Adds four Criterion.rs benchmark suites to measure hot-path performance
and demonstrate the impact of Sprints 1–3 optimizations:
- benches/protocol.rs — Command & StatusMessage serialize/deserialize
- benches/bandwidth.rs — BandwidthState atomics, budget, interval math
- benches/tcp_rx_scan.rs — memchr SIMD scan vs naive O(n) loop (55× faster
on 256KB buffers with status at end)
- benches/ecsrp5.rs — WCurve::new() heavy math vs cached LazyLock
(~123,000× faster access)
Also adds BENCHMARKS.md with usage instructions and example results.
Visibility changes (bench-only):
- scan_status_message is now pub (was #[cfg(test)] only)
- WCurve and WCURVE are now pub in ecsrp5.rs
dev-dependencies: criterion + pprof (optional flamegraph support)
This commit fixes the most significant hot-path bottleneck in the
client: the tcp_client_rx_loop was scanning up to 256KB byte-by-byte
on every read() call looking for interleaved 12-byte status messages.
Changes:
- client.rs (tcp_client_rx_loop): Replace the O(n) for-loop scan
with a three-stage approach:
1. Split-message check: An 11-byte carry buffer stores trailing
bytes from the previous read. We check every possible alignment
where a status message (0x07 + cpu_byte) could span the carry
and the start of the current buffer. This fixes a latent bug
where the old code would miss status messages split across TCP
read boundaries.
2. Fast scan: memchr::memchr (AVX2/NEON SIMD) finds 0x07 bytes
in the 256KB buffer. On all-zero data packets this exits in
~4096 SIMD-width operations instead of 262,144 byte compares.
~64x faster scan path.
3. Carry save: Save up to 11 trailing bytes for the next read.
- client.rs (unit tests): Add scan_status_message() helper and
five unit tests covering:
- Status message fully within buffer
- Status message split across reads (5+7 bytes)
- Status message split at boundary (1+11 bytes)
- All-zero buffer (no false positive)
- Short buffer (no panic)
- Cargo.toml / Cargo.lock: Add memchr as an explicit dependency.
Verified against live MikroTik RouterOS (TCP both + receive modes
with EC-SRP5 auth). Status messages detected correctly. No wire
protocol changes — 100% MikroTik compatible.
Disable release workflow trigger to prevent duplicate builds
when tagging manually with pre-built binaries.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New in v0.6.0:
- CPU usage: local/remote shown per interval (cpu: 12%/33%)
- Warning indicator (!) when CPU > 70% on either side
- MikroTik CPU encoding: 0x80 | percentage in status byte 1
- CSV includes local_cpu_pct and remote_cpu_pct columns
- Status message format corrected to match MikroTik wire format:
[type:1][cpu:1][00:2][seq:4 LE][bytes:4 LE]
- Removed btest-opensource submodule (fully reimplemented)
- Deleted research/ecsrp5 branch
- Updated all docs: architecture, user-guide, man page, protocol
- Version bumped to 0.6.0
58 tests, all passing. Zero warnings.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New features:
- --syslog <address:port> sends structured events to remote syslog (RFC 5424 UDP)
Events: AUTH_SUCCESS, AUTH_FAILURE, TEST_START, TEST_END, TEST_RESULT
- EC-SRP5 authentication for both client and server modes
- TCP multi-connection support (session tokens, all 3 directions)
Bug fixes since v0.2.0:
- EC-SRP5 server: fixed gamma parity (was 50% auth failure rate)
- EC-SRP5 server: use lift_x not redp1 for verification
- TCP send direction: server sends 12-byte status messages to client
- TCP both direction: TX loop injects status between data packets
- TCP data: send all zeros (no 0x07 header that MikroTik rejected)
- TCP disconnect detection: running flag set on EOF
- UDP multi-connection: unconnected socket accepts all source ports
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Rename package to btest-rs (Rust convention for reimplementations)
- MIT license matching the original btest-opensource license
- LICENSE explicitly credits Alex Samorukov's original work
- Comprehensive README with usage, performance numbers, and credits
- CLI --help references the original project
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Full reimplementation of the MikroTik Bandwidth Test protocol:
- Server mode: accepts connections from MikroTik devices on port 2000
- Client mode: connects to MikroTik btest servers
- TCP and UDP protocols with bidirectional support
- MD5 challenge-response authentication
- Dynamic speed adjustment (1.5x algorithm)
- Status exchange matching original C pselect() behavior
- Docker support with multi-stage build
Tested against MikroTik RouterOS achieving:
- 1.05 Gbps server RX (single connection)
- 530 Mbps client TCP download
- 840 Mbps client TCP upload
- 433 Mbps client UDP download
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
