[OpenGL] [sparse] Basic support for dynamic SNode #1256
Conversation
archibate
changed the title
archibate
added
the
error-prone
archibate
requested review from
k-ye and
yuanming-hu
and removed request for
yuanming-hu
archibate
changed the title
Codecov Report
@@ Coverage Diff @@
## master #1256 +/- ##
==========================================
+ Coverage 85.35% 85.48% +0.12%
==========================================
Files 19 19
Lines 3374 3375 +1
Branches 630 630
==========================================
+ Hits 2880 2885 +5
+ Misses 362 358 -4
Partials 132 132
Continue to review full report at Codecov.
|
|
Hello? I thought I made a big step.. |
|
Hello? 7 days passed and nobody react me except bots. |
There was a problem hiding this comment.
LGTM! Sorry, back then I was wondering whether it's good to support a somewhat limited dynamic. I'd also like to hear @yuanming-hu 's opinion on this. Can we discuss on this kind of decision before implementing it next time?
| @@ -2,6 +2,10 @@ | |||
|
|
|||
|
|
|||
| def ti_support_dynamic(test): | |||
There was a problem hiding this comment.
Add a test case for top-level dynamic?
The limited support here is good enough for ordinary purpose. Taichi THREE for example, only wants extend-able array for loading meshes. |
|
@yuanming-hu Could you take a pass? It's becoming stale. |
There was a problem hiding this comment.
Thanks for supporting dynamic here for OpenGL. I think it's very helpful in the short term. In the long term, we are refactoring the dynamic API, so I think we should stick to a basic implementation in OpenGL and switch to a more sophisticated implementation of the new system later.
Due to time constraints I'm sorry that I won't have time to take a close look at your implementation. Please feel free to merge.
Related issue = #711
[Click here for the format server]
Basic support!
for i in range(dynamic_len): list[i] = ifor listgen (low efficiency but works).ti.root.dense.dynamic(non-trivial, will do iapr).I also noticed that making msgbuf[] size to be explicit cause NV GLSL compiler to be slow. Should I open an extra buf for msg to make NV quicker? It may push mesa GL more close to it's buf number limit of 12.