VTK-m has been updated to replace old per device benchmark executables with a device
dependent shared library so that it's able to accept a device adapter at runtime through
the "--device=" argument.
Mask objects allow you to specify which output values should be
generated when a worklet is run. That is, the Mask allows you to skip
the invocation of a worklet for any number of outputs.
af4aef991 Add missing %= operator to ArrayPortalValueReference
9e02cd33a Declare ArrayPortalValueReference::operator= as const
b141f7515 Add more operator= to ArrayPortalValueReference
c8db70ae8 Fix warning about automatic conversions loosing precision
ddc6c91e3 Fix error about constexpr not available on CUDA device
1ca55ac31 Add specialized operators for ArrayPortalValueReference
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Allison Vacanti <allison.vacanti@kitware.com>
Merge-request: !1542
8a38717c3 have make_shared call default constructor
3fae21fa7 no pointers
4cd4637c1 ray tracing now using shared pointers for intersectors
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !1556
Declaring operator= as const seems a little weird because we are
changing the value. But remember that ArrayPortalReference is only a
reference class. The reference itself does not change, just the thing
that it is referencing. So declaring as const is correct and necessary
so that you can set the value of a reference returned from a function
(which is a right hand side value).
For some reason, these changes caused one of the CUDA compilers to
create an error about a variable declared constexpr not being available
on the device. That sounds like a bug in nvcc as the constexpr should
just be evaluated rather than stored in some part of memory. At any
rate, changing the constexpr to a preprocessing macro solves the
problem.
The ArrayPortalValueReference is supposed to behave just like the value
it encapsulates and does so by automatically converting to the base type
when necessary. However, when it is possible to convert that to
something else, it is possible to get errors about ambiguous overloads.
To avoid these, add specialized versions of the operators to specify
which ones should be used.
Also consolidated the CUDA version of an ArrayPortalValueReference to the
standard one. The two implementations were equivalent and we would like
changes to apply to both.
18ba2baf3 Suppress system lib memory leak warnings found by asan
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Robert Maynard <robert.maynard@kitware.com>
Merge-request: !1550
Due to a bug in pthread, loguru would leak 12 bytes memory from the main
thread when `loguru::init` is in use. Since GCC does not support
blacklist file, the suppression logic is added to
ctest_memcheck_supp_file file.
See details in loguru github issue #59.
d19524016 Refactor View class and fix its memory leak
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Robert Maynard <robert.maynard@kitware.com>
Merge-request: !1546
Do not check for the copyright statement for files in the third party
directory. These files shouldn't have the VTK-m copyright. Frankly, I
don't understand why this has not been a problem before.
451bff6f1 vtkmdiy: Use the new mangled VTKM_DIY_ defines
7d307b96c Merge branch 'upstream-diy' into diy_conditional_serialization
0adcc4d45 diy 2019-02-08 (72a201e1)
74acc2a7b vtkmdiy: Support only including the serialization headers of diy
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Matt Larsen <larsen30@llnl.gov>
Merge-request: !1541
