quicX Interoperability Status

This document records the latest baseline interoperability results of

quicX against the major QUIC implementations and is the canonical version

to cite externally.

For test methodology, commands, and environment, see

guide/interop_runbook.md.

Item	Value
Run mode	upstream quic-interop-runner full matrix (quic-network-simulator / ns-3 topology)
Peers under test	17 third-party implementations (14 bidirectional + nginx / haproxy server-only + chrome client-only)
Scenarios under test	24 (22 protocol-conformance + 2 performance measurements)
Total cases	744 (PASS 588 / FAIL 61 / UNSUPPORTED 95)
Overall effective pass rate	90.60% (588 / 649, excluding UNSUPPORTED)
RFC conformance pass rate	90.15% (531 / 589; client mode 292/312 = 93.59%, server mode 239/277 = 86.28%)
Performance pass rate	95.00% (57 / 60)

TL;DR

✅ Across the full 24-scenario × 17-peer matrix (744 cases), quicX achieves an overall effective pass rate of 90.60%.
✅ 12 scenarios at 100%: handshake, transfer, longrtt, chacha20, retry, http3, blackhole, ecn, transferloss, transfercorruption, ipv6, plus goodput (30/30, avg 8.67 Mbps, peak bandwidth utilization 94.7%).
✅ crosstraffic (vs TCP Cubic) 27/30 (90.0%), balanced in both directions; only kwik / mvfst (client) and xquic (server) miss the threshold.
✅ Versus the previous round (91.22%): quicx → mvfst retry, quicx → quic-go http3, quicx → picoquic blackhole, and quicx → mvfst transfercorruption recovered to PASS; client-mode zerortt and server-mode handshakeloss / handshakecorruption improved markedly.
⚠️ Current weak spots: connectionmigration (31.6%), rebind-addr (60.0%), rebind-port / amplificationlimit (66.7% each — amplificationlimit server mode fell to 6/14, the main regression of this round), and server-mode keyupdate (8/10).

1. Overview

Source: one full-matrix run of the upstream quic-interop-runner.
Environment: Linux host, Docker + ns-3 network simulation (leftnet ↔ sim ↔ rightnet); quicX runs from locally built interop_{server,client} binaries.
Coverage: 24 scenarios × (16 server-side peers + 15 client-side peers) = 744 combinations; no quicX ↔ quicX self-test.
Delta vs. previous baseline: the previous round used the same 24-scenario × 17-peer caliber (744 cases, 91.22% effective), so the two rounds are directly comparable. The overall rate dropped 0.62pp, driven mainly by server-mode amplificationlimit (14/14 → 6/14), keyupdate (9/10 → 8/10), and newly-failing xquic server multiplexing; meanwhile retry (25/26 → 26/26), http3 (26/27 → 27/27), blackhole (29/30 → 30/30), and transfercorruption (29/30 → 30/30) recovered to full marks, and zerortt (23/26 → 24/26) and handshakecorruption (24/28 → 26/28) improved. The upstream matrix does not include multiconnect / versionnegotiation.

2. Legend

Mark	Meaning
✅	SUCCEEDED
❌	FAILED (test failure / throughput below threshold)
➖	UNSUPPORTED — peer or scenario unsupported
—	N/A — combination not run (peer does not ship that role)

3. Overall Results

Metric	Value
Total cases	744
✅ Passed	588
❌ Failed	61
➖ Unsupported	95
Effective pass rate (excl. Unsupported)	588 / 649 ≈ 90.60%
RFC conformance pass rate	531 / 589 ≈ 90.15% (client 292/312 = 93.59%; server 239/277 = 86.28%)
Performance pass rate	57 / 60 ≈ 95.00% (goodput 30/30; crosstraffic 27/30)
Pass rate incl. Unsupported	588 / 744 ≈ 79.0%

4. Per-scenario pass rates

Test scenario	Short name	Client-mode pass rate	Server-mode pass rate	Combined effective pass rate	Description
handshake	`H`	16/16 (100.0%)	14/14 (100.0%)	30/30 (100.0%)	Handshake completes successfully.
transfer	`DC`	16/16 (100.0%)	14/14 (100.0%)	30/30 (100.0%)	Stream data is being sent and received correctly. Connection close completes with a zero error code.
longrtt	`LR`	16/16 (100.0%)	14/14 (100.0%)	30/30 (100.0%)	Handshake completes when RTT is long.
chacha20	`C20`	14/14 (100.0%)	10/10 (100.0%)	24/24 (100.0%)	Handshake completes using ChaCha20.
multiplexing	`M`	16/16 (100.0%)	13/14 (92.9%)	29/30 (96.7%)	Thousands of files are transferred over a single connection, and server increased stream limits to accomodate client requests.
retry	`S`	14/14 (100.0%)	12/12 (100.0%)	26/26 (100.0%)	Server sends a Retry, and a subsequent connection using the Retry token completes successfully.
resumption	`R`	14/15 (93.3%)	12/12 (100.0%)	26/27 (96.3%)	Connection is established using TLS Session Resumption.
zerortt	`Z`	14/14 (100.0%)	10/12 (83.3%)	24/26 (92.3%)	0-RTT data is being sent and acted on.
http3	`3`	14/14 (100.0%)	13/13 (100.0%)	27/27 (100.0%)	An H3 transaction succeeded.
blackhole	`B`	16/16 (100.0%)	14/14 (100.0%)	30/30 (100.0%)	Transfer succeeds despite underlying network blacking out for a few seconds.
keyupdate	`U`	16/16 (100.0%)	8/10 (80.0%)	24/26 (92.3%)	One of the two endpoints updates keys and the peer responds correctly.
ecn	`E`	6/6 (100.0%)	6/6 (100.0%)	12/12 (100.0%)	Handshake completes successfully.
amplificationlimit	`A`	14/16 (87.5%)	6/14 (42.9%)	20/30 (66.7%)	The server obeys the 3x amplification limit.
handshakeloss	`L1`	14/15 (93.3%)	10/13 (76.9%)	24/28 (85.7%)	Handshake completes under extreme packet loss.
transferloss	`L2`	16/16 (100.0%)	14/14 (100.0%)	30/30 (100.0%)	Transfer completes under moderate packet loss.
handshakecorruption	`C1`	15/15 (100.0%)	11/13 (84.6%)	26/28 (92.9%)	Handshake completes under extreme packet corruption.
transfercorruption	`C2`	16/16 (100.0%)	14/14 (100.0%)	30/30 (100.0%)	Transfer completes under moderate packet corruption.
ipv6	`6`	16/16 (100.0%)	14/14 (100.0%)	30/30 (100.0%)	A transfer across an IPv6-only network succeeded.
v2	`V2`	8/8 (100.0%)	7/8 (87.5%)	15/16 (93.8%)	Server should select QUIC v2 in compatible version negotiation.
rebind-port	`BP`	10/16 (62.5%)	10/14 (71.4%)	20/30 (66.7%)	Transfer completes under frequent port rebindings on the client side.
rebind-addr	`BA`	9/16 (56.2%)	10/14 (71.4%)	19/30 (63.3%)	Transfer completes under frequent IP address and port rebindings on the client side.
connectionmigration	`CM`	3/5 (60.0%)	3/14 (21.4%)	6/19 (31.6%)	A transfer succeeded during which the client performed an active migration.
goodput	`G`	16/16 (100.0%)	14/14 (100.0%)	30/30 (100.0%)	Measures connection goodput over a 10Mbps link.
crosstraffic	`C`	14/16 (87.5%)	13/14 (92.9%)	27/30 (90.0%)	Measures goodput over a 10Mbps link when competing with a TCP (cubic) connection.

5. Peer ranking

Rank	Implementation	Client mode (initiator)	Server mode (responder)	Protocol-feature pass rate	Combined effective pass rate	Status
1	aioquic	22/22 (100.0%)	22/23 (95.7%)	40/41 (97.6%)	44/45 (97.8%)	🌟 Excellent
2	picoquic	24/24 (100.0%)	22/24 (91.7%)	42/44 (95.5%)	46/48 (95.8%)	🌟 Excellent
3	lsquic	23/24 (95.8%)	22/23 (95.7%)	41/43 (95.3%)	45/47 (95.7%)	🌟 Excellent
4	ngtcp2	21/24 (87.5%)	24/24 (100.0%)	41/44 (93.2%)	45/48 (93.8%)	🌟 Excellent
5	kwik	19/22 (86.4%)	22/23 (95.7%)	38/41 (92.7%)	41/45 (91.1%)	🌟 Excellent
6	neqo	24/24 (100.0%)	20/24 (83.3%)	40/44 (90.9%)	44/48 (91.7%)	🌟 Excellent
7	quinn	22/22 (100.0%)	20/24 (83.3%)	38/42 (90.5%)	42/46 (91.3%)	🌟 Excellent
8	haproxy	20/22 (90.9%)	N/A	18/20 (90.0%)	20/22 (90.9%)	🌟 Excellent
9	quic-go	21/21 (100.0%)	18/22 (81.8%)	35/39 (89.7%)	39/43 (90.7%)	👍 Good
10	msquic	21/21 (100.0%)	18/22 (81.8%)	35/39 (89.7%)	39/43 (90.7%)	👍 Good
11	s2n-quic	20/22 (90.9%)	18/20 (90.0%)	34/38 (89.5%)	38/42 (90.5%)	👍 Good
12	nginx	19/21 (90.5%)	N/A	17/19 (89.5%)	19/21 (90.5%)	👍 Good
13	quiche	19/21 (90.5%)	18/20 (90.0%)	33/37 (89.2%)	37/41 (90.2%)	👍 Good
14	xquic	19/21 (90.5%)	17/22 (77.3%)	33/39 (84.6%)	36/43 (83.7%)	👍 Good
15	go-x-net	12/14 (85.7%)	13/15 (86.7%)	21/25 (84.0%)	25/29 (86.2%)	👍 Good
16	mvfst	16/19 (84.2%)	11/18 (61.1%)	24/33 (72.7%)	27/37 (73.0%)	⚠️ Needs improvement
17	chrome	N/A	1/1 (100.0%)	1/1 (100.0%)	1/1 (100.0%)	🌟 Excellent

Note: “Client mode” means quicX acting as client against the peer’s server; “Server mode” means the peer acting as client against the quicX server. The conformance rate excludes the goodput / crosstraffic performance scenarios.

6. Detailed matrices by scenario

📋 Detailed per-scenario matrix tables

🔹 handshake (`H`)

Note: Handshake completes successfully.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—	—	✅	➖

🔹 transfer (`DC`)

Note: Stream data is being sent and received correctly. Connection close completes with a zero error code.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—	—	✅	➖

🔹 longrtt (`LR`)

Note: Handshake completes when RTT is long.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—	—	✅	➖

🔹 chacha20 (`C20`)

Note: Handshake completes using ChaCha20.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	✅	✅	✅	✅	✅	✅	➖	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	➖	✅	➖	➖	✅	✅	✅	➖	✅	✅	✅	✅	—	—	✅	➖

🔹 multiplexing (`M`)

Note: Thousands of files are transferred over a single connection, and server increased stream limits to accomodate client requests.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	❌	—	—	✅	➖

🔹 retry (`S`)

Note: Server sends a Retry, and a subsequent connection using the Retry token completes successfully.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	✅	✅	✅	✅	✅	✅	➖	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	➖	✅	✅	✅	✅	✅	✅	➖	✅	✅	✅	✅	—	—	✅	➖

🔹 resumption (`R`)

Note: Connection is established using TLS Session Resumption.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	✅	✅	✅	✅	❌	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	➖	✅	✅	✅	✅	✅	✅	➖	✅	✅	✅	✅	—	—	✅	➖

🔹 zerortt (`Z`)

Note: 0-RTT data is being sent and acted on.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	✅	➖	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	➖	✅	➖	✅	✅	✅	✅	❌	✅	❌	✅	✅	—	—	✅	➖

🔹 http3 (`3`)

Note: An H3 transaction succeeded.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	✅	✅	✅	✅	✅	✅	✅	✅	➖	✅	✅	✅	✅	✅	—
quicX as Server	✅	➖	✅	✅	✅	✅	✅	✅	✅	✅	➖	✅	✅	—	—	✅	✅

🔹 blackhole (`B`)

Note: Transfer succeeds despite underlying network blacking out for a few seconds.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—	—	✅	➖

🔹 keyupdate (`U`)

Note: One of the two endpoints updates keys and the peer responds correctly.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	➖	➖	✅	➖	✅	➖	✅	✅	❌	✅	—	—	❌	➖

🔹 ecn (`E`)

Note: Handshake completes successfully.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	➖	✅	➖	✅	✅	➖	➖	➖	➖	✅	➖	➖	➖	✅	—
quicX as Server	✅	➖	➖	✅	➖	✅	✅	➖	➖	➖	➖	✅	➖	—	—	✅	➖

🔹 amplificationlimit (`A`)

Note: The server obeys the 3x amplification limit.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	❌	✅	✅	✅	❌	✅	✅	✅	—
quicX as Server	✅	❌	❌	❌	✅	❌	❌	✅	❌	✅	✅	❌	✅	—	—	❌	➖

🔹 handshakeloss (`L1`)

Note: Handshake completes under extreme packet loss.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	❌	✅	✅	✅	—
quicX as Server	✅	➖	✅	✅	❌	✅	✅	✅	❌	✅	✅	✅	❌	—	—	✅	➖

🔹 transferloss (`L2`)

Note: Transfer completes under moderate packet loss.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—	—	✅	➖

🔹 handshakecorruption (`C1`)

Note: Handshake completes under extreme packet corruption.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	➖	✅	✅	✅	✅	✅	✅	❌	✅	✅	✅	❌	—	—	✅	➖

🔹 transfercorruption (`C2`)

Note: Transfer completes under moderate packet corruption.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—	—	✅	➖

🔹 ipv6 (`6`)

Note: A transfer across an IPv6-only network succeeded.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—
quicX as Server	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅	—	—	✅	➖

🔹 v2 (`V2`)

Note: Server should select QUIC v2 in compatible version negotiation.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅	➖	➖	➖	➖	✅	✅	✅	➖	✅	✅	✅	➖	➖	✅	➖	—
quicX as Server	✅	➖	➖	➖	➖	✅	✅	✅	➖	✅	✅	✅	➖	—	—	❌	➖

🔹 rebind-port (`BP`)

Note: Transfer completes under frequent port rebindings on the client side.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	❌	❌	✅	✅	❌	✅	✅	❌	✅	✅	✅	✅	✅	❌	❌	✅	—
quicX as Server	✅	✅	❌	✅	✅	❌	✅	✅	❌	✅	❌	✅	✅	—	—	✅	➖

🔹 rebind-addr (`BA`)

Note: Transfer completes under frequent IP address and port rebindings on the client side.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	❌	❌	✅	❌	❌	✅	✅	❌	❌	✅	✅	✅	✅	❌	❌	✅	—
quicX as Server	✅	✅	❌	✅	✅	❌	✅	✅	❌	✅	❌	✅	✅	—	—	✅	➖

🔹 connectionmigration (`CM`)

Note: A transfer succeeded during which the client performed an active migration.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	❌	➖	➖	❌	➖	✅	✅	➖	➖	➖	➖	✅	➖	➖	➖	➖	—
quicX as Server	✅	❌	❌	❌	❌	❌	✅	❌	❌	❌	❌	✅	❌	—	—	❌	➖

🔹 goodput (`G`)

Note: Measures connection goodput over a 10Mbps link.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅ _{9.30 Mbps}	✅ _{7.97 Mbps}	✅ _{9.43 Mbps}	✅ _{9.19 Mbps}	✅ _{9.35 Mbps}	✅ _{8.35 Mbps}	✅ _{9.47 Mbps}	✅ _{7.97 Mbps}	✅ _{9.26 Mbps}	✅ _{8.55 Mbps}	✅ _{8.54 Mbps}	✅ _{9.17 Mbps}	✅ _{9.26 Mbps}	✅ _{9.40 Mbps}	✅ _{8.97 Mbps}	✅ _{8.69 Mbps}	—
quicX as Server	✅ _{9.20 Mbps}	✅ _{9.38 Mbps}	✅ _{8.17 Mbps}	✅ _{8.77 Mbps}	✅ _{6.81 Mbps}	✅ _{8.70 Mbps}	✅ _{8.88 Mbps}	✅ _{7.11 Mbps}	✅ _{6.82 Mbps}	✅ _{9.20 Mbps}	✅ _{8.23 Mbps}	✅ _{8.59 Mbps}	✅ _{9.03 Mbps}	—	—	✅ _{8.46 Mbps}	➖

📊 Measured goodput leaderboard (10 Mbps link test, average: 8.67 Mbps)

Rank	Implementation	Role	Status	Rate detail (± jitter)	Measured goodput (Mbps)	Bandwidth utilization
1	lsquic	quicX as Client	✅	9470 (± 9) kbps	9.47 Mbps	`94.7%`
2	quic-go	quicX as Client	✅	9426 (± 1) kbps	9.43 Mbps	`94.3%`
3	nginx	quicX as Client	✅	9404 (± 8) kbps	9.40 Mbps	`94.0%`
4	go-x-net	quicX as Server	✅	9377 (± 17) kbps	9.38 Mbps	`93.8%`
5	quiche	quicX as Client	✅	9351 (± 19) kbps	9.35 Mbps	`93.5%`
6	ngtcp2	quicX as Client	✅	9299 (± 93) kbps	9.30 Mbps	`93.0%`
7	xquic	quicX as Client	✅	9263 (± 5) kbps	9.26 Mbps	`92.6%`
8	mvfst	quicX as Client	✅	9256 (± 45) kbps	9.26 Mbps	`92.6%`
9	aioquic	quicX as Server	✅	9204 (± 85) kbps	9.20 Mbps	`92.0%`
10	ngtcp2	quicX as Server	✅	9198 (± 71) kbps	9.20 Mbps	`92.0%`
11	s2n-quic	quicX as Client	✅	9194 (± 40) kbps	9.19 Mbps	`91.9%`
12	picoquic	quicX as Client	✅	9170 (± 61) kbps	9.17 Mbps	`91.7%`
13	xquic	quicX as Server	✅	9029 (± 272) kbps	9.03 Mbps	`90.3%`
14	haproxy	quicX as Client	✅	8970 (± 53) kbps	8.97 Mbps	`89.7%`
15	lsquic	quicX as Server	✅	8884 (± 134) kbps	8.88 Mbps	`88.8%`
16	s2n-quic	quicX as Server	✅	8774 (± 142) kbps	8.77 Mbps	`87.7%`
17	neqo	quicX as Server	✅	8695 (± 164) kbps	8.70 Mbps	`87.0%`
18	quinn	quicX as Client	✅	8691 (± 6) kbps	8.69 Mbps	`86.9%`
19	picoquic	quicX as Server	✅	8589 (± 196) kbps	8.59 Mbps	`85.9%`
20	aioquic	quicX as Client	✅	8554 (± 17) kbps	8.55 Mbps	`85.5%`
21	msquic	quicX as Client	✅	8539 (± 38) kbps	8.54 Mbps	`85.4%`
22	quinn	quicX as Server	✅	8463 (± 176) kbps	8.46 Mbps	`84.6%`
23	neqo	quicX as Client	✅	8353 (± 13) kbps	8.35 Mbps	`83.5%`
24	msquic	quicX as Server	✅	8235 (± 300) kbps	8.23 Mbps	`82.3%`
25	quic-go	quicX as Server	✅	8168 (± 473) kbps	8.17 Mbps	`81.7%`
26	go-x-net	quicX as Client	✅	7972 (± 15) kbps	7.97 Mbps	`79.7%`
27	kwik	quicX as Client	✅	7966 (± 1550) kbps	7.97 Mbps	`79.7%`
28	kwik	quicX as Server	✅	7108 (± 130) kbps	7.11 Mbps	`71.1%`
29	mvfst	quicX as Server	✅	6823 (± 351) kbps	6.82 Mbps	`68.2%`
30	quiche	quicX as Server	✅	6810 (± 128) kbps	6.81 Mbps	`68.1%`

💡 Performance analysis: quicX performs excellently on the 10 Mbps bottleneck link — all 30 measured scenarios in both directions (Client mode 16/16, Server mode 14/14) pass with a 100% score, the top bandwidth utilization reaches 94.7% (lsquic: 9.47 Mbps), and the global average effective rate is 8.67 Mbps.

🔹 crosstraffic (`C`)

Note: Measures goodput over a 10Mbps link when competing with a TCP (cubic) connection.

Role \ Peer	ngtcp2	go-x-net	quic-go	s2n-quic	quiche	neqo	lsquic	kwik	mvfst	aioquic	msquic	picoquic	xquic	nginx	haproxy	quinn	chrome
quicX as Client	✅ _{4.38 Mbps}	✅ _{2.69 Mbps}	✅ _{4.58 Mbps}	✅ _{5.08 Mbps}	✅ _{4.74 Mbps}	✅ _{4.03 Mbps}	✅ _{8.42 Mbps}	❌	❌	✅ _{3.13 Mbps}	✅ _{3.31 Mbps}	✅ _{7.26 Mbps}	✅ _{8.56 Mbps}	✅ _{2.42 Mbps}	✅ _{4.62 Mbps}	✅ _{3.12 Mbps}	—
quicX as Server	✅ _{9.06 Mbps}	✅ _{6.11 Mbps}	✅ _{8.41 Mbps}	✅ _{8.62 Mbps}	✅ _{6.71 Mbps}	✅ _{8.77 Mbps}	✅ _{8.77 Mbps}	✅ _{6.58 Mbps}	✅ _{6.64 Mbps}	✅ _{8.96 Mbps}	✅ _{8.55 Mbps}	✅ _{8.50 Mbps}	❌	—	—	✅ _{8.10 Mbps}	➖

📊 Detailed results of the TCP (Cubic) competition runs

Rank	Implementation	Role	Status	Rate detail (± jitter)	Measured goodput (Mbps)	Notes
1	ngtcp2	quicX as Server	✅	9058 (± 129) kbps	9.06 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `90.6%`)
2	aioquic	quicX as Server	✅	8960 (± 84) kbps	8.96 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `89.6%`)
3	lsquic	quicX as Server	✅	8770 (± 168) kbps	8.77 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `87.7%`)
4	neqo	quicX as Server	✅	8765 (± 268) kbps	8.77 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `87.6%`)
5	s2n-quic	quicX as Server	✅	8621 (± 68) kbps	8.62 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `86.2%`)
6	xquic	quicX as Client	✅	8564 (± 116) kbps	8.56 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `85.6%`)
7	msquic	quicX as Server	✅	8548 (± 273) kbps	8.55 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `85.5%`)
8	picoquic	quicX as Server	✅	8503 (± 137) kbps	8.50 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `85.0%`)
9	lsquic	quicX as Client	✅	8423 (± 125) kbps	8.42 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `84.2%`)
10	quic-go	quicX as Server	✅	8414 (± 97) kbps	8.41 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `84.1%`)
11	quinn	quicX as Server	✅	8097 (± 226) kbps	8.10 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `81.0%`)
12	picoquic	quicX as Client	✅	7256 (± 129) kbps	7.26 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `72.6%`)
13	quiche	quicX as Server	✅	6706 (± 310) kbps	6.71 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `67.1%`)
14	mvfst	quicX as Server	✅	6642 (± 163) kbps	6.64 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `66.4%`)
15	kwik	quicX as Server	✅	6578 (± 108) kbps	6.58 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `65.8%`)
16	go-x-net	quicX as Server	✅	6106 (± 107) kbps	6.11 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `61.1%`)
17	s2n-quic	quicX as Client	✅	5076 (± 152) kbps	5.08 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `50.8%`)
18	quiche	quicX as Client	✅	4737 (± 481) kbps	4.74 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `47.4%`)
19	haproxy	quicX as Client	✅	4620 (± 77) kbps	4.62 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `46.2%`)
20	quic-go	quicX as Client	✅	4582 (± 191) kbps	4.58 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `45.8%`)
21	ngtcp2	quicX as Client	✅	4378 (± 746) kbps	4.38 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `43.8%`)
22	neqo	quicX as Client	✅	4033 (± 87) kbps	4.03 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `40.3%`)
23	msquic	quicX as Client	✅	3315 (± 70) kbps	3.31 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `33.1%`)
24	aioquic	quicX as Client	✅	3133 (± 157) kbps	3.13 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `31.3%`)
25	quinn	quicX as Client	✅	3125 (± 123) kbps	3.12 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `31.2%`)
26	go-x-net	quicX as Client	✅	2694 (± 94) kbps	2.69 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `26.9%`)
27	nginx	quicX as Client	✅	2416 (± 99) kbps	2.42 Mbps	Fair bandwidth share established on the 10 Mbps competing link (utilization `24.2%`)
-	`kwik`, `mvfst`	quicX as Client	❌	—	—	Below the interop-runner threshold when competing against TCP Cubic
-	`xquic`	quicX as Server	❌	—	—	Below the interop-runner threshold when competing against TCP Cubic

💡 Competition analysis: quicX shows balanced, bidirectional resilience against TCP Cubic competition: as Server it secures 13/14 (92.9%) of the 14 peer clients with an average captured rate > 8.0 Mbps (above 9.0 Mbps against ngtcp2, and 8.96 Mbps against aioquic); as Client 14/16 (87.5%) pass, with the highest capture efficiency against xquic (8.56 Mbps) and lsquic (8.42 Mbps). The combined anti-TCP-Cubic pass rate reaches 90.0% (27/30); only the kwik / mvfst (Client) and xquic (Server) combinations fall short.

quicX Interoperability Status

TL;DR

1. Overview

2. Legend

3. Overall Results

4. Per-scenario pass rates

5. Peer ranking

6. Detailed matrices by scenario

📋 Detailed per-scenario matrix tables

🔹 handshake (H)

🔹 transfer (DC)

🔹 longrtt (LR)

🔹 chacha20 (C20)

🔹 multiplexing (M)

🔹 retry (S)

🔹 resumption (R)

🔹 zerortt (Z)

🔹 http3 (3)

🔹 blackhole (B)

🔹 keyupdate (U)

🔹 ecn (E)

🔹 amplificationlimit (A)

🔹 handshakeloss (L1)

🔹 transferloss (L2)

🔹 handshakecorruption (C1)

🔹 transfercorruption (C2)

🔹 ipv6 (6)

🔹 v2 (V2)

🔹 rebind-port (BP)

🔹 rebind-addr (BA)

🔹 connectionmigration (CM)

🔹 goodput (G)