FATAL EXIT — "Allocator failed to allocate" memory-pool crash, mid-game (Build 16.2.10604)

Age of Empires IV IV - Report a Bug
1

  • Build Number: 16.2.10604.0

  • Platform: Xbox / Microsoft Store (Game Pass PC version)

  • Operating System: Windows 11

Issue:

The game crashes to desktop mid-match with a FATAL EXIT. It happens roughly twice per play session and has been constant since April 2026 - I can’t finish ranked or multiplayer games. (New account, so I can’t attach files yet - pasting the relevant log lines inline below.)

The fatal crash (warnings.log):

RangeFixedBlockAllocator::Alloc (requested:312), (capacity:1434451968) pool 5 was full, trying next pool with larger blocksize(1024).
RangeFixedBlockAllocator::Alloc (requested:312), (capacity:1434451968) No pool had room for this allocation.
(F) Allocator failed to allocate:  requested(312), current usage(443890754), capacity(1673527296)
(F) -- FATAL EXIT --
Application crashed.

This looks identical to other reports - same requested(312) and the same capacity(1673527296) ceiling - so the internal fixed-block memory pool seems to exhaust regardless of system RAM (I have 64 GB).

Framerate visibly decays across the match right before the crash (warnings.log):

Local user framerate is [41.18]
Local user framerate is [40.59]
Local user framerate is [37.27]

Every crash is preceded by multi-second main-thread stalls / lock waits (LogFiles\unhandled.*.txt):

NetworkManager::DispatchEvents() - Warning - Time between calls was 5363 ms
SessionManager::Process() - Warning - Time between calls was 2480 ms, time waiting for lock was 2469 ms

Frequency: 11 crashes logged, roughly 2 per session: 2026-04-11 (x2), 04-18 (x2), 05-02 (x2), 05-03 (x2), 06-06 (x2), 06-14 (x1).

The game is correctly using the dedicated GPU (warnings.log), so this is not a wrong-GPU issue:

Chosen adapter: NVIDIA RTX PRO 2000 Blackwell Generation Laptop GPU
NVIDIA RTX PRO 2000 ... 7951MB dedicated video memory available
Vendor: NVIDIA (0x10DE)   Installed Driver Version: r595_00

System (well above spec, not a hardware limit):
CPU: Intel Core Ultra 9 285H / RAM: 64 GB
GPU: NVIDIA RTX PRO 2000 Blackwell Laptop, 8 GB VRAM, driver r595
OS: Windows 11

2

Same issue here. Game just hard resets while trying to play on a custom map with my friend.

3

Update — 2026-07-03. Still crashing, now on Build 16.2.10884.0 (Game Pass PC).

Two things I ruled out today; neither stopped it:

  • Updated NVIDIA driver r595 → 610.47 (crash log confirms Installed Driver Version: r610_45). No change.
    • Enabled texture streaming (it had been off). No change.
      The very first match after both changes crashed with the identical failure — and pool usage was actually lower this time, which makes the fragmentation angle clearer. Pool walk from warnings.log:
RangeFixedBlockAllocator::Alloc (requested:312), (capacity:1434451968) pool 5 was full, trying next pool with larger blocksize(1024).
RangeFixedBlockAllocator::Alloc (requested:312), (capacity:1434451968) No pool had room for this allocation.
(F) Allocator failed to allocate:  requested(312), current usage(423546740), capacity(1673527296)
(F) -- FATAL EXIT --

Note current usage(423546740) ≈ 423 MB against capacity(1673527296) ≈ 1.67 GB — a 312-byte allocation fails while the pool is ~75% empty. This isn’t running out of memory; no fixed-block size-class pool has a free slot. Same 312-byte request and same 1.67 GB ceiling as every earlier report.

Same lead-up as always — framerate decay, then multi-second main-thread stalls right before the exit:

NetworkManager::DispatchEvents() - Warning - Time between calls was 5322 ms
SessionManager::Process() - Warning - Time between calls was 5312 ms waiting for lock

Updated frequency: 13+ crashes now, ~2/session, spanning 2026-04-11 → 2026-07-03, across two builds (16.2.10604 and 16.2.10884) and two driver branches (r595 and r610). No config or hardware change stops it — this is a game-side allocator bug.

(@MaskedBasil3614 above plays these custom-map co-op games with me and sees the same hard-reset. It may just be his client dropping when my host process dies rather than an independent hit, so I won’t overstate it — flagging in case the shared-session timing is useful.)

4

The Game is busted!!!

5

Very frustrating, we haven’t been able to get through a full game in ‘ages’. I switched to a whole different computer, having claude look at logs again here. Anyone else facing this issue? We’re at a point where we simply can’t play anymore. I guess we’ll switch to AOEIII until this is figured out. It has been crashy for us since november with a recent uptick of not being able to finish a game at all now.

6

Update: confirmed on 3 machines, including two players crashing simultaneously in the same match.

Earlier reports in this thread are from my laptop (Core Ultra 9 285H / 64 GB / RTX PRO 2000). Today, on Build 16.2.10884.0 (Game Pass PC), my desktop (i7-11700 / 32 GB / RTX 3070 / driver r590) and my co-op partner’s PC (32 GB / AMD Radeon RX 9070 XT 16 GB / Adrenalin 26.10.21.05, no mods active) both hit the identical FATAL EXIT — within ~0.5 s of each other, ~42 minutes into the same 4-player custom match, seconds after a player reached Imperial:

RangeFixedBlockAllocator::Alloc (requested:312), (capacity:1434451968) pool 5 was full, trying next pool with larger blocksize(1024).
RangeFixedBlockAllocator::Alloc (requested:312), (capacity:1434451968) No pool had room for this allocation.
(F) Allocator failed to allocate:  requested(312), current usage(423033104), capacity(1673527296)
(F) -- FATAL EXIT --

Key evidence for the devs:

  • current usage(423033104) is byte-identical on both machines (Intel/NVIDIA and AMD) at the moment of death — the arena is being filled by the deterministic lockstep sim, so all peers exhaust it at the same tick and the whole lobby crashes together. This is why it looks like a “hard reset” for everyone in the match.
  • Same requested(312) and capacity(1673527296) across three machines, two GPU vendors, three driver versions, mods on and off — this is a hard-coded engine arena, hardware-independent.
  • The failure is size-class exhaustion, not memory pressure: the arena was only ~25% used (423 MB of 1.67 GB), but the 512-byte pool (pool 5) and all larger pools were full.
  • No degradation before death: one machine was at 50 fps, the other at 108 fps one second before the crash; network counters healthy on both; the only stall recorded is at the fatal instant (caused by the crash, not causing it).

Reproduction profile: long (40+ min) 4-player custom lobbies, large map (size 640), late Castle/Imperial with big armies. Happy to provide full warnings.log files from all machines on request.

7

77 views and nothing? Are me and my buddy seriously alone in this crashing?

Another weekend has arrived and again we can’t play our fav game. I guess AOE 3 will become our favorite game again.