The August 2025 Patch Tuesday cumulative for Windows 11 (KB5063878) has introduced an unexpected and painful wrinkle for live broadcasters: NDI-based streaming through OBS and other production tools can stutter, lag, and drop frames — even on healthy networks — and Microsoft has acknowledged the issue while recommending an immediate, manual change to NDI’s transport mode as a workaround. (neowin.net) (support.microsoft.com)
NDI (Network Device Interface) is a widely used, high‑performance video-over-IP protocol adopted across live production, remote contribution, and content creation. It enables multiple PCs and devices to exchange low‑latency, high‑quality video and audio using commodity Ethernet networks. Production toolchains commonly combine NDI with Open Broadcaster Software (OBS), vMix, and the official NDI Tools suite for capture, routing, and monitoring.
In mid‑August 2025, Microsoft shipped KB5063878 as the cumulative quality and security update for Windows 11 version 24H2 (OS Build 26100.4946). The update also coincided with several other regressions reported by community and vendor channels — from recovery and reset workflow problems to storage regressions affecting certain NVMe controllers — and the NDI stuttering issue emerged as a particularly visible pain point for streamers and broadcast operators.
Microsoft’s guidance identifies the failing interaction as occurring when NDI is configured to use its default Receive Mode, which relies on the protocol known as Reliable UDP (RUDP). The symptom set described by Microsoft includes “severe stuttering, lag, and choppy playback, even under low‑bandwidth conditions,” with the impact especially pronounced when using Display Capture in tools like OBS. The company has said it is actively investigating and has pointed users to a temporary change of NDI’s transport from RUDP to either TCP or legacy UDP. (neowin.net) (support.microsoft.com)
The precise root cause — whether a timing/regulator change in the network stack, a change in socket behavior, a driver interaction, or a new security/validation path interfering with RUDP’s reliability layer — is under investigation. Microsoft has not, at the time of writing, published a line‑by‑line technical root‑cause document; the company has listed the symptom and an interim mitigation and is working toward a permanent fix. (neowin.net) (support.microsoft.com)
At the same time, separate but contemporaneous reports about NVMe drives disappearing under heavy writes after KB5063878 circulated among enthusiasts and affected users; those reports prompted vendor and Microsoft investigations in parallel. They should be treated as related only by coincidence in the update window unless vendors confirm a single causal change. Until then, treat storage claims as hardware‑specific and validate against vendor advisories and Microsoft notices.
Source: Windows Report Windows 11 KB5063878 Update Reportedly Breaks OBS NDI Streaming
NDI (Network Device Interface) is a widely used, high‑performance video-over-IP protocol adopted across live production, remote contribution, and content creation. It enables multiple PCs and devices to exchange low‑latency, high‑quality video and audio using commodity Ethernet networks. Production toolchains commonly combine NDI with Open Broadcaster Software (OBS), vMix, and the official NDI Tools suite for capture, routing, and monitoring.In mid‑August 2025, Microsoft shipped KB5063878 as the cumulative quality and security update for Windows 11 version 24H2 (OS Build 26100.4946). The update also coincided with several other regressions reported by community and vendor channels — from recovery and reset workflow problems to storage regressions affecting certain NVMe controllers — and the NDI stuttering issue emerged as a particularly visible pain point for streamers and broadcast operators.
Microsoft’s guidance identifies the failing interaction as occurring when NDI is configured to use its default Receive Mode, which relies on the protocol known as Reliable UDP (RUDP). The symptom set described by Microsoft includes “severe stuttering, lag, and choppy playback, even under low‑bandwidth conditions,” with the impact especially pronounced when using Display Capture in tools like OBS. The company has said it is actively investigating and has pointed users to a temporary change of NDI’s transport from RUDP to either TCP or legacy UDP. (neowin.net) (support.microsoft.com)
Why this matters to creators and production teams
- NDI is commonly used for multi‑PC workflows: capture machines, encoder boxes, graphics machines, and monitoring stations all stream sources over the LAN. Interruptions to NDI streams can immediately derail a live show.
- Many streamers and small production teams rely on Display Capture for game capture, slides, and app windows; this configuration appears to amplify the regression’s impact.
- Unlike an isolated app crash, transport‑level stuttering is unpredictable and hard to remediate on the fly; it breaks lip sync, ruins multi‑camera switching, and can force aborting a live transmission.
Technical explanation: RUDP, NDI, and where the regression appears
How NDI typically transports media
NDI supports multiple transport modes to balance latency, reliability, and network behavior:- RUDP (Reliable UDP): NDI’s default in many configurations; attempts to combine UDP’s low latency with an application‑layer reliability mechanism that retransmits lost packets, avoids head‑of‑line blocking, and keeps latency low where possible.
- UDP (Legacy): Simple, unacknowledged datagrams; low overhead and low latency but no retransmission or reliability.
- TCP (Single TCP): Standard reliable, ordered byte stream; higher latency and potential head‑of‑line blocking, but robust across networks with middleboxes that mishandle UDP variants.
What appears to have changed in KB5063878
Microsoft’s advisory (reported publicly and mirrored by other outlets) points to an interaction between Windows 11 24H2 post‑KB5063878 and NDI’s RUDP receive mode. The operating system change appears to disrupt RUDP’s expected behavior, producing consistent packet handling delays or retransmission storms that manifest as stutter and choppy playback even when bandwidth is plentiful.The precise root cause — whether a timing/regulator change in the network stack, a change in socket behavior, a driver interaction, or a new security/validation path interfering with RUDP’s reliability layer — is under investigation. Microsoft has not, at the time of writing, published a line‑by‑line technical root‑cause document; the company has listed the symptom and an interim mitigation and is working toward a permanent fix. (neowin.net) (support.microsoft.com)
Verified details and cross‑references
Several independent sources corroborate the same essential claim and mitigation path:- Microsoft’s KB/support entry for the August 12, 2025 cumulative update (KB5063878) shows the release metadata and the OS build involved; it also links to the Windows release health dashboard for ongoing notes. The support page documents the update’s scope and confirms Microsoft’s awareness of subsequent issues tied to the rollup. (support.microsoft.com)
- Reporting by a major Windows news outlet documented community reports (including an early forum post flagged by a community member) and Microsoft’s confirmation that the issue is tied to the NDI default receive mode (RUDP). That report includes the exact workaround steps recommended (change NDI Receive Mode via NDI Tools). (neowin.net)
- The official NDI documentation for Access Manager describes the Advanced receive options and confirms that NDI client applications read their configuration from the shared config file and that a restart of the receiving application is necessary for changes to take effect — supporting the recommended workaround process. (docs.ndi.video)
- Community threads and vendor forums independently reproduce the phenomenon: multiple users report that uninstalling the KB or switching NDI transport resolves the problem in their setups, and vendors (vMix, NDI tool users) have posted advisories in parallel. Those community repros are consistent in symptomology and remediation, reinforcing the diagnosis while also flagging the risk of the update to storage and recovery flows in some environments. (forum.derivative.ca)
The practical workaround (step‑by‑step)
Microsoft and NDI recommend changing the NDI Receive Mode from RUDP to Single TCP or UDP (Legacy) as an immediate mitigation. The procedure below is drawn from the official NDI Tools guidance and Microsoft’s advisory:- Download and install the free NDI Tools package from NDI’s distribution.
- Run NDI Access Manager on the receiving machine.
- Select the Advanced tab.
- Under Receive Mode, change the setting from RUDP (the default) to either Single TCP or UDP (Legacy).
- Press OK to save.
- Restart any NDI‑receiving applications (for example, OBS, Studio Monitor, vMix capture) so they re‑read the configuration.
- Repeat steps 2–6 on every machine participating in the NDI flow where RUDP had been used.
- Single TCP is likely the safest single‑change mitigation on difficult networks and across network equipment that treats nonstandard UDP differently, but it can increase latency and lead to head‑of‑line blocking if packet loss occurs. UDP (Legacy) keeps latency low but restores no reliability guarantees; it may be preferable in low‑loss, low‑latency LANs. Pick the tradeoff appropriate for the production environment and test thoroughly. (docs.ndi.video) (neowin.net)
- After changing the setting, restart OBS/NDI clients to force a config refresh — NDI accesses a shared config file at %programdata%/NDI/ndi-config.v1.json and reads it on application start. The NDI doc explicitly notes that running applications don't automatically pick up changes. (docs.ndi.video)
- If the workaround is not feasible in a time‑critical show (for example, remote contributors that can't be reconfigured quickly), temporarily removing the KB may restore previous behavior; however, removing security updates carries risk and is operationally disruptive. Use rollback as a last resort and follow organization policy for update management. (learn.microsoft.com)
Risk assessment and tradeoffs
- Security vs. Availability: Uninstalling a security rollup to restore NDI behavior is an option with real tradeoffs. Security updates often close vulnerabilities; rolling them back increases exposure. For production facilities, weigh the risk of being offline or producing broken live content against the potential security implications of a temporary rollback.
- Latency and reliability: Switching to TCP may eliminate stutter but can increase end‑to‑end latency and create head‑of‑line blocking during packet loss. For low‑latency competitive gaming streams or tightly synchronized multi‑camera sets, this is a measurable change. Test both settings before a critical broadcast.
- Scale of remediation: Single‑PC hobby setups can reconfigure NDI Tools quickly. Enterprise or distributed production teams must coordinate across many machines and subnets; orchestrating a safe, authenticated roll‑out is nontrivial and requires change control.
- Underlying regressions: The presence of multiple concurrent regressions in the same cumulative window (NDI transport behavior, storage regressions, recovery failures) suggests systemic testing gaps in the rollout pipeline. Organizations should treat August 2025 updates as high‑risk and validate in pilot rings before wide deployment.
Recommended operational checklist for streamers and engineers
- Short checklist to vet machines before a live show:
- Confirm OS build and whether KB5063878 is installed (winver /systeminfo).
- If KB5063878 is installed, change NDI Receive Mode to Single TCP or UDP and restart NDI clients. (docs.ndi.video)
- Perform an end‑to‑end check of NDI sources in the exact capture mode used for the show (Display Capture in OBS, webcam, game capture, etc.).
- Run a 10–15 minute sustained playback test and note any stutter/dropouts.
- If the team uses remote contributors, publish configuration guidance and version checks and coordinate a simultaneous switch window.
- For mission‑critical production, stage the update rollback/mitigation in a maintenance window and document the decision rationale.
- For administrators managing many devices:
- Pause automatic approvals for KB5063878 in WSUS/Windows Update for Business test rings until the fix is confirmed.
- Pilot the NDI Receive Mode change on a small subset of affected hosts and evaluate quality impact.
- If rollback is necessary, document compensating security controls and pursue the Microsoft Known Issue Rollback (KIR) or future patch as soon as it’s available. (neowin.net)
What vendors and communities report
NDI tooling vendors and independent producers mirrored the issue quickly. Community forums, broadcast engineering subreddits, and vendor pages show clusters of reproducible reports consistent with the Microsoft advisory: changing the transport resolves the symptom in a majority of cases, and uninstalling the KB often restores prior behavior where reconfiguring NDI is impractical. Vendors such as vMix and community threads have posted direct confirmation of the problem and recommended the same mitigations; those independent vendor confirmations strengthen confidence that the fault is a Windows/OS-level regression rather than an NDI or OBS code defect. (forum.derivative.ca)At the same time, separate but contemporaneous reports about NVMe drives disappearing under heavy writes after KB5063878 circulated among enthusiasts and affected users; those reports prompted vendor and Microsoft investigations in parallel. They should be treated as related only by coincidence in the update window unless vendors confirm a single causal change. Until then, treat storage claims as hardware‑specific and validate against vendor advisories and Microsoft notices.
How long until a permanent fix?
Microsoft has acknowledged the issue and is investigating. Historically, Microsoft’s cadence for out‑of‑band fixes or Known Issue Rollbacks (KIR) ranges from days to a few weeks depending on severity, repro rate, and whether a simple KIR or a full cumulative patch is required. For organizations that require a formal rollback approach, Microsoft has used KIR deployments via Group Policy or Intune in previous incidents; watch the Windows release health dashboard and official KB updates for a published KIR or a new cumulative that explicitly lists the fix. (neowin.net)Final recommendations (concise)
- For single creators and small teams: change NDI Receive Mode to Single TCP or UDP (Legacy) and restart your apps. Test thoroughly before going live. (docs.ndi.video)
- For productions with strict latency constraints: test both transports; consider localized reconfiguration (e.g., dedicated gigabit switch, direct 1:1 NIC connections) before rolling back security updates. Community reports show network hardware/NIC differences can influence NDI behavior, so hardware alternatives can help. (forum.derivative.ca)
- For IT admins: hold KB5063878 in non‑critical rings, pilot the change, and prepare rollback or KIR deployment playbooks. Coordinate with security teams about compensating controls if rollback is necessary.
- Avoid uninstalling security updates unless absolutely necessary and approved by risk and security governance; switching NDI transport is the recommended first line of defense. (neowin.net)
Conclusion
The KB5063878 incident underlines the fragile coupling between operating system network behavior and specialized, performance‑sensitive protocols like NDI. For broadcasters and creators, the immediate practical takeaway is straightforward: change NDI’s Receive Mode away from RUDP to restore reliable playback and avoid uninstalling security updates unless other mitigations fail and governance allows it. Behind that simple mitigation, the episode raises broader questions about update validation for niche, high‑value workflows and the need for rapid vendor collaboration when regressions affect critical live services. Microsoft and the NDI ecosystem have provided clear interim guidance; the remaining task is a durable fix from Microsoft and careful, documented testing by every production team before re‑introducing the affected cumulative into live environments. (neowin.net) (docs.ndi.video)Source: Windows Report Windows 11 KB5063878 Update Reportedly Breaks OBS NDI Streaming