Eglantine crashes on launch
The window opens, then the process is gone
Issue 73 describes a session that runs, hits 0 FPS, returns to 60, and then the emulator simply closes. Issue 596 describes a crash right after the loading screen on an RTX 50-series card with v0.1.0-rc1. Those are both “it closed,” and they are not the same bug. The rc release title says DONT USE, so the first swap on that machine is any later build with files, not another copy of the rc.
A second cluster is the graphics driver. Issue 341 is an AMDGPU kernel crash on RDNA3 under Linux. Issue 370 is Mario Kart World dying inside MoltenVK while a pipeline is created, on an Apple M4, builds 158 and 168. A Windows MSVC zip that exits before any window can still be the missing Visual C++ redistributable the handbook warns about. The MinGW PGO zip says it bundles those DLLs, so switching zip is a test, not a ritual.
Builds 73 and 64 tell people who crash as a game starts, after the regression that began around build 47, to launch once with EDEN_NX2_SPARSE_LAZY set to 0. I keep that flag attached to those notes. It is not a universal first step for an M4 or for a missing DLL. When it still closes, the issue form wants build, CPU, GPU, driver, settings, steps, a screenshot and the log.
Questions
Do I reinstall Windows?
Nothing in these issues asks for that. Swap the zip, leave the rc behind, then collect the log.
Is every crash the sparse-lazy flag?
No. That flag is documented for the post-47 game-launch regression. Driver crashes and the Mac pipeline crash are separate.
Sources
Links from the same publisher count as one source; a community compilation counts as one source even when it contains many reports.
| Source | Status | Checked |
|---|---|---|
| Issue 73 | Public page · 1 source | 2026-11-07 |
| Issue 370 | Public page · 1 source | 2026-11-07 |
| Eglantine releases | Public page · 1 source | 2026-11-07 |