Black screen after Great Deku Tree dialogue on AMD RX 5700 XT – game keeps rendering (~40 fps, 600–990 draws) but output is black; ~100 shader programs fail to translate #148
Labels
No labels
accessibility
autorelease: pending
autorelease: tagged
bug
dependencies
documentation
duplicate
enhancement
github_actions
good first issue
help wanted
invalid
question
testing wanted
wontfix
No milestone
No assignees
1 participant
Due date
No due date set.
Dependencies
No dependencies set
Reference
MagicBeanDev/Hookshot#148
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Description
On an AMD RX 5700 XT the game boots and plays the intro, but the screen turns black right after the Great Deku Tree finishes speaking. The game does not crash or freeze: the log shows frames still being presented at ~40 fps with 600–990 draws per frame, and the audio threads keep running. So the scene seems to be rendered, but nothing visible reaches the screen.
RX_VK_NO_FOLLOW=1(set through a .bat launcher before starting the game) does not change anything. It is still black.System
Steps to reproduce
launch_nofollow.bat, which setsRX_VK_NO_FOLLOW=1, or directly).Expected: gameplay is visible.
Actual: black screen.
What the log shows (full log attached: hookshot_log_1000af.txt)
Rendering keeps running:
14520 presents, 39.5 fps; a frame: 619 draws (61 skipped) ... 21840 programs (100 failed), 799 textures (2004 MB)Programs fail to translate (about 100 failed in total):
mesh program ... not translated: Failed to track indirect branch is not implemented... not translated: a network left out on this cardNVN2 program ... (stage 7) does not lower, with these reasons:vk: AMD Radeon RX 5700 XT, Vulkan 1.4, driver 0x800161; mesh shaders no, host memory import yes, pipeline libraries yesvk: matrix multiplies in software (the cooperative matrices lack 16x8x16 f16)Other messages near the end of the log:
vk: graphics pipeline for 5fa56485a6b07619 not made (-13)(an earlier run also had6eaa646d07ea4e43 not made (-13))vk: a draw (program e99f305c16ced78f) samples a target the pass wrote, after 2 draws of the pass (1152x648): a pass of its ownvk: upload over a drawn imagemessagesNotes
hookshot_programs.bin: 0 program keys kept). An earlier run with 21,628 cached program keys showed only 10 failed programs, so the cache may hide the real failures in later runs.Tried
RX_VK_NO_FOLLOW=1: no change, still black.Attachments: hookshot_log_1000af.txt
hookshot_log.txt
Thanks for the detailed log. The RX 5700 XT (RDNA1) has no mesh shader support in Vulkan (your log says 'mesh shaders no'), so the ~100 programs that fail to translate are the mesh-shader ones and that scenery is skipped. The black screen after the Deku Tree is a separate problem, probably the same one as #127, where an RX 9060 XT got in game with
RX_VK_OOT_VISUAL=0andRX_VK_NO_FOLLOW=1. Could you try those two (inhookshot_env.txtnext tohookshot.exe, one per line)?Yes i already tried
RX_VK_OOT_VISUAL=0RX_VK_NO_FOLLOW=1.(in
hookshot_env.txt)so rdna1 is not supported?
it's possible the problem on the RX 6000 series GPUs is a different one, as they have "mesh shaders yes", and seem to get stuck at different point, with the RX 6000 hanging before tree scene, while this issue with RX 5700 XT is after the tree scene
log using https://github.com/NotRexed/Hookshot/releases/tag/foliage-linux-updater
hookshot_log.txt
Tracked in #1001, follow there. That issue collects every report of GPUs without mesh shaders, and this one is linked from it with its GPU, version and log details. v2.0.6 brings back the fallback for GPUs without mesh shaders (PR #779). It is untested on these cards, so a v2.0.6 result from you would help. If it still happens on v2.0.6 (https://github.com/NotRexed/Hookshot/releases/tag/v2.0.6; Update next to Play in the launcher), please add a comment there with your
hookshot_log.txt.