As we’re closing in on the 580 driver release, the last to support my 1050 Ti GPU…
One more CPU usage comparison. This time apples-to-apples with the Papers GTK4 app.
Vulkan uses ~2x cpu cycles versus NGL. Ultimately this is not terrible, and improved from OP.
- PID 110808 is NGL
- PID 110851 is Vulkan
CPU USAGE (nVidia internal tracking bug #5238502 (futex)):
Crash-On-Exit (nVidia internal tracking bug #unassigned):
I will likely need to open a separate thread about the crash-on-exit issue. It’s not limited to GTK.
sway crashes on exit nearly every instance when the default wlroots vulkan backend is used.
This is a complete deal-breaker for usage of Vulkan with nVidia.
Coredumps from my brief testing under both sway and hyprland over the past week.
NOTE: vkcube crashes under Sway are a separate driver issue tracked here.
Mon 2025-07-21 06:31:20 EDT 69691 1000 1000 SIGSEGV inaccessible /usr/bin/sway
Tue 2025-07-22 11:49:10 EDT 966 1000 1000 SIGFPE present /usr/bin/vkcube
Wed 2025-07-23 14:28:36 EDT 24558 1000 1000 SIGFPE present /usr/bin/vkcube
Wed 2025-07-23 14:30:26 EDT 25037 1000 1000 SIGFPE present /usr/bin/vkcube
Thu 2025-07-24 09:45:29 EDT 19211 1000 1000 SIGSEGV inaccessible /usr/bin/sway
Thu 2025-07-24 14:51:37 EDT 30660 1000 1000 SIGSEGV present /usr/bin/kgx
Thu 2025-07-24 16:58:28 EDT 35683 1000 1000 SIGSEGV present /usr/bin/gnome-calculator
Thu 2025-07-24 17:01:37 EDT 36105 1000 1000 SIGSEGV present /usr/bin/gnome-text-editor
Thu 2025-07-24 17:02:11 EDT 36359 1000 1000 SIGSEGV present /usr/bin/gnome-text-editor
Thu 2025-07-24 17:29:44 EDT 32144 1000 1000 SIGSEGV inaccessible /usr/bin/sway
Sat 2025-07-26 08:35:57 EDT 109402 1000 1000 SIGSEGV present /usr/bin/gnome-calculator
Sat 2025-07-26 08:56:27 EDT 110166 1000 1000 SIGSEGV present /usr/bin/gnome-text-editor
Sat 2025-07-26 10:48:59 EDT 110851 1000 1000 SIGSEGV present /usr/bin/papers
Sat 2025-07-26 10:59:21 EDT 113689 1000 1000 SIGSEGV present /usr/bin/papers
Sat 2025-07-26 11:14:55 EDT 115658 1000 1000 SIGSEGV inaccessible /usr/bin/sway
Sat 2025-07-26 11:19:57 EDT 117281 1000 1000 SIGSEGV present /usr/bin/loupe
Mon 2025-07-28 09:31:42 EDT 194991 1000 1000 SIGSEGV present /usr/bin/loupe
VK_LAYER_KHRONOS_validation Khronos Validation Layer 1.4.321 version 1
VK_LAYER_NV_optimus NVIDIA Optimus layer 1.4.303 version 1
VK_LAYER_NV_present NVIDIA GR2608 layer 1.4.303 version 1
I’ve been using VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation to hopefully provide additional debug logging.
Verified it’s functioning…
$ VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation vkcube
Selected WSI platform: wayland
Selected GPU 0: NVIDIA GeForce GTX 1050 Ti, type: DiscreteGpu
Validation Error: [ VUID-VkRenderPassBeginInfo-pNext-02852 ] | MessageID = 0xdb2c5767
vkCmdBeginRenderPass(): pRenderPassBegin->renderArea offset.x (0) + extent.width (1440) is greater than framebuffer width (500).
The Vulkan spec states: If the pNext chain does not contain VkDeviceGroupRenderPassBeginInfo or its deviceRenderAreaCount member is equal to 0, renderArea.offset.x + renderArea.extent.width must be less than or equal to VkFramebufferCreateInfo::width the framebuffer was created with (https://docs.vulkan.org/spec/latest/chapters/renderpass.html#VUID-VkRenderPassBeginInfo-pNext-02852)
Objects: 2
[0] VkRenderPass 0xd000000000d
[1] VkFramebuffer 0x270000000027
…with vkcube, but outputs exactly nothing additional when Vulkan apps crash on exit.
The backtraces are near useless without debug symbols for the driver (only available to nVidia devs).
Please, could the crash-on-exit issue could be assigned a bug #?