: x38_low + x40_low + x42_low + x44_low + x80_low
"mov %[mx7], %rdx\t\n"
"mov %[mx19], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x41_tmp = x7 * x19
"adx %r14, %[rx86]\t\n" // bucket: x38_high + x40_high + x42_high + x44_high + x80_high
"adc %r13, %[rx82]\t\n" // bucket: x38_low + x40_low + x42_low + x44_low + x80_low
"mov %[mx11], %rdx\t\n"
"mov %[mx15], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x39_tmp = x11 * x15
"adx %r13, %[rx86]\t\n" // bucket: x38_high + x40_high + x42_high + x44_high + x80_high
"adc %r14, %[rx82]\t\n" // bucket: x38_low + x40_low + x42_low + x44_low + x80_low
"mov %[mx5], %rdx\t\n"
"mov %[mx18], %r14\t\n"
"mulx %r14, %r14, %r13\t\n" // x36_tmp = x5 * x18
"adx %r14, %[rx86]\t\n" // bucket: x38_high + x40_high + x42_high + x44_high + x80_high
"adc %r13, %[rx82]\t\n" // bucket: x38_low + x40_low + x42_low + x44_low + x80_low
"mov %[mx10], %rdx\t\n"
"mov %[mx13], %r13\t\n"
"mulx %r13, %r13, %r14\t\n" // x37_tmp = x10 * x13
"adx %r13, %[rx86]\t\n" // bucket: x38_high + x40_high + x42_high + x44_high + x80_high
"adc %r14, %[rx82]\t\n" // bucket: x38_low + x40_low + x42_low + x44_low + x80_low
"adcx $0, %[rx86]\t\n" // bucket: x38_high + x40_high + x42_high + x44_high + x80_high
"and $0x7ffffffffffff, %[rx82]\t\n" // x82 = x80_low & 0x7ffffffffffff
"mov %[rx82], %rdx\t\n"
"mov %[rx82], %[rx86]\t\n"
"mov %rdx, %[rx82]\t\n"
"shrd $0x33, %[rx86], %[rx82]\t\n" // x81 = x80_low:x80_high >> 0x33
// FIXME: lea for x83 = x81 * 0x13
"mov $0x13, %rdx\t\n"
"mulx %[rx86], %rdx, %[rx86]\t\n" // x83 = x81 * 0x13
"adx %[rx86], %[rx90]\t\n" // bucket: x84 = x70 + x83
"shr $0x33, %[rx90]\t\n" // x85 = x70 >> 0x33
"mov %[rx90], %[rx86]\t\n"
"and $0x7ffffffffffff, %[rx86]\t\n" // x86 = x84 & 0x7ffffffffffff
"adx %[rx89], %[rx90]\t\n" // bucket: x87 = x85 + x73
"shr $0x33, %[rx90]\t\n" // x88 = x85 >> 0x33
"mov %[rx90], %[rx89]\t\n"
"and $0x7ffffffffffff, %[rx89]\t\n" // x89 = x87 & 0x7ffffffffffff
"adx %r12, %[rx90]\t\n" // bucket: x90 = x88 + x76
: [rx89] "=&r" (x89), [rx79] "=&r" (x79), [rx82] "=&r" (x82), [rx90] "=&r" (x90), [rx86] "=&r" (x86)
: [mx9] "m" (x9), [mx7] "m" (x7), [mx5] "m" (x5), [mx10] "m" (x10), [mx11] "m" (x11), [mx13] "m" (x13), [mx15] "m" (x15), [mx17] "m" (x17), [mx18] "m" (x18), [mx19] "m" (x19)
: "cc", "rdx", "r7", "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15"
);