Age | Commit message (Collapse) | Author |
|
|
|
|
|
Support thread local storage for BLI_task_parallel_mempool,
as well as support for the reduce and free callbacks.
mempool_iter_threadsafe_* functions have been moved into a private
header thats only shared between task_iterator.c and BLI_mempool.c
so the TLS can be made part of the iterator array without having to
rely on passing in struct offsets.
Add test task.MempoolIterTLS that ensures reduce and free
are working as expected.
Reviewed By: mont29
Ref D11548
|
|
Fix the data management bug where evaluation of lattice objects would
write back to the CoW copy of the Lattice ID, even when that copy was
shared between objects.
Each lattice object evaluation now stores its own evaluated data copy
via `BKE_object_eval_assign_data()`.
Reviewed By: sergey
Maniphest Tasks: T78650
Differential Revision: https://developer.blender.org/D10790
|
|
In preparation of adding Neon support.
Ref D8237, T78710
|
|
|
|
Regression introduced by {rB042143440d76}. The deform group lookup was
performed on the wrong object. Before the lookup was performed on
the lattice object. This patch changes this back.
Reviewed By: Bastien Montagne
Differential Revision: https://developer.blender.org/D9638
|
|
|
|
Follow-up to rBc0beeeb5de0cbc, fixing overflow accesses on arrays
introduced by rB042143440d76.
It's never 'OK' to access invalid memory...
|
|
Caused Victor.blend from the cloud to crash with ASan. Four floats are
fetched from the stack, but the buffer was only three wide.
Caused by 042143440d. Issue was probably harmless since the fourth
element wasn't actually touched.
|
|
This patch improves the single core performance of the lattice deform.
1. Prefetching deform vert during initialization. This data is constant for
each innerloop. This reduces the complexity of the inner loop what makes
more CPU resources free for other optimizations.
2. Prefetching the Lattice instance. It was constant. Although performance
wise this isn't noticeable it is always good to free some space in the
branch prediction tables.
3. Remove branching in all loops by not exiting when the effect of the loop
isn't there. The checks in the inner loops detected if this loop didn't
have any effect on the final result and then continue to the next loop.
This made the branch prediction unpredictable and a lot of mis
predictions were done. For smaller inner loops it is always better
to remove unpredictable if statements by using branchless code patterns.
4. Use SSE2 instruction when available.
This gives 50% performance increase measured on a
Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz with GCC 9.3.
Also check other compilers.
Before:
```
performance_no_dvert_10000 (4 ms)
performance_no_dvert_100000 (30 ms)
performance_no_dvert_1000000 (268 ms)
performance_no_dvert_10000000 (2637 ms)
```
After:
```
performance_no_dvert_10000 (3 ms)
performance_no_dvert_100000 (21 ms)
performance_no_dvert_1000000 (180 ms)
performance_no_dvert_10000000 (1756 ms)
```
Reviewed By: Campbell Barton
Differential Revision: https://developer.blender.org/D9087
|
|
No functional changes
|
|
Also order sizeof(..) first to promote other values to size_t.
|
|
|
|
This changes curve deform code not to set the objects inverse matrix,
this shouldn't cause problems as it's not used elsewhere afterwards.
|
|
- BKE_lattice_deform_data_create was init_latt_deform.
- BKE_lattice_deform_data_destroy was end_latt_deform.
- BKE_lattice_deform_data_eval_co was calc_latt_deform.
|
|
This is an improvement over 2.7x which converted edit-mesh to mesh
(CDDM) for all 3 modifiers.
Overall this increases performance in edit-mode by around 15-20%.
|
|
Move armature/curve functions into their headers,
they were previously in BKE_lattice.h
|