The GPU address map for device faults is off by default (#1174)

With VK_EXT_device_address_binding_report enabled, AMD's Windows driver (26.10.1) tracks every bound range in a
chained hash map that it searches on nearly every recorded command. The chains grow as textures stream in and out, so
after 5-10 minutes in a busy area recording a submission cost 3-4 times more and an RX 9070 XT fell from 60 to 20-30
fps. The map only serves the fault report after a device lost, so TOTK_ADDR_REPORT now defaults to 0 and =1 turns it
on for a run that chases a fault. ADDR_REPORT is a known key now, so the ini line no longer logs "unknown key".

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Valiantsin DzerakhandClaude Opus 5.5 committed 2026-11-15 20:13:00 +03:00
1 parent 6151ac4fbc
commit c5b1cc7fac
5 files changed
+18 -7

No files matched your search

+5
View File
@@ -516,6 +516,11 @@
# last one it finished. Unset: on AMD GPUs (VK_AMD_buffer_marker). 1: also on NVIDIA GPUs
# (VK_NV_device_diagnostic_checkpoints), which NVIDIA says costs a little. 0: off.
# pass_crumbs = 1
# GPU address map: after a GPU memory fault (device lost) the log names the image or buffer that held the faulting
# address and whether it was freed before the fault. 1: on, where the driver has VK_EXT_device_address_binding_report.
# It slows the recording of GPU commands, on AMD's Windows driver more and more over a session, so turn it on only for
# the run that reproduces the crash. Default: 0.
# addr_report = 0
# Render targets up to this size (bytes, 0 to 1073741824) go back to their game memory whenever the game waits for a
# GPU fence. A larger target goes back too unless the GPU drew it in each of the last 16 frames. A game that reads a
# larger target drawn every frame without invalidating it sees stale bytes at the default. With trace = gfx the log