dynamo/Cargo.toml at main · nv-yna/dynamo · GitHub

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
# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0

[workspace]
members = [
    "launch/dynamo-run",
    "lib/llm",
    "lib/runtime",
    "lib/config",
    "lib/tokens",
    "lib/mocker",
    "lib/kv-router",
    "lib/memory",
    "lib/async-openai",
    "lib/parsers",
    "lib/bindings/c",
    "lib/bindings/python/codegen",
    "lib/engines/*",
    "lib/config",
]
# Exclude certain packages that are slow to build and we don't ship as flagship
# features from default build, but keep them in workspace for convenience.
# - launch/dynamo-run
# - lib/engines/*
default-members = [
    "lib/llm",
    "lib/runtime",
    "lib/config",
    "lib/tokens",
    "lib/mocker",
    "lib/memory",
    "lib/async-openai",
    "lib/parsers",
    "lib/bindings/c",
]
resolver = "3"

[workspace.package]
version = "0.9.0"
edition = "2024"
description = "Dynamo Inference Framework"
authors = ["NVIDIA Inc. <sw-dl-dynamo@nvidia.com>"]
license = "Apache-2.0"
homepage = "https://github.com/ai-dynamo/dynamo"
repository = "https://github.com/ai-dynamo/dynamo.git"
keywords = ["llm", "genai", "inference", "nvidia", "distributed"]

[workspace.dependencies]
# Local crates
dynamo-runtime = { path = "lib/runtime", version = "0.9.0" }
dynamo-llm = { path = "lib/llm", version = "0.9.0" }
dynamo-config = { path = "lib/config", version = "0.9.0" }
dynamo-tokens = { path = "lib/tokens", version = "0.9.0" }
dynamo-memory = { path = "lib/memory", version = "0.9.0" }
dynamo-mocker = { path = "lib/mocker", version = "0.9.0" }
dynamo-kv-router = { path = "lib/kv-router", version = "0.9.0", features = ["metrics"] }
dynamo-async-openai = { path = "lib/async-openai", version = "0.9.0", features = ["byot"] }
dynamo-parsers = { path = "lib/parsers", version = "0.9.0" }

# External dependencies
anyhow = { version = "1" }
async-nats = { version = "0.45.0", features = ["service"] }
async-stream = { version = "0.3" }
async-trait = { version = "0.1" }
async_zmq = { version = "0.4.0" }
blake3 = { version = "1" }
bytes = { version = "1" }
chrono = { version = "0.4", default-features = false, features = [
    "alloc",
    "std",
    "clock",
    "now",
    "serde",
] }
cudarc = { version = "0.17.8", features = ["cuda-12020"] }
dashmap = { version = "6.1" }
derive_builder = { version = "0.20" }
derive-getters = { version = "0.5" }
either = { version = "1.13", features = ["serde"] }
etcd-client = { version = "0.17.0", features = ["tls"] }
futures = { version = "0.3" }
hf-hub = { version = "0.4.2", default-features = false, features = [
    "tokio",
    "rustls-tls",
    "ureq",
] }

# ModelExpress for model downloading
modelexpress-client = { version = "0.2.0" }
modelexpress-common = { version = "0.2.0" }

humantime = { version = "2.2.0" }
libc = { version = "0.2" }
oneshot = { version = "0.1.13", features = ["std", "async"] }
parking_lot = "0.12.5"
prometheus = { version = "0.14" }
rand = { version = "0.9.2" }
reqwest = { version = "0.12.24", default-features = false, features = [
    "json",
    "stream",
    "rustls-tls",
] }
rmp-serde = { version = "1" }
# "rc" is for async-openai. Allows serializing Rc and Arc. Generally avoid doing that.
serde = { version = "1", features = ["derive", "rc"] }
serde_json = { version = "1" }
strum = { version = "0.27", features = ["derive"] }
tempfile = "3"
thiserror = { version = "2.0.17" }
tmq = { version = "0.5.0" }
zmq = { version = "0.10" }
tokio = { version = "1", features = ["full"] }
tokio-stream = { version = "0.1" }
tokio-util = { version = "0.7", features = ["codec", "net", "rt", "io-util"] }
tower-http = { version = "0.6", features = ["trace"] }
axum = { version = "=0.8.4", features = ["macros"] }
axum-core = { version = "0.5.2" }
hyper = { version = "=1.7.0" }
hyper-util = { version = "=0.1.17" }
tracing = { version = "0.1" }
tracing-subscriber = { version = "0.3", features = [
    "env-filter",
    "local-time",
    "json",
] }
tracing-opentelemetry = { version = "0.32.0" }
opentelemetry = { version = "0.31.0", features = ["trace"] }
opentelemetry_sdk = { version = "0.31.0", features = ["trace", "rt-tokio"] }
opentelemetry-otlp = { version = "0.31.0", features = ["trace", "grpc-tonic"] }
validator = { version = "0.20.0", features = ["derive"] }
uuid = { version = "1.18.1", features = ["v4", "serde"] }
url = { version = "2.5", features = ["serde"] }
xxhash-rust = { version = "0.8", features = ["xxh3", "const_xxh3"] }

[profile.dev.package]
insta.opt-level = 3

[profile.dev]
# release level optimizations otherwise everything feels slow
# opt-level = 3

[profile.release]
# These make the build much slower but shrink the binary, and could help performance
codegen-units = 1
lto = "thin"