AMD GPU (7900XTX) issue/crash with miner v3.0.1

[Fatfingerjoe]

Hi,

i got some trouble on the new quantus-miner v3.0.1 with an AMD Radeon 7900XTX GPU environment on Ubuntu Linux.

Environment

quantus-miner: v3.0.1
quantus-node: v0.6.3
quantus-cli: v1.3.3
GPU generation: AMD Radeon RX 7900 XTX (RADV NAVI31) (Backend: Vulkan)
Driver: amd mesa
v: 25.2.8-0ubuntu0.25.10.1 glx-v: 1.4 direct-render: yes renderer: AMD
Radeon RX 7900 XTX (radeonsi navi31 LLVM 20.1.8 DRM 3.64
6.17.0-22-generic)
Backend: Vulkan
OS: Ubuntu 25.10
Node state: fully synced
Miner mode: external miner via QUIC to 127.0.0.1:9833
What the miner tells me:

[2026-04-20T19:33:43Z INFO engine_gpu] Initializing GPU adapter 0: AMD Radeon RX 7900 XTX (RADV NAVI31) (Backend: Vulkan)
[2026-04-20T19:33:43Z INFO gpu_engine] Adapter 0 detailed info:
[2026-04-20T19:33:43Z INFO gpu_engine] Name: AMD Radeon RX 7900 XTX (RADV NAVI31)
[2026-04-20T19:33:43Z INFO gpu_engine] Vendor: 4098
[2026-04-20T19:33:43Z INFO gpu_engine] Device: 29772
[2026-04-20T19:33:43Z INFO gpu_engine] Device Type: DiscreteGpu
[2026-04-20T19:33:43Z INFO gpu_engine] Driver: radv
[2026-04-20T19:33:43Z INFO gpu_engine] Driver Info: Mesa 25.2.8-0ubuntu0.25.10.1
[2026-04-20T19:33:43Z INFO gpu_engine] Backend: Vulkan
[2026-04-20T19:33:43Z INFO gpu_engine] Device limits for adapter 0:
[2026-04-20T19:33:43Z INFO gpu_engine] Max workgroups per dimension: 65535
[2026-04-20T19:33:43Z INFO gpu_engine] Max workgroup size X: 256
[2026-04-20T19:33:43Z INFO gpu_engine] Max workgroup size Y: 256
[2026-04-20T19:33:43Z INFO gpu_engine] Max workgroup size Z: 64
[2026-04-20T19:33:43Z INFO gpu_engine] Max compute invocations per workgroup: 256
[2026-04-20T19:33:43Z INFO gpu_engine] Max buffer size: 268435456
[2026-04-20T19:33:44Z INFO gpu_engine] Vendor-specific dispatch configuration:
[2026-04-20T19:33:44Z INFO gpu_engine] Max hardware workgroups: 65535
[2026-04-20T19:33:44Z INFO gpu_engine] Optimal workgroups: 6553
[2026-04-20T19:33:44Z INFO engine_gpu] Initializing GPU adapter 1: llvmpipe (LLVM 20.1.8, 256 bits) (Backend: Vulkan)
[2026-04-20T19:33:44Z INFO gpu_engine] Adapter 1 detailed info:
[2026-04-20T19:33:44Z INFO gpu_engine] Name: llvmpipe (LLVM 20.1.8, 256 bits)
[2026-04-20T19:33:44Z INFO gpu_engine] Vendor: 65541
[2026-04-20T19:33:44Z INFO gpu_engine] Device: 0
[2026-04-20T19:33:44Z INFO gpu_engine] Device Type: Cpu
[2026-04-20T19:33:44Z INFO gpu_engine] Driver: llvmpipe
[2026-04-20T19:33:44Z INFO gpu_engine] Driver Info: Mesa 25.2.8-0ubuntu0.25.10.1 (LLVM 20.1.8)
[2026-04-20T19:33:44Z INFO gpu_engine] Backend: Vulkan
[2026-04-20T19:33:44Z INFO gpu_engine] Device limits for adapter 1:
[2026-04-20T19:33:44Z INFO gpu_engine] Max workgroups per dimension: 65535
[2026-04-20T19:33:44Z INFO gpu_engine] Max workgroup size X: 256
[2026-04-20T19:33:44Z INFO gpu_engine] Max workgroup size Y: 256
[2026-04-20T19:33:44Z INFO gpu_engine] Max workgroup size Z: 64
[2026-04-20T19:33:44Z INFO gpu_engine] Max compute invocations per workgroup: 256
[2026-04-20T19:33:44Z INFO gpu_engine] Max buffer size: 268435456
[2026-04-20T19:33:45Z INFO gpu_engine] Vendor-specific dispatch configuration:
[2026-04-20T19:33:45Z INFO gpu_engine] Max hardware workgroups: 65535
[2026-04-20T19:33:45Z INFO gpu_engine] Optimal workgroups: 4095
....
[2026-04-20T19:35:02Z INFO gpu_engine] GPU 0 search started: range 7866e152..1340e154, nonces: 4294967295, cancel check interval: 10000
radv/amdgpu: The CS has been cancelled because the context is lost. This context is guilty of a hard recovery.

thread '' (6770) panicked at /home/runner/.cargo/registry/src/index.crates.io-1949cf8c6b5b557f/wgpu-27.0.1/src/backend/wgpu_core.rs:1826:30:
Error in Device::poll: Validation Error

Caused by:
Parent device is lost

0: 0x62e057688aa2 - unknown
1: 0x62e05736f8ba - unknown
2: 0x62e057687e85 - unknown
3: 0x62e057687b00 - unknown
4: 0x62e0576b0407 - unknown
5: 0x62e0576b03b9 - unknown
6: 0x62e0576b03ac - unknown
7: 0x62e05737080b - unknown
8: 0x62e057704910 - unknown
9: 0x62e0573aa000 - unknown
10: 0x62e0573fee5d - unknown
11: 0x62e0573fdea5 - unknown
12: 0x62e0573fddfb - unknown
13: 0x62e0576b7be8 - unknown
14: 0x7a89510a3d64 - start_thread
at ./nptl/pthread_create.c:448:8
15: 0x7a89511373fc - __GI___clone3
at ./misc/../sysdeps/unix/sysv/linux/x86_64/clone3.S:78:0
16: 0x0 -

if you need more information let me now (my name in Telegram is ZR1R)

something i noticed which is not connected to this problem:

for now i can mine with CPU, my name appears on the telemetry dashboard but after a while my name disappear and never show up again. (but my node and miner seems to work just fine without any trouble) maybe just a small bug?