New morton class with arithmetic and comparison operators #860
Conversation
… on HLSL side by specializing , a bunch of morton operators
…or both cpp and hlsl
|
|
||
| NBL_CONSTEXPR_FUNC T operator()(NBL_CONST_REF_ARG(T) operand) | ||
| { | ||
| return -operand; |
There was a problem hiding this comment.
doesn't it need to be operand.operator-() because of HLSL?
or maybe you want to do a partial spec with requires Fundamental<T> and requires !Fundamental<T> where you do operator-() call ?
| template<> | ||
| struct is_emulating_integral_scalar<emulated_uint64_t> |
There was a problem hiding this comment.
where's the primary template declaration?
btw you could actually replace impl::is_emulating_integral_scalar with just ImitationIntegral64Scalar concept
There was a problem hiding this comment.
it is below is_emulating_floating_point_scalar
There was a problem hiding this comment.
just use the EmulatedIntegral64Scalar concept and call it a day, one less thing in impl
There was a problem hiding this comment.
is_emulating_integral_scalar might be needed if we want to emulate uint128_t in the future. Or this is unlikeyly?
|
|
||
| template<typename Scalar, typename U> | ||
| struct Promote<vector <Scalar, 2>, U> | ||
| // TODO(kevinyu): Should we enable truncation from uint64_t to emulated_vector<emulated_uint64_t, N>? |
There was a problem hiding this comment.
promotion yes, allow it
| } | ||
| }; | ||
|
|
||
| // The default version above only works for fundamental scalars, vectors and matrices. This is because you can't call `~x` unless `x` is one of the former. |
There was a problem hiding this comment.
then constrain the version above to only do FundamentalType
the thing which you have as a specialization you should have as the general case, and the general you should have as a specialization
There was a problem hiding this comment.
is_fundamental_v is set to is_scalar_v or void. Should we add fundamental to include concepts::Vector and concepts::Matrix?
| template<typename T NBL_STRUCT_CONSTRAINABLE > | ||
| struct ternary_operator | ||
| { | ||
| using type_t = T; | ||
|
|
||
| T operator()(bool condition, NBL_CONST_REF_ARG(T) lhs, NBL_CONST_REF_ARG(T) rhs) | ||
| NBL_CONSTEXPR_FUNC T operator()(NBL_CONST_REF_ARG(bool) condition, NBL_CONST_REF_ARG(T) lhs, NBL_CONST_REF_ARG(T) rhs) | ||
| { | ||
| return select<bool, T>(condition, lhs, rhs); | ||
| } | ||
| }; |
There was a problem hiding this comment.
hmm actually I noticed something, in C++ and HLSL the ? shortcircuits (whereas select shouldn't)
This means that ternary_operator can only be defined as
template<typename F1, typename F2> requires is_same_v<decltype(std::declval<F1>()()),decltype(std::declval<F2>()())>
struct ternary_operator
{
using type_t = decltype(std::declval<F1>().operator());
NBL_CONSTEXPR_FUNC type_t operator()(const bool condition, NBL_CONST_REF_ARG(F1) lhs, NBL_CONST_REF_ARG(F2) rhs)
{
if (condition)
return lhs();
else
return rhs();
}
};simply to take lambdas and short-circuit (not call/evaluate the branch that is false)
so sorry about telling you to reimplement ternary_operator in terms of select
(but most ccrrent uses of ternary_operator should be replaced with select)
There was a problem hiding this comment.
In Nabla itself there should only be this, which needs to change to select_helper partial spec
Nabla/include/nbl/builtin/hlsl/complex.hlsl
Line 440 in 407051c
There was a problem hiding this comment.
In the example the ternary should be replaced with select
https://github.com/search?q=repo%3ADevsh-Graphics-Programming%2FNabla-Examples-and-Tests%20ternary_operator%20&type=code
| #define NBL_CONSTEXPR_OOL_MEMBER const | ||
| #define NBL_CONSTEXPR_INLINE_OOL_MEMBER const | ||
| #define NBL_IF_CONSTEXPR(...) if (__VA_ARGS__) | ||
| #define NBL_ASSERT(...) |
There was a problem hiding this comment.
wel aready have a shader assert somwhere, it does a vk::RawBufferStore to an invalid (0) BDA
There was a problem hiding this comment.
include/nbl/builtin/hlsl/morton.hlsl
Outdated
| NBL_CONSTEXPR_FUNC bool verifyAnyBitIntegralVec(vector<T, Dim> vec) | ||
| { | ||
| array_get<vector<T, Dim>, T> getter; | ||
| [[unroll]] |
There was a problem hiding this comment.
NBL_UNROLL or it won't compile in C++
There was a problem hiding this comment.
see
Nabla/include/nbl/builtin/hlsl/macros.h
Line 38 in 1e637c4
Description
Adds a new class for 2,3 and 4-dimensional morton codes, with arithmetic and comparison operators
Testing
TODO
TODO list:
Need to make sure all operators work properly before merging