Hugging Face's logo

kernels-community
/
quantization-eetq

kernel
Model card Files Community
1
quantization-eetq
File size: 4,262 Bytes 
1dc29e9
34fd1ec
0c60fb4
1dc29e9
 
 
0c60fb4
 
1dc29e9
 
0c60fb4
 
 
 
 
 
 
1dc29e9
0c60fb4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1dc29e9
 
0c60fb4
 
 
 
 
 
 
 
 
 
 
1dc29e9
0c60fb4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1dc29e9
 
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
 
[general]
name = "quantization_eetq"
universal = false

[torch]
src = [
    "torch-ext/torch_binding.cpp",
    "torch-ext/torch_binding.h",
]

[kernel.weight_only_batched_gemv]
backend = "cuda"
depends = [
    "cutlass_2_10",
    "torch",
]
include = ["cutlass_extensions/include"]
src = [
    "cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/kernel/mixed_gemm_B_layout.h",
    "weightOnlyBatchedGemv/common.h",
    "weightOnlyBatchedGemv/enabled.h",
    "weightOnlyBatchedGemv/kernel.h",
    "weightOnlyBatchedGemv/kernelLauncher.cu",
    "weightOnlyBatchedGemv/kernelLauncher.h",
    "weightOnlyBatchedGemv/utility.h",
    "weightOnlyBatchedGemv/weightOnlyBatchedGemvBs1Int4b.cu",
    "weightOnlyBatchedGemv/weightOnlyBatchedGemvBs1Int8b.cu",
    "weightOnlyBatchedGemv/weightOnlyBatchedGemvBs2Int4b.cu",
    "weightOnlyBatchedGemv/weightOnlyBatchedGemvBs2Int8b.cu",
    "weightOnlyBatchedGemv/weightOnlyBatchedGemvBs3Int4b.cu",
    "weightOnlyBatchedGemv/weightOnlyBatchedGemvBs3Int8b.cu",
    "weightOnlyBatchedGemv/weightOnlyBatchedGemvBs4Int4b.cu",
    "weightOnlyBatchedGemv/weightOnlyBatchedGemvBs4Int8b.cu",
]

[kernel.cutlass_kernels]
backend = "cuda"
depends = [
    "cutlass_2_10",
    "torch",
]
include = [
    ".",
    "utils",
    "cutlass_extensions/include",
]
src = [
    "cutlass_extensions/include/cutlass_extensions/arch/mma.h",
    "cutlass_extensions/include/cutlass_extensions/compute_occupancy.h",
    "cutlass_extensions/include/cutlass_extensions/epilogue/epilogue_quant_helper.h",
    "cutlass_extensions/include/cutlass_extensions/epilogue/thread/ft_fused_activations.h",
    "cutlass_extensions/include/cutlass_extensions/epilogue/threadblock/epilogue_per_row_per_col_scale.h",
    "cutlass_extensions/include/cutlass_extensions/epilogue/threadblock/epilogue_tensor_op_int32.h",
    "cutlass_extensions/include/cutlass_extensions/epilogue_helpers.h",
    "cutlass_extensions/include/cutlass_extensions/ft_gemm_configs.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_fpA_intB_traits.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/kernel/fpA_intB_gemm.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/kernel/fpA_intB_gemm_with_broadcast.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/kernel/mixed_gemm_B_layout.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_base.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/warp/default_mma_tensor_op.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_compute_B_with_f16.h",
    "cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h",
    "cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h",
    "cutlass_extensions/include/cutlass_extensions/tile_interleaved_layout.h",
    "cutlass_kernels/cutlass_heuristic.cu",
    "cutlass_kernels/cutlass_heuristic.h",
    "cutlass_kernels/cutlass_preprocessors.cc",
    "cutlass_kernels/cutlass_preprocessors.h",
    "cutlass_kernels/fpA_intB_gemm.cu",
    "cutlass_kernels/fpA_intB_gemm.h",
    "cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm.h",
    "cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_template.h",
    "cutlass_kernels/fpA_intB_gemm_wrapper.cu",
    "cutlass_kernels/fpA_intB_gemm_wrapper.h",
    "weightOnlyBatchedGemv/common.h",
    "weightOnlyBatchedGemv/enabled.h",
    "utils/activation_types.h",
    "utils/cuda_utils.h",
    "utils/logger.cc",
    "utils/logger.h",
    "utils/string_utils.h",
    "utils/torch_utils.h",
]