-
Notifications
You must be signed in to change notification settings - Fork 2.1k
/
docker-compose-gpu-runner-cuda-12-0.yml
75 lines (71 loc) · 2.29 KB
/
docker-compose-gpu-runner-cuda-12-0.yml
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
version: '3.2'
services:
reth:
restart: always
image: "ghcr.io/paradigmxyz/reth:v1.0.6"
ports:
- 127.0.0.1:8545:8545
volumes:
- type: volume
source: reth-data
target: /rethdata
- type: bind
source: ./etc/reth/chaindata
target: /chaindata
command: node --dev --datadir /rethdata --http --http.addr 0.0.0.0 --http.port 8545 --http.corsdomain "*" --dev.block-time 300ms --chain /chaindata/reth_config
zk:
image: ghcr.io/matter-labs/zk-environment:cuda-12_0-latest
depends_on:
- reth
- postgres
security_opt:
- seccomp:unconfined
command: tail -f /dev/null
volumes:
- .:/usr/src/zksync
- /usr/src/cache:/usr/src/cache
- /var/run/docker.sock:/var/run/docker.sock
- /usr/src/keys:/mnt/prover_setup_keys
environment:
- CACHE_DIR=/usr/src/cache
- SCCACHE_CACHE_SIZE=50g
- SCCACHE_GCS_BUCKET=matterlabs-infra-sccache-storage
- SCCACHE_GCS_SERVICE_ACCOUNT=gha-ci-runners@matterlabs-infra.iam.gserviceaccount.com
- SCCACHE_ERROR_LOG=/tmp/sccache_log.txt
- SCCACHE_GCS_RW_MODE=READ_WRITE
- CI=1
- GITHUB_WORKSPACE=$GITHUB_WORKSPACE
# We set CUDAARCHS for l4 gpu's
- CUDAARCHS=89
# We need to forward all nvidia-devices, as due to bug with cgroups and nvidia-container-runtime (https://github.com/NVIDIA/libnvidia-container/issues/176#issuecomment-1159454366), cgroups are disabled and thou GPU isn't properly forwarded to dind
devices:
- /dev/nvidia0:/dev/nvidia0
- /dev/nvidiactl:/dev/nvidiactl
- /dev/nvidia-caps:/dev/nvidia-caps
- /dev/nvidia-modeset:/dev/nvidia-modeset
- /dev/nvidia-uvm:/dev/nvidia-uvm
- /dev/nvidia-uvm-tools:/dev/nvidia-uvm-tools
env_file:
- ./.env
extra_hosts:
- "host:host-gateway"
profiles:
- runner
network_mode: host
pid: host
deploy:
resources:
reservations:
devices:
- capabilities: [ gpu ]
postgres:
image: "postgres:14"
command: postgres -c 'max_connections=200'
ports:
- 127.0.0.1:5432:5432
environment:
# We bind only to 127.0.0.1, so setting insecure password is acceptable here
- POSTGRES_PASSWORD=notsecurepassword
volumes:
postgres-data:
reth-data: