Make the spline indicies a build option. Enableing this negatively affects cuda kernels. (9e86ec1b) · Commits · Cianciosa, Mark / graph_framework

CMakeLists.txt

+1 −0

Original line number	Diff line number	Diff line
		@@ -10,6 +10,7 @@ option (SAVE_KERNEL_SOURCE "Writes the kernel source code to a file." OFF)
		option (USE_INPUT_CACHE "Cache the values kernel input values." OFF)
		option (USE_CONSTANT_CACHE "Cache the value of constantants in kernel registers." OFF)
		option (SHOW_USE_COUNT "Add a comment showing the use count in kernel sources." OFF)
		option (USE_INDEX_CACHE "Cache index values instead of computing them every time." OFF)

		#-------------------------------------------------------------------------------
		# Set the cmake module path.

graph_framework/CMakeLists.txt

+1 −0

Original line number	Diff line number	Diff line
		@@ -24,6 +24,7 @@ target_compile_definitions (rays
		$<$<BOOL:${SAVE_KERNEL_SOURCE}>:SAVE_KERNEL_SOURCE>
		$<$<BOOL:${USE_CONSTANT_CACHE}>:USE_CONSTANT_CACHE>
		$<$<BOOL:${SHOW_USE_COUNT}>:SHOW_USE_COUNT>
		$<$<BOOL:${USE_INDEX_CACHE}>:USE_INDEX_CACHE>
		)

		target_include_directories (rays

graph_framework/piecewise.hpp

+52 −2

Original line number	Diff line number	Diff line
		@@ -293,12 +293,15 @@ void compile_index(std::ostringstream &stream,
		jit::register_map &indices,
		const jit::register_usage &usage) {
		if (registers.find(this) == registers.end()) {
		#ifdef USE_INDEX_CACHE
		if (indices.find(this->arg.get()) == indices.end()) {
		#endif
		const size_t length = leaf_node<T, SAFE_MATH>::backend_cache[data_hash].size();
		shared_leaf<T, SAFE_MATH> a = this->arg->compile(stream,
		registers,
		indices,
		usage);
		#ifdef USE_INDEX_CACHE
		indices[a.get()] = jit::to_string('i', a.get());
		stream << " const "
		<< jit::smallest_int_type<T> (length) << " "
		@@ -306,6 +309,7 @@ void compile_index(std::ostringstream &stream,
		compile_index<T> (stream, registers[a.get()], length);
		a->endline(stream, usage);
		}
		#endif

		registers[this] = jit::to_string('r', this);
		stream << " const ";
		@@ -326,22 +330,39 @@ void compile_index(std::ostringstream &stream,
		#endif
		stream << registers[leaf_node<T, SAFE_MATH>::backend_cache[data_hash].data()];
		if constexpr (jit::use_metal<T> ()) {
		#ifdef USE_INDEX_CACHE
		stream << ".read("
		<< indices[this->arg.get()];
		<< indices[this->arg.get()]
		<< ").r";
		#else
		stream << ".read(";
		compile_index<T> (stream, registers[a.get()], length);
		stream << ").r";
		#endif
		#ifdef USE_CUDA_TEXTURES
		} else if constexpr (jit::use_cuda()) {
		#ifdef USE_INDEX_CACHE
		stream << ", "
		<< indices[this->arg.get()];
		#else
		stream << ", ";
		compile_index<T> (stream, registers[a.get()], length);
		#endif
		if constexpr (jit::is_complex<T> () \|\| jit::is_double<T> ()) {
		stream << ")";
		}
		stream << ")";
		#endif
		} else {
		#ifdef USE_INDEX_CACHE
		stream << "["
		<< indices[this->arg.get()]
		<< "]";
		#else
		stream << "[";
		compile_index<T> (stream, registers[a.get()], length);
		stream << "]";
		#endif
		}
		this->endline(stream, usage);
		}
		@@ -839,11 +860,14 @@ void compile_index(std::ostringstream &stream,
		const size_t length = leaf_node<T, SAFE_MATH>::backend_cache[data_hash].size();
		const size_t num_rows = length/num_columns;

		#ifdef USE_INDEX_CACHE
		if (indices.find(this->left.get()) == indices.end()) {
		#endif
		shared_leaf<T, SAFE_MATH> x = this->left->compile(stream,
		registers,
		indices,
		usage);
		#ifdef USE_INDEX_CACHE
		indices[x.get()] = jit::to_string('i', x.get());
		stream << " const "
		<< jit::smallest_int_type<T> (num_rows) << " "
		@@ -852,10 +876,12 @@ void compile_index(std::ostringstream &stream,
		x->endline(stream, usage);
		}
		if (indices.find(this->right.get()) == indices.end()) {
		#endif
		shared_leaf<T, SAFE_MATH> y = this->right->compile(stream,
		registers,
		indices,
		usage);
		#ifdef USE_INDEX_CACHE
		indices[y.get()] = jit::to_string('i', y.get());
		stream << " const "
		<< jit::smallest_int_type<T> (num_columns) << " "
		@@ -882,6 +908,7 @@ void compile_index(std::ostringstream &stream,
		<< ";" << std::endl;
		}
		}
		#endif

		registers[this] = jit::to_string('r', this);
		stream << " const ";
		@@ -902,6 +929,7 @@ void compile_index(std::ostringstream &stream,
		#endif
		stream << registers[leaf_node<T, SAFE_MATH>::backend_cache[data_hash].data()];
		if constexpr (jit::use_metal<T> ()) {
		#ifdef USE_INDEX_CACHE
		stream << ".read("
		<< jit::smallest_int_type<T> (std::max(num_rows,
		num_columns))
		@@ -910,21 +938,43 @@ void compile_index(std::ostringstream &stream,
		<< ","
		<< indices[this->left.get()]
		<< ")).r";
		#else
		stream << ".read(uint2(";
		compile_index<T> (stream, registers[y.get()], num_columns);
		stream << ",";
		compile_index<T> (stream, registers[x.get()], num_rows);
		stream << ")).r";
		#endif
		#ifdef USE_CUDA_TEXTURES
		} else if constexpr (jit::use_cuda()) {
		#ifdef USE_INDEX_CACHE
		stream << ", "
		<< indices[this->right.get()]
		<< ", "
		<< indices[this->left.get()];
		#else
		stream << ", ";
		compile_index<T> (stream, registers[y.get()], num_columns);
		stream << ", ";
		compile_index<T> (stream, registers[x.get()], num_rows);
		#endif
		if constexpr (jit::is_complex<T> () \|\| jit::is_double<T> ()) {
		stream << ")";
		}
		stream << ")";
		#endif
		} else {
		#ifdef USE_INDEX_CACHE
		stream << "["
		<< indices[temp.get()]
		<< "]";
		#else
		stream << "[";
		compile_index<T> (stream, registers[x.get()], num_rows);
		stream << "*" << num_columns << " + ";
		compile_index<T> (stream, registers[y.get()], num_columns);
		stream << "]";
		#endif
		}
		this->endline(stream, usage);
		}