Online sim racing punishes network latency more than almost any consumer game. iRacing’s match-pace tick rate is 60 Hz, Assetto Corsa Competizione runs server-side at 50 Hz, and any single packet round-trip above 35 ms shows up as a “warp” where another car snaps to a new position. Sustained latency under 25 ms with jitter under 3 ms is the realistic target, and most home networks are 60–120 ms away from it without specific tuning.
This guide covers the four network choices that actually move the latency number — physical link type, router/firewall QoS, Wi-Fi configuration, and ISP plan selection — in the order they should be addressed. For the sim-racer side of the equation, the sim racing internet setup, latency, and ISP choice reference covers the in-rig wiring and broadband shopping. This article goes upstream from the rig to the firewall to the WAN.
Wired Beats Wi-Fi by 3-5 ms (Always)
The single biggest gain comes from running Cat6 from the sim rig to the router. Even Wi-Fi 6E in optimal conditions adds 3–5 ms median latency and 8–15 ms p99 jitter compared to wired. In a real home with neighbors, microwaves, and Bluetooth, the gap widens to 10–25 ms p99.
If running cable is impossible, the second-best option is a MoCA 2.5 adapter pair through coaxial — 2.5 Gbps, 2 ms added latency, and works without re-cabling. Powerline adapters (HomePlug AV2 etc.) are the worst option for sim racing — 8–15 ms added latency, sensitive to noise from any other appliance on the circuit.
The shorthand: Cat6 first, MoCA second, mesh Wi-Fi third, powerline never.
QoS Configuration on the Firewall
Quality-of-Service rules prioritize the sim-racing traffic when the link saturates (which happens whenever someone in the house starts a 4K stream or a Steam download). Without QoS, your 25 ms latency jumps to 200+ ms the moment bandwidth is contested.
The principles that actually work:
- Cake or fq_codel queue discipline — modern algorithms that fight bufferbloat. Set on both upload and download. Most consumer routers don’t expose this; pfSense, OPNsense, OpenWRT, and Eero do.
- Set bandwidth caps 5–10% below your real link speed — measured at peak usage, not advertised. The cap is what gives the queue control.
- Tag sim-racing traffic to high priority by IP or DSCP — easier to set by source IP (the sim rig’s static IP) than by port (iRacing uses random UDP ports).
- Don’t use legacy “WMM” wireless QoS — it does little for the sim-racing workload and complicates other tuning.
For pfSense users, the limiter setup with a CoDel-cake child queue and a high-priority tag for the sim-rig IP is the standard build — see the pfSense cluster for the configuration walkthrough. OPNsense’s traffic shaper handles the same intent through a slightly different UI.
Wi-Fi 6 (and 6E) Configuration If You Must Go Wireless
If wired and MoCA are off the table, Wi-Fi 6 with strict configuration gets close to acceptable for sim racing. The configuration that matters:
- 5 GHz or 6 GHz only — disable 2.4 GHz on the sim rig’s network. The 2.4 GHz band is congested and adds 4–8 ms variability.
- 80 MHz channel width on 5 GHz, 160 MHz on 6 GHz — wider isn’t always better; in dense apartments, 40 MHz can be lower-jitter than 80.
- Pin a single AP for the sim rig — disable client steering and band steering for the rig’s MAC. Roaming events add 50–80 ms latency spikes.
- WPA3-only or WPA2-Personal — never WPA3-Transition mode, which causes intermittent reassociation.
- OFDMA enabled, MU-MIMO disabled — counterintuitive but MU-MIMO on consumer APs adds occasional jitter without sufficient throughput benefit.
Realistic Wi-Fi 6E latency is 4–8 ms median in a good RF environment, 12–20 ms in a bad one. Compare against Cat6’s flat 1 ms.
ISP Plan Selection (Most Don’t Matter, A Few Do)
For sim racing, advertised “speed” is mostly irrelevant — 50 Mbps down is plenty. What matters is upload speed, route quality, and whether the ISP throttles UDP. Specifics:
- Symmetric fiber (gigabit/gigabit, 1 Gbps/200 Mbps, etc.) — best for sim racing. Upload doesn’t get cramped on race-server packets.
- Cable (Comcast, Spectrum, etc.) — works fine for sim racing if the upload tier is 35 Mbps+. DOCSIS 3.1 has acceptable jitter; DOCSIS 3.0 in a busy node sometimes doesn’t.
- 5G fixed wireless (T-Mobile Home Internet, Verizon) — usable but variable. 30–80 ms latency and bursty jitter make it the second-worst sim racing option.
- Starlink — 25–40 ms latency, occasional satellite-handoff spikes to 100+ ms. Tolerable for casual leagues, unsuitable for tight pro multi-class.
- Satellite GEO (Hughes, Viasat) — unusable. 600+ ms latency.
Before committing to an ISP plan, run a 24-hour ping test (`ping -i 0.5` against the sim-server region’s edge) and look at p99 latency, not average. The average looks fine; the p99 is what causes warps.
Route Shopping (The Underrated Tool)
Two ISPs at the same price often have different upstream peering, which means different latency to the iRacing US-East server farm or whatever your league uses. Test before you commit.
The shopping workflow:
- Get a 30-day trial or a friend’s connection for 1 hour
- Run `mtr` (or WinMTR on Windows) to the sim server’s IP for 5 minutes
- Note hop count and per-hop p99 latency
- Compare against your current ISP — fewer hops and lower p99 = better route
Sometimes a cheaper “lesser” plan on a better-peered ISP is faster than a premium plan on a worse-peered one. The numbers on the speed-test ad never tell this story.
What to Do on the Rig Side
Once the upstream is clean, three rig-side settings matter:
- Disable Windows Game Mode if it interferes with NIC scheduling — controversial, test both ways
- Set the NIC to 1 Gbps full-duplex explicitly — auto-negotiate sometimes settles on 100 Mbps after a switch reboot
- Update NIC drivers — Realtek drivers from 2021 have a known latency regression; Intel I225-V firmware before v2.5 has bufferbloat
For the deeper rig hardware that wraps around all this — UPS, NIC selection, cable runs — pair with the homelabrouter hardware cluster.
When the Network Is Already Good Enough
A useful sanity check before any tuning: run a 30-minute online race and watch the in-game latency display. If it shows 18–25 ms with no warps, the network is already good. Tuning further has diminishing returns and won’t measurably improve race outcomes. Spend the time on car setups instead.
Tune when the display shows 35+ ms or sustained jitter spikes during contested bandwidth (someone streaming, someone downloading). That’s where QoS and wired upgrades earn their cost.
Frequently Asked Questions
What latency do I need for online sim racing?
Sustained latency under 25 milliseconds and jitter under 3 milliseconds is the practical target for clean iRacing or Assetto Corsa Competizione racing. Above 35 milliseconds you start seeing other cars warp. Above 60 milliseconds the experience becomes unplayable in close-pack racing.
Will Wi-Fi 6 work for sim racing?
Acceptably in a good RF environment. Wi-Fi 6 adds 3 to 5 milliseconds median latency and 8 to 15 milliseconds jitter compared with Cat6. In dense apartments, the gap is wider. Wired ethernet is always the better choice when physically possible.
Does QoS actually help in real games?
Yes when bandwidth is contested. Without QoS, a 4K Netflix stream or a Steam download will spike latency to 200 milliseconds. With Cake or fq_codel queue discipline and the sim rig tagged high priority, latency stays at 25 to 30 milliseconds even during contention.
Is gigabit fiber overkill for sim racing?
For latency, no, because gigabit fiber typically has the best route peering and lowest jitter regardless of bandwidth headroom. For bandwidth, yes, because sim racing uses 1 to 3 megabits per second of actual race traffic. The fiber benefits come from peering quality.
Can a VPN improve sim racing latency?
Almost never. VPNs add encryption overhead and an extra hop, both of which increase latency. The only exception is when your ISP routes through a poorly peered transit; then a VPN that egresses on a better network can reduce hop count to the sim server. Test before assuming.
What is bufferbloat and why does it matter?
Bufferbloat is when network buffers grow large enough during congestion that latency spikes from 25 milliseconds to 500 milliseconds in seconds. Sim racing is unplayable through bufferbloat. Cake or fq_codel queue discipline on the firewall fights it by capping the buffer.
Should I run a separate router for the sim rig?
Only if your existing router cannot run Cake or fq_codel. A router running OPNsense, pfSense, or OpenWRT handles QoS for the entire household and a separate sim-rig router is unnecessary. If your ISP-supplied router cannot run modern queue discipline, replace it rather than chain a second one.