Blender Git Loki

Cycles: Enable OptiX on first generation Maxwell GPUs again

Fix OptiX being shown as available on first generation Maxwell GPUs

The OptiX kernels are compiled for target "compute_sm_52", which is only available on second
generation Maxwell GPUs, so disable support for older ones.

Cycles: Use pre-compiled PTX kernel for older generation when no matching one is found

This patch changes the discovery of pre-compiled kernels, to look for any PTX, even if
it does not match the current architecture version exactly. It works because the driver can
JIT-compile PTX generated for architectures less than or equal to the current one.
This e.g. makes it possible to render on a new GPU architecture even if no pre-compiled
binary kernel was distributed for it as part of the Blender installation.

Reviewed By: brecht

Differential Revision: https://developer.blender.org/D8332

Cycles: Enable OptiX on all Maxwell+ GPUs

format

groundwork for a secondary cache to prefetch to data

frame based cache: load data in the cache when running out of frames

WIP: curve sculpting

GPencil: For Invert filling be sure there are two pixels thickness

Now a pixel is added to get a minimum thickness of 2 pixels. This avoid false contacts with empty pixels.

GPencil: Small adjust to zoom calculation

Old value was too high

GPencil: Avoid infinite loop checking first pixel

Merge branch 'master' into temp_bmesh_multires

* Yet another attempt at writing a faster distance to tri function.
I think I will end up writing a less accurate version and be done
with it. This is a consistent hotspot in profiling.
* Fixed a few more undo bugs

avoid loading curve data when it is homogenous

add support for instancing

load data for the frames based on the cache method

EEVEE: Split closure_lit_lib.glsl

Merge branch 'master' into temp-geometry-nodes-instances-api-v2

EEVEE: Fix lightgrid accumulation not using diffuse_accum

Also add safe normalize to interpolated normals.

EEVEE: Replace constant booleans by const float

This makes principled optimization easier and less verbose. Tests shows
no differences in performance.