算子核(nax-core):零依赖 no_std
- Z_2^64 定点算术 + AVX2 矩阵乘,与 Python 参考实现 215 条逐位一致
- 加性分享 / Beaver 三元组 / 截断对 / edaBit,ChaCha20 PRG,SHA-256
- 自实现银行家舍入(f64::round_ties_even 是 std-only,no_std 编不过)
协议栈
- nax-proto:手写帧协议(12 字节头 + u64 payload),无 protobuf 依赖
- nax-vault:Beaver 乘 / 定点截断 / open / reveal
- nax-node:TCP 传输 + 密文会话 + PSI 会话
- nax-psi:X25519 ECDH-PSI(唯一外部依赖 dalek,因恒定时间约束)
- nax-tee:HAL trait(seal/unseal/attest/sign)+ Mock + TrustPlan 四级降级
验收
- 83 用例全绿;armv7 全 workspace 编译通过
- 两进程密文计算:(1·4+1)+(2·5+1)+(3·6+1) = 35
- 两进程 PSI:[1..5] ∩ [3..7] = {3,4,5}
- 单二进制 450 KB,零运行时依赖
文档
- docs/ARCHITECTURE.md:分层、Ray 定位、TEE 分工与降级梯度
- docs/ZYNQ.md:A9 性能预算(比 i5 慢 8-12x,非 50x)与部署形态
51 lines
1.7 KiB
Bash
Executable File
51 lines
1.7 KiB
Bash
Executable File
#!/usr/bin/env bash
|
||
# 从 Ray head 提交 nax 集群:每个 site 一个 nax-node 进程,按资源标签锁定落点。
|
||
#
|
||
# 前置:
|
||
# - 各机构机器已跑 start-node.sh,带 {<site>_node: 1} 资源标签
|
||
# - nax 二进制已随 runtime-env 分发(或各机器预置)
|
||
set -euo pipefail
|
||
|
||
NAX_BIN="${NAX_BIN:-nax}"
|
||
# 站点列表与监听端口,形如 "alice:9443 bob:9443"
|
||
SITES="${NAX_SITES:-alice:9443 bob:9443}"
|
||
HEAD_NODE="${NAX_HEAD:-alice}"
|
||
SESSION_KEY_FILE="${NAX_SESSION_KEY:-/run/secrets/nax-session}"
|
||
|
||
echo "== nax 集群启动 =="
|
||
echo " binary : ${NAX_BIN}"
|
||
echo " sites : ${SITES}"
|
||
echo " head : ${HEAD_NODE}"
|
||
|
||
# 每个 site 一个长期驻留的 Ray actor,它只负责 spawn 进程并 supervise,
|
||
# 不参与任何计算、不持有数据。
|
||
for entry in $SITES; do
|
||
site="${entry%%:*}"
|
||
port="${entry##*:}"
|
||
echo "-- 提交 site=${site} port=${port}"
|
||
|
||
python3 - "$site" "$port" "$HEAD_NODE" "$NAX_BIN" "$SESSION_KEY_FILE" <<'PY'
|
||
import json, subprocess, sys
|
||
|
||
site, port, head, nax_bin, keyfile = sys.argv[1:6]
|
||
|
||
# Ray 的 actor 声明自己需要哪个资源标签 —— 这是"钉机器"的唯一机制
|
||
resources = {f"{site}_node": 1.0}
|
||
|
||
cmd = [
|
||
nax_bin, "node",
|
||
"--site", site,
|
||
"--listen", f"0.0.0.0:{port}",
|
||
"--head", "true" if site == head else "false",
|
||
"--session-key-file", keyfile,
|
||
]
|
||
|
||
# 骨架:真实实现走 ray.remote(...).options(resources=resources)
|
||
print(json.dumps({"site": site, "resources": resources, "cmd": cmd}, ensure_ascii=False))
|
||
PY
|
||
done
|
||
|
||
echo
|
||
echo "注:当前为骨架(P5 前)。等 nax-node 可运行后,此处改为"
|
||
echo " ray job submit + ray.remote(resources=...) 的实际提交。"
|