Repository navigation
feat: add 8500B jumbo frame packet size to perf tests - #31
Merged
Merged
Conversation
- Add 8500 to default TREX packet sizes (64,512,1400,8500) - Increase echo app recv buffers from 2048 to 10000 bytes for jumbo payloads - Set MTU 9001 on DUT kernel interfaces for jumbo frame support - Add --max-pkt-len=9100 --mbuf-size=10240 to testpmd for jumbo frames - Add --mbuf-factor 8 to TRex startup for large packet memory allocation https://claude.ai/code/session_01294ECCWxYiENcnK5HVVyoL
[CI] Stage: DeployInfrastructure ready.
|
[CI] Stage: SummaryAll tests PASSED. ARP seeding: kernel /proc/net/arp (automatic)
|
✅ Integration Tests Passed (Run 23693401580)Branch: Test Results
Application Logs (last 20 lines)receiver-echo-server.log sender-echo-server.log sender-test-client.log receiver-test-client-iperf.log sender-test-client-iperf.log Full Application Logs (last 200 lines each)receiver-echo-server.logsender-echo-server.logsender-test-client.logreceiver-test-client-iperf.logsender-test-client-iperf.log
|
…ling - Add mbuf_9k: 4096 to trex_cfg.yaml memory config so TRex can allocate mbufs large enough for 8500B packets - Remove --mbuf-factor 8 (explicit pool config is the proper fix) - Add try/except per packet size in run_benchmark.py so one size failing doesn't crash the entire benchmark — partial results are still written https://claude.ai/code/session_01294ECCWxYiENcnK5HVVyoL
[Perf] Stage: DeployDeploying |
[CI] Stage: DeployInfrastructure ready.
|
[Perf] Stage: Instances Ready
|
[Perf] Stage: TRex ConfigStarting TRex configuration (MAC discovery + NIC binding)... |
[Perf] Stage: TRex Config OK
|
[Perf] Stage: TRex StartedTRex server running. Beginning benchmarks... |
[Perf] DUT ReadyDUT instance |
[Perf] Stage: Benchmark (1/3)Running |
[Perf] Benchmark Diag:
|
[Perf] Benchmark Diag:
|
[Perf] Stage: Benchmark (2/3)Running |
[CI] Stage: SummaryAll tests PASSED. ARP seeding: kernel /proc/net/arp (automatic)
|
[Perf] Benchmark Diag:
|
✅ Integration Tests Passed (Run 23694538130)Branch: Test Results
Application Logs (last 20 lines)receiver-echo-server.log sender-echo-server.log sender-test-client.log receiver-test-client-iperf.log sender-test-client-iperf.log Full Application Logs (last 200 lines each)receiver-echo-server.logsender-echo-server.logsender-test-client.logreceiver-test-client-iperf.logsender-test-client-iperf.log
|
[Perf] Benchmark Diag:
|
[Perf] Stage: Benchmark (3/3)Running |
[Perf] Benchmark Diag:
|
[Perf] Benchmark Diag:
|
[Perf] Diag: testpmd logtestpmd output (last 30 lines) |
[Perf] Stage: Results[21:58:57] INFO Generating markdown summary... Performance Test Results — c5n.2xlargeCommit: 1400B packets
512B packets
64B packets
|
- Set DPDK port MTU to 9001 (AWS VPC max) in both the UdpSocket and DpdkBackend paths so the NIC accepts jumbo frames - Increase mempool data_room_size from 2176 to 9344 bytes (9216 + headroom) so mbufs can hold full jumbo frames without truncation - Use force=True in TRex client.start() to bypass the 16 Gbps line rate check — ENA PMD falsely reports 16 Gbps but c5n actually has 25 Gbps Root cause of 100% drop for rust-dpdk 8500B: the DPDK port's default MTU was 1500, so the NIC silently dropped all jumbo frames on RX. https://claude.ai/code/session_01294ECCWxYiENcnK5HVVyoL
[Perf] Stage: DeployDeploying |
- Change instance type from c5n.2xlarge to c6in.xlarge - Remove force=True from TRex to see if c6in ENA reports accurate line rate - c6in.xlarge: 4 vCPUs, 8 GB, 30 Gbps — $0.284/hr vs $0.432/hr https://claude.ai/code/session_01294ECCWxYiENcnK5HVVyoL
- Move try/except to per-rate-step so partial results (e.g. 70K, 140K succeed but 350K hits line rate) are still saved to the JSON - Sort packet sizes numerically in the summary table (64B < 512B < 1400B < 8500B) instead of lexicographic string sort https://claude.ai/code/session_01294ECCWxYiENcnK5HVVyoL
Integration Test Failure (Run 23695262254)Branch: failure-summary.json{
"failed_step": "deploy_infrastructure",
"error": "Infrastructure deployment failed",
"exit_code": 2,
"timestamp": "2026-03-28T22:07:48.749420Z",
"sender_instance_id": "",
"receiver_instance_id": "",
"commit": "c9f6f9d7c322f6704a19edc659177cebd356c0bf",
"run_url": "https://github.com/gspivey/dpdk-stdlib-rust/actions/runs/23695262254"
}```
### instance-0-user-data.log (8463 bytes, last 80 lines)Compiling cexpr v0.6.0 Compiling heck v0.5.0 error[E0308]: mismatched types For more information about this error, try Compiling regex v1.12.3 Compiling smallvec v1.15.1 error[E0308]: mismatched types For more information about this error, try [ 101.646945] cloud-init[1750]: Downloaded unicode-ident v1.0.24 [ 103.193734] cloud-init[1767]: Downloaded socket2 v0.6.3 total 108 |
[CI] Stage: DeployInfrastructure ready.
|
[CI] Stage: SummaryAll tests PASSED. ARP seeding: kernel /proc/net/arp (automatic)
|
[Perf] Stage: DeployDeploying |
✅ Integration Tests Passed (Run 24240693445)Branch: Test Results
Application Logs (last 20 lines)receiver-echo-server.log sender-echo-server.log sender-test-client.log receiver-test-client-iperf.log sender-test-client-iperf.log Full Application Logs (last 200 lines each)receiver-echo-server.logsender-echo-server.logsender-test-client.logreceiver-test-client-iperf.logsender-test-client-iperf.log
|
[Perf] Stage: Instances Ready
|
[Perf] Stage: TRex ConfigStarting TRex configuration (MAC discovery + NIC binding)... |
[Perf] Stage: TRex Config OK
|
[Perf] Stage: TRex StartedTRex server running. Beginning benchmarks... |
[Perf] DUT ReadyDUT instance |
[Perf] Stage: Benchmark (1/3)Running |
[Perf] Benchmark Diag:
|
[Perf] Benchmark Diag:
|
[Perf] Stage: Benchmark (2/3)Running |
[Perf] Benchmark Diag:
|
[Perf] Benchmark Diag:
|
[Perf] Stage: Benchmark (3/3)Running |
[Perf] Benchmark Diag:
|
[Perf] Benchmark Diag:
|
[Perf] Diag: testpmd logtestpmd output (last 30 lines) |
[Perf] Stage: Results[12:34:49] INFO Generating markdown summary... Performance Test Results — unknownCommit: 64B packets
512B packets
1400B packets
8500B packets
|
8500B jumbo frames on c6in.xlarge: rust-dpdk delivers 125K PPS at 8.5 Gbps with 0.00% drop, matching native-dpdk (testpmd). Jumbo frames achieve higher sustained bandwidth than 1400B with 6x fewer packets. Bandwidth ceiling is ENA single-flow limit, not our stack. Also documents ENA Express MTU <= 8900 incompatibility finding. https://claude.ai/code/session_01294ECCWxYiENcnK5HVVyoL
[CI] Stage: DeployInfrastructure ready.
|
[CI] Stage: SummaryAll tests PASSED. ARP seeding: kernel /proc/net/arp (automatic)
|
✅ Integration Tests Passed (Run 24244488181)Branch: Test Results
Application Logs (last 20 lines)receiver-echo-server.log sender-echo-server.log sender-test-client.log receiver-test-client-iperf.log sender-test-client-iperf.log Full Application Logs (last 200 lines each)receiver-echo-server.logsender-echo-server.logsender-test-client.logreceiver-test-client-iperf.logsender-test-client-iperf.log
|
gspivey
pushed a commit
that referenced
this pull request
Jun 18, 2026
gspivey
added a commit
that referenced
this pull request
Jun 19, 2026
## ROADMAP Item #31 Implements `dpdk-stdlib-tcp: DUT test binaries and synthetic benchmark` from ROADMAP.md. **Spec:** `.kiro/specs/tcp-support/` tasks 12.1, 12.2, 13.1, 13.2, 13.3 ### Changes 1. **apps/tcp-echo** — Sync TCP echo server with `--ip`/`--port` flags and graceful SIGTERM/SIGINT shutdown 2. **apps/tcp-test-client** — TCP client with 4 test modes: `handshake` (connect-then-close), `bidir` (bidirectional echo verification), `shutdown` (graceful FIN teardown), `std-parity` (byte-for-byte comparison vs std::net::TcpStream) 3. **apps/tokio-tcp-echo** — Async TCP echo server via dpdk-tokio compat layer 4. **apps/tcp-synthetic-bench** — Mock PacketBackend benchmark measuring connection establishment latency, single-stream throughput, and engine tick processing time. Outputs markdown (stdout) + JSON (stderr) ### CI Workflow (requires manual apply) The `tcp-synthetic-perf` job for `integration-tests.yml` is prepared but cannot be pushed due to PAT `workflow` scope limitation. The job definition: - Runs on `ubuntu-latest` (no AWS needed) - Builds and runs `tcp-synthetic-bench` in release mode - Posts results as PR comment with commit hash and run link - Uploads results artifact with 30-day retention The workflow change is included as a local diff and should be applied by a human maintainer or token with `workflow` scope. ### Tests - `cargo build` — all workspace crates compile cleanly - `cargo test` — all existing tests pass (0 failures) ### Tradeoffs - `tcp-synthetic-bench` drives the TcpEngine directly (no real TCP connections) to measure pure framework overhead - `MockBackend` struct is defined for spec compliance but the benchmark uses engine-direct approach for deterministic measurements - `try_clone()` in tcp-test-client's std-parity mode works because the std fallback path is used for the kernel comparison --------- Co-authored-by: Agent Router <agent@gspivey.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
dut_bind_kernel()--max-pkt-len=9100 --mbuf-size=10240to testpmd for jumbo frame support--mbuf-factor 8to TRex startup for large packet memory allocationMotivation
With 1400B packets we're hitting 6.5-7 Gbps. Jumbo frames (8500B) should dramatically increase bandwidth by amortizing per-packet overhead over ~6x more payload. Theoretical max on c5n.2xlarge at 25 Gbps = ~367K PPS at 8500B.
Test plan
cargo build && cargo test— all 252 tests passhttps://claude.ai/code/session_01294ECCWxYiENcnK5HVVyoL