diff options
| author | Stefan Werner <stefan.werner@tangent-animation.com> | 2019-10-16 11:29:04 +0300 |
|---|---|---|
| committer | Stefan Werner <stefan.werner@tangent-animation.com> | 2019-10-16 11:29:04 +0300 |
| commit | 35a545b752319833f8a231ea4ebbb46fc2650515 (patch) | |
| tree | 398f85040d71b69a8e59153f3bf3e7120e8a2fdb /intern/cycles/device | |
| parent | 76e8d2cc174315e3f71b561872e4fb88cfd75743 (diff) | |
Cycles: Allow PTX targets for CUDA kernel build.
This is intended for developers on Windows primarily:
Now, CUDA architectures of type compute_xx are supported. This allows for quicker builds,
at the expense of the CUDA driver running ptxas the first time a kernel is loaded.
Differential Revision: https://developer.blender.org/D5953
Diffstat (limited to 'intern/cycles/device')
| -rw-r--r-- | intern/cycles/device/device_cuda.cpp | 6 |
1 files changed, 6 insertions, 0 deletions
diff --git a/intern/cycles/device/device_cuda.cpp b/intern/cycles/device/device_cuda.cpp index 00dd37f089c..b5e10b0c2cb 100644 --- a/intern/cycles/device/device_cuda.cpp +++ b/intern/cycles/device/device_cuda.cpp @@ -454,6 +454,12 @@ class CUDADevice : public Device { VLOG(1) << "Using precompiled kernel."; return cubin; } + const string ptx = path_get(string_printf("lib/%s_compute_%d%d.ptx", name, major, minor)); + VLOG(1) << "Testing for pre-compiled kernel " << ptx << "."; + if (path_exists(ptx)) { + VLOG(1) << "Using precompiled kernel."; + return ptx; + } } const string common_cflags = compile_kernel_get_common_cflags( |