LuisaCompute
LuisaCompute copied to clipboard
LuisaGroup
High-Performance Rendering Framework on Stream Architectures
**Describe the bug** 报错信息： > panic occurred: 'swapchain context is not initialized' in file 'luisa_compute_backend_impl\src\cpu\mod.rs' at line 190 我尝试过以clang-cl和llvm作为toolchain，都会出现这个panic，而使用cmake构建是能正常运行的，我怀疑可能是xmake构建时一些构建配置没能正确传递给rust那边 为了能成功构建，我改了`scripts\xmake_func.lua`里对 “Windows SDK”的要求： ``` if #vers > 2 then --if tonumber(vers[3])...
Currently it is working normally on the CUDA backend. Tested with `test_path_tracing_*`, `test_rtx_*` and `test_photon_mapping`. 2 tests demonstrate motion blur: `test_rtx_vertex_motion_blur` and `test_rtx_instance_motion_blur`. Other backends do not function due to...
**Describe the bug** Some backends copy host data to staging buffer at submission time. If the host data is written by device then later upload to device again without syncing...
Being a gorgeous GRAPHICS framework that supports RAY TRACING, LuisaCompute should support tracing-time motion blur! And we desperately need it for our renderer. Could this be scheduled? ༼ಢ_ಢ༽
**Is your feature request related to a problem? Please describe.** LuisaCompute can't operate on Linux + AMD GPU's **Describe the solution you'd like** Vulkan / HIP backend(s) additionally to current...
...instead of simply crash See also #119
So we can better support the use cases like in #117.
**Is your feature request related to a problem? Please describe.** I would like to make an interactive app that computes things on GPU and can blit the results to screen....
Implement blockwise synchronize on CPU backend.
Metadata
Owner
Metadata
High-Performance Rendering Framework on Stream Architectures