9
9
10
10
| Operation | BLAS | CPU | CUDA | Metal |
11
11
| -----------| ------| ------| ------| ------|
12
- | ABS | ❌ | ✅ | 🟡 | ❌ |
12
+ | ABS | ❌ | ✅ | ✅ | ❌ |
13
13
| ACC | ❌ | ✅ | ✅ | ✅ |
14
14
| ADD | ❌ | ✅ | ✅ | 🟡 |
15
15
| ADD1 | ❌ | ✅ | ✅ | ❌ |
@@ -31,20 +31,20 @@ Legend:
31
31
| DIV | ❌ | ✅ | ✅ | 🟡 |
32
32
| DUP | ❌ | ✅ | 🟡 | 🟡 |
33
33
| ELU | ❌ | ✅ | ❌ | 🟡 |
34
- | EXP | ❌ | ✅ | 🟡 | ❌ |
34
+ | EXP | ❌ | ✅ | ✅ | ❌ |
35
35
| FLASH_ATTN_EXT | ❌ | ✅ | 🟡 | 🟡 |
36
36
| GATED_LINEAR_ATTN | ❌ | ✅ | ✅ | ❌ |
37
37
| GEGLU | ❌ | ✅ | ✅ | 🟡 |
38
38
| GEGLU_ERF | ❌ | ✅ | ✅ | 🟡 |
39
39
| GEGLU_QUICK | ❌ | ✅ | ✅ | 🟡 |
40
- | GELU | ❌ | ✅ | 🟡 | 🟡 |
41
- | GELU_ERF | ❌ | ✅ | 🟡 | 🟡 |
42
- | GELU_QUICK | ❌ | ✅ | 🟡 | 🟡 |
40
+ | GELU | ❌ | ✅ | ✅ | 🟡 |
41
+ | GELU_ERF | ❌ | ✅ | ✅ | 🟡 |
42
+ | GELU_QUICK | ❌ | ✅ | ✅ | 🟡 |
43
43
| GET_ROWS | ❌ | ✅ | 🟡 | ✅ |
44
44
| GET_ROWS_BACK | ❌ | 🟡 | 🟡 | ❌ |
45
45
| GROUP_NORM | ❌ | ✅ | ✅ | ✅ |
46
- | HARDSIGMOID | ❌ | ✅ | 🟡 | ❌ |
47
- | HARDSWISH | ❌ | ✅ | 🟡 | ❌ |
46
+ | HARDSIGMOID | ❌ | ✅ | ✅ | ❌ |
47
+ | HARDSWISH | ❌ | ✅ | ✅ | ❌ |
48
48
| IM2COL | ❌ | ✅ | ✅ | 🟡 |
49
49
| L2_NORM | ❌ | ✅ | ✅ | ✅ |
50
50
| LEAKY_RELU | ❌ | ✅ | ✅ | ✅ |
@@ -53,15 +53,15 @@ Legend:
53
53
| MUL | ❌ | ✅ | ✅ | 🟡 |
54
54
| MUL_MAT | 🟡 | 🟡 | 🟡 | 🟡 |
55
55
| MUL_MAT_ID | ❌ | ✅ | ✅ | ✅ |
56
- | NEG | ❌ | ✅ | 🟡 | 🟡 |
56
+ | NEG | ❌ | ✅ | ✅ | 🟡 |
57
57
| NORM | ❌ | ✅ | ✅ | 🟡 |
58
58
| OPT_STEP_ADAMW | ❌ | ✅ | ✅ | ❌ |
59
59
| OUT_PROD | 🟡 | 🟡 | 🟡 | ❌ |
60
60
| PAD | ❌ | ✅ | ✅ | ✅ |
61
61
| PAD_REFLECT_1D | ❌ | ✅ | ❌ | ✅ |
62
62
| POOL_2D | ❌ | ✅ | ✅ | ✅ |
63
63
| REGLU | ❌ | ✅ | ✅ | 🟡 |
64
- | RELU | ❌ | ✅ | 🟡 | 🟡 |
64
+ | RELU | ❌ | ✅ | ✅ | 🟡 |
65
65
| REPEAT | ❌ | ✅ | 🟡 | ✅ |
66
66
| REPEAT_BACK | ❌ | ✅ | ✅ | ❌ |
67
67
| RMS_NORM | ❌ | ✅ | ✅ | 🟡 |
@@ -74,9 +74,9 @@ Legend:
74
74
| SCALE | ❌ | ✅ | ✅ | ✅ |
75
75
| SET | ❌ | ✅ | ❌ | ✅ |
76
76
| SET_ROWS | ❌ | 🟡 | ❌ | 🟡 |
77
- | SGN | ❌ | ✅ | 🟡 | ❌ |
78
- | SIGMOID | ❌ | ✅ | 🟡 | 🟡 |
79
- | SILU | ❌ | ✅ | 🟡 | 🟡 |
77
+ | SGN | ❌ | ✅ | ✅ | ❌ |
78
+ | SIGMOID | ❌ | ✅ | ✅ | 🟡 |
79
+ | SILU | ❌ | ✅ | ✅ | 🟡 |
80
80
| SILU_BACK | ❌ | ✅ | ✅ | ❌ |
81
81
| SIN | ❌ | ✅ | ✅ | 🟡 |
82
82
| SOFT_MAX | ❌ | ✅ | ✅ | ✅ |
@@ -85,11 +85,11 @@ Legend:
85
85
| SQRT | ❌ | ✅ | ✅ | 🟡 |
86
86
| SSM_CONV | ❌ | ✅ | ✅ | ✅ |
87
87
| SSM_SCAN | ❌ | ✅ | ✅ | ✅ |
88
- | STEP | ❌ | ✅ | 🟡 | ❌ |
88
+ | STEP | ❌ | ✅ | ✅ | ❌ |
89
89
| SUB | ❌ | ✅ | ✅ | 🟡 |
90
90
| SUM | ❌ | ✅ | ✅ | ❌ |
91
91
| SUM_ROWS | ❌ | ✅ | ✅ | ✅ |
92
92
| SWIGLU | ❌ | ✅ | ✅ | 🟡 |
93
- | TANH | ❌ | ✅ | 🟡 | 🟡 |
93
+ | TANH | ❌ | ✅ | ✅ | 🟡 |
94
94
| TIMESTEP_EMBEDDING | ❌ | ✅ | ✅ | ✅ |
95
95
| UPSCALE | ❌ | ✅ | ✅ | 🟡 |
0 commit comments