Game: Arc B390 (Panther Lake iGPU, Windows): the GPU is lost a few minutes into play, twice at a dispatch of compute program 499dbfd0721d2858 #606

New issue

Open

opened 2026-11-14 23:34:57 +00:00 by ertomas · 0 comments

ertomas commented

2026-11-14 23:34:57 +00:00

Longshot version

Built from source, scripts/dev_packs.py packs over two bases:

base f647a072 (v3.0.0-rc.3-13-gf647a072), with main fbd42024's renderer (longshot_vk.dll)
base 414f56e2 (v3.0.0-rc.3-82-g414f56e2), with #443's renderer (its head c8201e51 merged onto 414f56e2)

Channel

Built from source

What kind of problem

Crash, freeze or black screen

What happened

A few minutes in, the GPU is lost (VK_ERROR_DEVICE_LOST) or hangs, and the game closes. In the four runs below that didn't end normally, the last unfinished GPU work was a compute dispatch every time the game could record it. In two of them it was the same program, 499dbfd0721d2858, on two different builds.

Run	Build	After	How it ended	Last unfinished work (the device-loss checkpoints)
1	`f647a072` + main fbd42024's renderer	frame 5775, ~4.4 min	`FATAL: the GPU hung: no submission finished in 30 s, and the driver never reported the device lost`	stage 0x1: `dispatch of program 6637ac1d052d22d4`; stage 0x2000: `INDIRECT dispatch of program 685adb41edc19de7`
2	`f647a072` + main fbd42024's renderer	frame 2987, ~2.2 min	`FATAL: the GPU device was lost (VK_ERROR_DEVICE_LOST)`	stage 0x1 and 0x2000: `dispatch of program 499dbfd0721d2858`
3	`414f56e2` + #443	frame 10145, ~3.5 min, right as the opening movie ended	`VK_ERROR_DEVICE_LOST`, crash record `reason: device lost`, thread Presentation Thread	stage 0x1 and 0x2000: `dispatch of program 499dbfd0721d2858`
4	`414f56e2` + #443, `RX_VK_SPLIT_BEFORE_DRAW=0`	frame 17987, ~6 min, in play	the window closed by itself. The log stops mid-line: no FATAL, no crash record, no Windows Application error	not recorded

Other runs on the same PC ran to a normal close: about 22 min on f647a072 alone, about 3 min with #189, and about 6 min on 414f56e2 + #443. So it doesn't happen every time.

Not the cause: sleep. The System log has no Modern Standby or sleep event during runs 1–4. The shader preparation also finished on both bases, so there were no pending compiles in play: 0.00 waits for shaders or pipelines.

Possibly related: #343 (B580, black screen and crash right after the opening cutscene, every time), and #189's earlier series (device losses in 5 of 8 runs without its change, on an Intel iGPU). Run 2 predates #189 being merged, but run 3 doesn't: #189 is in 414f56e2.

Where in the game

Kokiri Forest, a new game or a save near Link's house. Run 3 died as the opening movie ended; the others while walking around.

How often

4 of the last 7 runs on this PC.

Graphics card and driver version

Intel Arc B390 (Panther Lake iGPU, PCI VEN_8086&DEV_B080, no VRAM mirror: VRAM mirror of buffer pools off), driver 32.0.101.9150

CPU, RAM and operating system

Intel Core Ultra X7 358H (16 threads), 32 GB, Windows 11 Pro Insider Preview build 26220

Changes from the defaults

None (default options, output 2048x1152 or 2560x1440).

longshot_log.txt

The four runs' logs are on this PC, and I can attach them or run with more logging on request. A switch that names what program 499dbfd0721d2858 is, or dumps the dispatch, would help.

Before you send it

I searched the open issues for it.
I'm not attaching game files, keys, the game's executable, its .bin caches or saves.
The problem is still there on the newest release or nightly. (Seen on main 414f56e2 from source; not checked on a nightly.)

### Longshot version Built from source, `scripts/dev_packs.py` packs over two bases: - base f647a072 (`v3.0.0-rc.3-13-gf647a072`), with main fbd42024's renderer (`longshot_vk.dll`) - base 414f56e2 (`v3.0.0-rc.3-82-g414f56e2`), with #443's renderer (its head c8201e51 merged onto 414f56e2) ### Channel Built from source ### What kind of problem Crash, freeze or black screen ### What happened A few minutes in, the GPU is lost (`VK_ERROR_DEVICE_LOST`) or hangs, and the game closes. In the four runs below that didn't end normally, the last unfinished GPU work was a compute dispatch every time the game could record it. In two of them it was the same program, `499dbfd0721d2858`, on two different builds. | Run | Build | After | How it ended | Last unfinished work (the device-loss checkpoints) | |---|---|---|---|---| | 1 | f647a072 + main fbd42024's renderer | frame 5775, ~4.4 min | `FATAL: the GPU hung: no submission finished in 30 s, and the driver never reported the device lost` | stage 0x1: `dispatch of program 6637ac1d052d22d4`; stage 0x2000: `INDIRECT dispatch of program 685adb41edc19de7` | | 2 | f647a072 + main fbd42024's renderer | frame 2987, ~2.2 min | `FATAL: the GPU device was lost (VK_ERROR_DEVICE_LOST)` | stage 0x1 and 0x2000: `dispatch of program 499dbfd0721d2858` | | 3 | 414f56e2 + #443 | frame 10145, ~3.5 min, right as the opening movie ended | `VK_ERROR_DEVICE_LOST`, crash record `reason: device lost`, thread Presentation Thread | stage 0x1 and 0x2000: `dispatch of program 499dbfd0721d2858` | | 4 | 414f56e2 + #443, `RX_VK_SPLIT_BEFORE_DRAW=0` | frame 17987, ~6 min, in play | **the window closed by itself.** The log stops mid-line: no FATAL, no crash record, no Windows Application error | not recorded | Other runs on the same PC ran to a normal close: about 22 min on f647a072 alone, about 3 min with #189, and about 6 min on 414f56e2 + #443. So it doesn't happen every time. Not the cause: sleep. The System log has no Modern Standby or sleep event during runs 1–4. The shader preparation also finished on both bases, so there were no pending compiles in play: `0.00 waits for shaders or pipelines`. Possibly related: #343 (B580, black screen and crash right after the opening cutscene, every time), and #189's earlier series (device losses in 5 of 8 runs without its change, on an Intel iGPU). Run 2 predates #189 being merged, but run 3 doesn't: #189 is in 414f56e2. ### Where in the game Kokiri Forest, a new game or a save near Link's house. Run 3 died as the opening movie ended; the others while walking around. ### How often 4 of the last 7 runs on this PC. ### Graphics card and driver version Intel Arc B390 (Panther Lake iGPU, PCI `VEN_8086&DEV_B080`, no VRAM mirror: `VRAM mirror of buffer pools off`), driver 32.0.101.9150 ### CPU, RAM and operating system Intel Core Ultra X7 358H (16 threads), 32 GB, Windows 11 Pro Insider Preview build 26220 ### Changes from the defaults None (default options, output 2048x1152 or 2560x1440). ### longshot_log.txt The four runs' logs are on this PC, and I can attach them or run with more logging on request. A switch that names what program `499dbfd0721d2858` is, or dumps the dispatch, would help. ### Before you send it - [x] I searched the open issues for it. - [x] I'm not attaching game files, keys, the game's executable, its .bin caches or saves. - [ ] The problem is still there on the newest release or nightly. (Seen on main 414f56e2 from source; not checked on a nightly.)