ferrum-kernels 0.7.7

Unified compute kernels (CUDA/Metal/CPU) and model runner for Ferrum inference
Documentation 
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
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201

# THIS FILE IS AUTOMATICALLY GENERATED BY CARGO
#
# When uploading crates to the registry Cargo will automatically
# "normalize" Cargo.toml files for maximal compatibility
# with all versions of Cargo and also rewrite `path` dependencies
# to registry (e.g., crates.io) dependencies.
#
# If you are reading this file be aware that the original Cargo.toml
# will likely look very different (and much more reasonable).
# See Cargo.toml.orig for the original contents.

[package]
edition = "2021"
name = "ferrum-kernels"
version = "0.7.7"
authors = ["Ferrum Team"]
build = "build.rs"
autolib = false
autobins = false
autoexamples = false
autotests = false
autobenches = false
description = "Unified compute kernels (CUDA/Metal/CPU) and model runner for Ferrum inference"
readme = "README.md"
license = "MIT"

[features]
cuda = [
    "candle-core/cuda",
    "dep:cudarc",
    "marlin",
    "cudarc?/nccl",
]
default = []
fa2-source = ["cuda"]
marlin = []
metal = ["dep:metal"]
tensor-parallel = ["cuda"]
triton-kernels = ["cuda"]
vllm-marlin = ["cuda"]
vllm-moe-marlin = [
    "cuda",
    "vllm-marlin",
]
vllm-paged-attn-v2 = ["cuda"]

[lib]
name = "ferrum_kernels"
path = "src/lib.rs"

[[test]]
name = "attention_gemm_test"
path = "tests/attention_gemm_test.rs"

[[test]]
name = "attention_metal_test"
path = "tests/attention_metal_test.rs"

[[test]]
name = "attention_paged_attention_test"
path = "tests/attention_paged_attention_test.rs"

[[test]]
name = "attention_paged_kv_append_test"
path = "tests/attention_paged_kv_append_test.rs"

[[test]]
name = "attention_transformer_test"
path = "tests/attention_transformer_test.rs"

[[test]]
name = "cudarc_graph_repro"
path = "tests/cudarc_graph_repro.rs"

[[test]]
name = "flash_attn_batched_eq"
path = "tests/flash_attn_batched_eq.rs"

[[test]]
name = "int8_kv_parity"
path = "tests/int8_kv_parity.rs"

[[test]]
name = "kv_cache_append_batched_eq"
path = "tests/kv_cache_append_batched_eq.rs"

[[test]]
name = "moe_align_block_size_eq"
path = "tests/moe_align_block_size_eq.rs"

[[test]]
name = "paged_decode_attn_bench"
path = "tests/paged_decode_attn_bench.rs"

[[test]]
name = "qk_norm_rope_batched_eq"
path = "tests/qk_norm_rope_batched_eq.rs"

[[test]]
name = "triton_add_bias_eq"
path = "tests/triton_add_bias_eq.rs"

[[test]]
name = "triton_fused_add_rms_norm_eq"
path = "tests/triton_fused_add_rms_norm_eq.rs"

[[test]]
name = "triton_fused_moe_bench"
path = "tests/triton_fused_moe_bench.rs"

[[test]]
name = "triton_fused_moe_eq"
path = "tests/triton_fused_moe_eq.rs"

[[test]]
name = "triton_fused_moe_tile_sweep"
path = "tests/triton_fused_moe_tile_sweep.rs"

[[test]]
name = "triton_fused_silu_mul_eq"
path = "tests/triton_fused_silu_mul_eq.rs"

[[test]]
name = "triton_gelu_eq"
path = "tests/triton_gelu_eq.rs"

[[test]]
name = "triton_layer_norm_eq"
path = "tests/triton_layer_norm_eq.rs"

[[test]]
name = "triton_residual_add_eq"
path = "tests/triton_residual_add_eq.rs"

[[test]]
name = "triton_residual_add_inplace_eq"
path = "tests/triton_residual_add_inplace_eq.rs"

[[test]]
name = "triton_rms_norm_eq"
path = "tests/triton_rms_norm_eq.rs"

[[test]]
name = "triton_softmax_eq"
path = "tests/triton_softmax_eq.rs"

[[test]]
name = "triton_w4a16_eq"
path = "tests/triton_w4a16_eq.rs"

[[test]]
name = "vllm_gptq_repack_smoke"
path = "tests/vllm_gptq_repack_smoke.rs"

[dependencies.candle-core]
version = "0.9.2"

[dependencies.cudarc]
version = "0.19"
features = [
    "driver",
    "cublas",
    "f16",
    "nvrtc",
    "cuda-version-from-build-system",
]
optional = true
default-features = false

[dependencies.ferrum-bench-core]
version = "0.7.7"

[dependencies.ferrum-interfaces]
version = "0.7.7"

[dependencies.ferrum-types]
version = "0.7.7"

[dependencies.half]
version = "2.5.0"
features = [
    "num-traits",
    "use-intrinsics",
    "rand_distr",
]

[dependencies.metal]
version = "0.31"
optional = true

[dependencies.rayon]
version = "1.11"

[dependencies.serde_json]
version = "1.0"

[dependencies.tracing]
version = "0.1"

[build-dependencies.bindgen_cuda]
version = "0.1.6"