-
Notifications
You must be signed in to change notification settings - Fork 48
/
fma.c
146 lines (132 loc) · 4.65 KB
/
fma.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
#include "emulate.h"
#define FMA32_X_FP16 (1ull << 61)
#define FMA32_Y_FP16 (1ull << 60)
static double fma64_alu(double x, double y, double z, uint64_t operand) {
switch ((operand >> 27) & 7) {
case 1: return x * y;
case 2: return z + x;
case 3: return x;
case 4: return z + y;
case 5: return y;
case 6: return z;
case 7: return 0.;
}
double result;
__asm("fmadd %d0, %d1, %d2, %d3" : "=w"(result) : "w"(x), "w"(y), "w"(z));
return result;
}
static float fma32_alu(float x, float y, float z, uint64_t operand) {
switch ((operand >> 27) & 7) {
case 1: return x * y;
case 2: return z + x;
case 3: return x;
case 4: return z + y;
case 5: return y;
case 6: return z;
case 7: return 0.f;
}
float result;
__asm("fmadd %s0, %s1, %s2, %s3" : "=w"(result) : "w"(x), "w"(y), "w"(z));
return result;
}
static _Float16 fma16_alu(_Float16 x, _Float16 y, _Float16 z, uint64_t operand) {
switch ((operand >> 27) & 7) {
case 1: return x * y;
case 2: return z + x;
case 3: return x;
case 4: return z + y;
case 5: return y;
case 6: return z;
case 7: return 0.f;
}
_Float16 result;
__asm("fmadd %h0, %h1, %h2, %h3" : "=w"(result) : "w"(x), "w"(y), "w"(z));
return result;
}
static void load_xy_reg_fma32(float* dst, const void* src, uint64_t offset, uint64_t fp16) {
load_xy_reg(dst, src, offset);
if (fp16) {
for (uint32_t i = 0; i < 16; ++i) {
float val = dst[i];
__asm("fcvt %s0, %h0" : "=w"(val) : "0"(val));
dst[i] = val;
}
}
}
void emulate_AMX_FMA64(amx_state* state, uint64_t operand) {
uint64_t y_offset = operand & 0x1FF;
uint64_t x_offset = (operand >> 10) & 0x1FF;
uint64_t z_row = (operand >> 20) & 63;
uint64_t x_enable = parse_writemask(operand >> 41, 8, 7);
uint64_t y_enable = parse_writemask(operand >> 32, 8, 7);
double x[8];
double y[8];
load_xy_reg(x, state->x, x_offset);
load_xy_reg(y, state->y, y_offset);
for (int i = 0; i < 8; i++) {
if (!((x_enable >> (i * 8)) & 1)) continue;
if (operand & FMA_VECTOR_PRODUCT) {
double* z = &state->z[z_row].f64[i];
*z = fma64_alu(x[i], y[i], *z, operand);
} else {
for (int j = 0; j < 8; j++) {
if (!((y_enable >> (j * 8)) & 1)) continue;
double* z = &state->z[(j * 8) + (z_row & 7)].f64[i];
*z = fma64_alu(x[i], y[j], *z, operand);
}
}
}
}
void emulate_AMX_FMA32(amx_state* state, uint64_t operand) {
uint64_t y_offset = operand & 0x1FF;
uint64_t x_offset = (operand >> 10) & 0x1FF;
uint64_t z_row = (operand >> 20) & 63;
uint64_t x_enable = parse_writemask(operand >> 41, 4, 7);
uint64_t y_enable = parse_writemask(operand >> 32, 4, 7);
float x[16];
float y[16];
load_xy_reg_fma32(x, state->x, x_offset, operand & FMA32_X_FP16);
load_xy_reg_fma32(y, state->y, y_offset, operand & FMA32_Y_FP16);
for (int i = 0; i < 16; i++) {
if (!((x_enable >> (i * 4)) & 1)) continue;
if (operand & FMA_VECTOR_PRODUCT) {
float* z = &state->z[z_row].f32[i];
*z = fma32_alu(x[i], y[i], *z, operand);
} else {
for (int j = 0; j < 16; j++) {
if (!((y_enable >> (j * 4)) & 1)) continue;
float* z = &state->z[(j * 4) + (z_row & 3)].f32[i];
*z = fma32_alu(x[i], y[j], *z, operand);
}
}
}
}
void emulate_AMX_FMA16(amx_state* state, uint64_t operand) {
uint64_t y_offset = operand & 0x1FF;
uint64_t x_offset = (operand >> 10) & 0x1FF;
uint64_t z_row = (operand >> 20) & 63;
uint64_t x_enable = parse_writemask(operand >> 41, 2, 7);
uint64_t y_enable = parse_writemask(operand >> 32, 2, 7);
_Float16 x[32];
_Float16 y[32];
load_xy_reg(x, state->x, x_offset);
load_xy_reg(y, state->y, y_offset);
for (int i = 0; i < 32; i++) {
if (!((x_enable >> (i * 2)) & 1)) continue;
if (operand & FMA_VECTOR_PRODUCT) {
_Float16* z = &state->z[z_row].f16[i];
*z = fma16_alu(x[i], y[i], *z, operand);
} else {
for (int j = 0; j < 32; j++) {
if (!((y_enable >> (j * 2)) & 1)) continue;
if (operand & FMA_WIDEN_16_32) {
float* z = &state->z[(j * 2) + (i & 1)].f32[i >> 1];
*z = fma32_alu(x[i], y[j], *z, operand);
} else {
_Float16* z = &state->z[(j * 2) + (z_row & 1)].f16[i];
*z = fma16_alu(x[i], y[j], *z, operand);
}
}
}
}
}