[PC/Steam] Recurring 15-second FPS drops with ~315-second page-fault bursts
Hello Apex/Respawn support,
Apex previously ran well near my 240 FPS cap. Recently it repeatedly slows for approximately 10–15 seconds, often into the 180–200 FPS range, then recovers. The exact first day symptoms began is uncertain. Please investigate and advise on a supported remedy.
It also happens while spectating teammates, or spectating enemies after my entire squad dies: the in-game FPS counter still drops to around 190–200. The issue is intermittent across sessions: I had a whole session without noticeable drops, then frequent episodes the next day.
Platform: PC / Steam, default public branch; Windows 11 Home build 26200.
Hardware: Intel Core Ultra 7 265K; ASUS ROG STRIX Z890-E GAMING WIFI; RTX 5070; 64 GB (2x32 GB) DDR5-5200; Samsung 990 PRO 2 TB.
NVIDIA driver during recordings: 616.64. Captured process: r5apex_dx12.exe.
Steam build verified after reinstall on September 7: 24828373 (not independently established for the earlier captures).
Reproduction: play normally at a 240 FPS cap for at least 10–20 minutes and watch for a sustained slowdown followed by spontaneous recovery. It can occur without an unusually demanding scene. Recorded page-fault burst onsets were about 315 seconds apart; this is not a guaranteed timer from game launch or a guarantee that every session reproduces it.
Measured examples, September 7, 2026, Eastern time (UTC−4):
• 17:50:27–17:50:43: approximately 238 → 210 → 238 FPS before/during/after, with about 241,000 process page faults/sec during the burst.
• 17:55:43–17:55:59: approximately 239 → 191 → 230 FPS, with about 254,000 faults/sec. These onsets were 315.36 seconds apart; more than 45 GiB physical RAM remained available. These recordings did not use heavy WPR CPU tracing.
• 18:45:50–18:46:06: approximately 236 → 193 → 237 FPS. A short fault trace triggered during this episode; zero lost events were reported.
The short trace records 867,781 Apex transition faults and 41,003 demand-zero faults, with no Apex hard-fault disk-read events. 403 sampled stacks on a System thread contain EasyAntiCheat_EOS.sys+0x3ad7e3 → NtUnlockVirtualMemory, often followed by MiFreeWsleList/MiWsleFree. This suggests a possible working-set interaction, but does NOT prove an EAC defect: call arguments and target address ranges were not captured and have not been directly linked to Apex's faulting pages. Analysis used AI assistance, Microsoft TraceEvent and matching Microsoft kernel symbols; please independently review the raw evidence.
Already tried: Steam file verification; clean boot; prior DX11/DX12 testing; overlay/background-app testing; different FPS caps (including 158, with drops still reported); power settings; and a full Steam uninstall/reinstall with leftover local Apex settings/assets/shader cache moved aside. The same slowdown recurred after reinstall. It also persisted in a test at native 2560×1440 with only +fps_max 240, instead of my usual stretched 2080×1440 and letterbox overrides. Those later observations were not recorded. Graphics settings varied between some sessions; the measured comparisons above use neighboring windows within each recording.
Could you check whether this matches a known performance regression or working-set behavior in the current Apex/EAC integration, and advise on a supported fix? I have the short ETL, CapFrameX recordings, memory counters, exact timestamps and decoder details available. Please provide a private upload route for the raw diagnostic files if needed.
