Self-hosted benchmark results for wirestead v0.9.4.
Benchmark Target
- wirestead ref:
v0.9.4
- wirestead commit:
1799b5982b8a8d72495fb723ed02d90113d2693e
- reference platform:
Jetson Orin Nano Super
- platform suffix:
linux-arm64-jetson-orin-nano-super
- benchmark repo commit:
f5bb58e120fee8cf9827f7621732de48b5e22bf3
- GitHub run id:
31925426126
- measured at UTC:
2026-08-16T04:00:52Z
Runner Environment
- runner:
ubuntu-arm64
- host:
ubuntu
- OS:
Ubuntu 22.04.5 LTS
- kernel:
5.15.185-tegra
- architecture:
aarch64
- CPU:
Cortex-A78AE
- logical CPUs:
6
- estimated physical cores:
6
- memory KiB:
7789888
- compiler:
c++ (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0
- CMake:
cmake version 4.3.4
Jetson Environment
- Jetson model:
NVIDIA Jetson Orin Nano Engineering Reference Developer Kit Super
- L4T release: `# R36 (release), REVISION: 5.0, GCID: 43688277, BOARD: generic, EABI: aarch64, DATE: Fri Jan 16 03:50:45 UTC 2026
KERNEL_VARIANT: oot
TARGET_USERSPACE_LIB_DIR=nvidia
TARGET_USERSPACE_LIB_DIR_PATH=usr/lib/aarch64-linux-gnu/nvidia`
- nvpmodel:
NV Power Mode: MAXN_SUPER
- jetson_clocks:
SOC family:tegra234 Machine:NVIDIA Jetson Orin Nano Engineering Reference Developer Kit Super
Latency Summary
| transport |
payload |
runs |
msg/s median |
MiB/s median |
p50 us |
p95 us |
p99 us |
p99.9 us |
max us |
>5ms |
>10ms |
>50ms |
| tcp |
64 |
3 |
8156 |
0.50 |
120 |
135 |
155 |
222 |
881 |
0 |
0 |
0 |
| tcp |
256 |
3 |
7557 |
1.85 |
130 |
139 |
157 |
224 |
496 |
0 |
0 |
0 |
| tcp |
1024 |
3 |
7870 |
7.69 |
127 |
137 |
172 |
288 |
898 |
0 |
0 |
0 |
| tcp |
4096 |
3 |
5296 |
20.69 |
189 |
204 |
224 |
355 |
1726 |
0 |
0 |
0 |
| tcp |
16384 |
3 |
3482 |
54.41 |
283 |
303 |
340 |
570 |
1391 |
0 |
0 |
0 |
| tcp |
65536 |
3 |
1771 |
110.67 |
560 |
595 |
627 |
996 |
2106 |
0 |
0 |
0 |
| udp |
64 |
3 |
9149 |
0.56 |
115 |
123 |
145 |
265 |
862 |
0 |
0 |
0 |
| udp |
256 |
3 |
9394 |
2.29 |
102 |
115 |
149 |
230 |
1293 |
0 |
0 |
0 |
| udp |
1024 |
3 |
8877 |
8.67 |
110 |
119 |
177 |
270 |
844 |
0 |
0 |
0 |
| uds |
64 |
3 |
12224 |
0.75 |
81 |
87 |
103 |
182 |
879 |
0 |
0 |
0 |
| uds |
256 |
3 |
13879 |
3.39 |
71 |
114 |
135 |
171 |
380 |
0 |
0 |
0 |
| uds |
1024 |
3 |
12345 |
12.06 |
75 |
101 |
144 |
220 |
1039 |
0 |
0 |
0 |
| uds |
4096 |
3 |
9130 |
35.66 |
106 |
166 |
180 |
225 |
781 |
0 |
0 |
0 |
| uds |
16384 |
3 |
4914 |
76.78 |
208 |
241 |
262 |
385 |
737 |
0 |
0 |
0 |
| uds |
65536 |
3 |
3463 |
216.43 |
280 |
297 |
307 |
358 |
595 |
0 |
0 |
0 |
Strategy Summary
| transport |
strategy |
payload |
runs |
accepted MiB/s |
received MiB/s |
delivery % |
failed sends |
runtime stats |
dropped msg |
dropped MiB |
max queue KiB |
bp events |
| tcp |
besteffort |
64 |
3 |
32.60 |
32.60 |
100.00 |
5475 |
yes |
5475 |
0.33 |
1024 |
150 |
| tcp |
besteffort |
256 |
3 |
89.11 |
89.11 |
100.00 |
6499 |
yes |
6499 |
1.59 |
1024 |
642 |
| tcp |
besteffort |
1024 |
3 |
340.70 |
340.70 |
100.00 |
5430 |
yes |
5430 |
5.30 |
1024 |
1396 |
| tcp |
besteffort |
4096 |
3 |
1209.76 |
1209.76 |
100.00 |
13786 |
yes |
13786 |
53.85 |
1024 |
6172 |
| tcp |
besteffort |
16384 |
3 |
1217.80 |
1217.80 |
100.00 |
19477 |
yes |
19477 |
304.33 |
1024 |
11048 |
| tcp |
besteffort |
65536 |
3 |
1240.92 |
1240.92 |
100.00 |
18607 |
yes |
18607 |
1162.94 |
1024 |
12006 |
| tcp |
reliable |
64 |
3 |
34.90 |
34.90 |
100.00 |
0 |
yes |
0 |
0 |
1026.44 |
220 |
| tcp |
reliable |
256 |
3 |
89.49 |
89.49 |
100.00 |
0 |
yes |
0 |
0 |
1047 |
354 |
| tcp |
reliable |
1024 |
3 |
339.46 |
339.46 |
100.00 |
0 |
yes |
0 |
0 |
1061 |
1496 |
| tcp |
reliable |
4096 |
3 |
1238.67 |
1238.67 |
100.00 |
0 |
yes |
0 |
0 |
1408 |
7986 |
| tcp |
reliable |
16384 |
3 |
1246.61 |
1246.61 |
100.00 |
0 |
yes |
0 |
0 |
2400 |
13660 |
| tcp |
reliable |
65536 |
3 |
1204.48 |
1204.48 |
100.00 |
0 |
yes |
0 |
0 |
3072 |
12492 |
| udp |
besteffort |
64 |
3 |
8.00 |
7.97 |
99.84 |
22697 |
yes |
22697 |
1.39 |
1024 |
88 |
| udp |
besteffort |
256 |
3 |
29.50 |
29.50 |
100.00 |
22452 |
yes |
22452 |
5.48 |
1024 |
326 |
| udp |
besteffort |
1024 |
3 |
116.48 |
116.37 |
99.91 |
21841 |
yes |
21841 |
21.33 |
1024 |
1210 |
| udp |
besteffort |
4096 |
3 |
380.76 |
366.36 |
96.50 |
21294 |
yes |
21294 |
83.18 |
1024 |
3904 |
| udp |
besteffort |
16384 |
3 |
977.75 |
977.54 |
99.87 |
21085 |
yes |
21085 |
329.45 |
1024 |
8932 |
| udp |
reliable |
64 |
3 |
6.91 |
6.91 |
100.00 |
0 |
yes |
0 |
0 |
1024.81 |
70 |
| udp |
reliable |
256 |
3 |
26.78 |
26.78 |
100.00 |
0 |
yes |
0 |
0 |
1028.50 |
286 |
| udp |
reliable |
1024 |
3 |
97.03 |
97.03 |
100.00 |
0 |
yes |
0 |
0 |
1043 |
1002 |
| udp |
reliable |
4096 |
3 |
350.96 |
350.96 |
100.00 |
0 |
yes |
0 |
0 |
1184 |
3812 |
| udp |
reliable |
16384 |
3 |
1070.77 |
1070.65 |
99.99 |
0 |
yes |
0 |
0 |
1504 |
11750 |
| uds |
besteffort |
64 |
3 |
33.22 |
33.22 |
100.00 |
0 |
yes |
0 |
0 |
797.88 |
0 |
| uds |
besteffort |
256 |
3 |
95.37 |
95.37 |
100.00 |
3134 |
yes |
3134 |
0.77 |
1024 |
442 |
| uds |
besteffort |
1024 |
3 |
337.61 |
337.61 |
100.00 |
1819 |
yes |
1819 |
1.78 |
1024 |
94 |
| uds |
besteffort |
4096 |
3 |
1127.92 |
1127.92 |
100.00 |
8019 |
yes |
8019 |
31.32 |
1024 |
4564 |
| uds |
besteffort |
16384 |
3 |
1301.39 |
1301.39 |
100.00 |
19402 |
yes |
19402 |
303.16 |
1024 |
13362 |
| uds |
besteffort |
65536 |
3 |
1340.52 |
1340.52 |
100.00 |
20139 |
yes |
20139 |
1258.69 |
1024 |
14588 |
| uds |
reliable |
64 |
3 |
25.00 |
25.00 |
100.00 |
0 |
yes |
0 |
0 |
1024.56 |
30 |
| uds |
reliable |
256 |
3 |
91.86 |
91.86 |
100.00 |
0 |
yes |
0 |
0 |
1026.25 |
190 |
| uds |
reliable |
1024 |
3 |
381.84 |
381.84 |
100.00 |
0 |
yes |
0 |
0 |
1035 |
546 |
| uds |
reliable |
4096 |
3 |
1128.99 |
1128.99 |
100.00 |
0 |
yes |
0 |
0 |
1196 |
5720 |
| uds |
reliable |
16384 |
3 |
1488.15 |
1488.15 |
100.00 |
0 |
yes |
0 |
0 |
1664 |
15642 |
| uds |
reliable |
65536 |
3 |
1579.90 |
1579.90 |
100.00 |
15 |
yes |
0 |
0 |
4096 |
18714 |
For BestEffort, accepted throughput alone can be misleading. Use received throughput, delivery rate, dropped messages/bytes, and max queued bytes together.
For Reliable, dropped message counters should normally remain zero. Failed sends indicate payloads rejected before local acceptance.
UDP Payload Smoke Summary
Raw UDP Control
| payload |
send success |
server received |
client received |
echo matches |
delivery % |
match % |
timeouts |
| 1024 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
| 4096 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
| 16384 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
Notes
- raw 4096 succeeds but Wirestead 4096 fails -> likely Wirestead UDP path issue.
- raw 4096 fails and Wirestead 4096 fails -> likely environment / WSL2 / OS behavior.
- raw succeeds and Wirestead succeeds -> strategy pressure model issue.
Wirestead UDP
| strategy |
payload |
send success |
server received |
client received |
echo matches |
delivery % |
match % |
timeouts |
| reliable |
1024 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
| reliable |
4096 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
| reliable |
16384 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
| besteffort |
1024 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
| besteffort |
4096 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
| besteffort |
16384 |
1000 |
1000 |
1000 |
1000 |
100.00 |
100.00 |
0 |
Notes
- No UDP payload smoke anomalies detected.
Classification Guide
- raw UDP succeeds + Wirestead UDP fails -> investigate the Wirestead UDP path.
- raw UDP fails + Wirestead UDP fails -> investigate environment, OS, WSL2, socket buffer, or datagram behavior.
- raw UDP succeeds + Wirestead UDP succeeds + strategy fails -> investigate the pressure benchmark model.
RuntimeStats
| strategy |
payload |
client failed sends |
client drops |
server messages received |
server bytes received |
server failed sends |
server drops |
| reliable |
1024 |
0 |
0 |
1000 |
1024000 |
0 |
0 |
| reliable |
4096 |
0 |
0 |
1000 |
4096000 |
0 |
0 |
| reliable |
16384 |
0 |
0 |
1000 |
16384000 |
0 |
0 |
| besteffort |
1024 |
0 |
0 |
1000 |
1024000 |
0 |
0 |
| besteffort |
4096 |
0 |
0 |
1000 |
4096000 |
0 |
0 |
| besteffort |
16384 |
0 |
0 |
1000 |
16384000 |
0 |
0 |
UDP Classification Guide
- raw UDP succeeds + Wirestead UDP fails -> investigate the Wirestead UDP path.
- raw UDP fails + Wirestead UDP fails -> investigate environment, OS, WSL2, socket buffer, or datagram behavior.
- raw UDP succeeds + Wirestead UDP succeeds + strategy fails -> investigate the pressure benchmark model.
Notes
- Full raw CSV, metadata, hardware record, manifest, and checksum are attached as release assets.
- Latency values are round-trip local echo measurements in microseconds.
- Strategy measurements are one-way streaming measurements.
- UDP latency payloads above
1024 bytes were skipped for this run.