- User Since
- Jan 18 2016, 12:14 PM (282 w, 11 h)
That's a nice improvement! I even think that this kind of loop optimization will be useful in other parts of the manta code too.
We should place it in a more generic container so that it can be used everywhere (e.g. plugin/secondaryparticles.cpp has similar for-loops with radius offsets).
Thu, May 27
Tue, May 25
+1, geonodes should always be considered non-static
Tue, May 18
Running macOS with AMD graphics here. I am not seeing the issue, all good for me.
Mon, May 17
Apr 29 2021
Apr 22 2021
Apr 14 2021
Apr 12 2021
Apr 3 2021
Mar 29 2021
Mar 26 2021
Mar 24 2021
Mar 23 2021
@Marc (Polyflogger) It seems this issue comes from OpenVDB and blosc (v1.5.0) directly: A build with the latest blosc (v1.21.0) has no problems whatsoever.
Mar 20 2021
Mar 18 2021
Hi together! It looks like this problem comes from the OpenVDB compression.
Until this is fixed, can you try using a compression method that is not "Blosc" (Cache -> Advanced -> Compression Volumes)?
One thing: macOS arm64 currently uses different versions for LLVM and OpenMP (rB21236af80c8e).
make source_archive_complete will not fetch these when creating the archive on a system != macOS arm64.
Mar 17 2021
Just ran this too. LGTM, have nothing to add to the discussion.
It's unclear to me why this landed without proper testing a day before bcon2. We should have tested this patch with the upcoming macOS arm buildbot and functionality directly on arm devices.
Mar 16 2021
Mar 12 2021
Mar 11 2021
Mar 9 2021
Didn't see any issues either!
Mar 8 2021
Mar 5 2021
Mar 4 2021
Thanks for the explanations and the patch.
Mar 3 2021
@Phil Hargett (hargettp) ARM libraries have been updated, so after make update builds should work again!
Mar 2 2021
Mar 1 2021
To prevent platform flags from being overridden, use them within the Python flags variables.
Feb 26 2021
Okay, this patch is actually just for me :)
Feb 25 2021
Builds fine on macOS. Yes, I'd say if there is a heads-up elsewhere (e.g. bchat) then I wouldn't mind getting in such smaller changes quicker.
Feb 24 2021
Builds fine on macOS, let's get this in!
Feb 23 2021
Now appending the platform flags to the python flags. @Brecht Van Lommel (brecht) not sure if this is the nicest way to do this. Appending to CFLAGS within the export didn't work for me - maybe some weird formatting issue than I'm overseeing.
Feb 22 2021
Right, it built fine but probably just ignored the CMAKE_ARGS.
Feb 19 2021
@Sybren A. Stüvel (sybren) On macOS x86 it builds just fine. For macOS arm I can test next week (I am fairly certain there will be a problem with OSL C++11 and LLVM 11).
What about ILMBASE_HOME, OPENEXR_HOME and LLVM_ROOT? Can they be removed? (also @Ray molenkamp (LazyDodo))
Confirmed on macOS Catalina (AMD GPU) with rB55c1021be449
Oh, the upgrade to llvm 11 in that commit will most likely break the OSL build on macOS arm ... @Ray molenkamp (LazyDodo) this is not over yet :)
Feb 18 2021
@Ankit Meel (ankitm) Interesting .. not yet sure if it's is related to the issue but this looks odd to me:
- CMake/deps: Removed duplicate shared libs variables