20623330469

Committed 31 Dec 2025 04:51PM UTC coverage: 39.635% (-0.08%) from 39.712%

Build # 20623330469

Build Type

Pull #421

github

Committed by

web-flow

Commit Message

Merge 6f97fcd37 into 3b72c335e

Pull Request Pull Request #421: Extend tensor library nodes with primitive type support and refactor CMathNode to use enums

Run Details

14997 of 49248 branches covered (30.45%)

Branch coverage included in aggregate %.

247 of 620 new or added lines in 52 files covered. (39.84%)

37 existing lines in 4 files now uncovered.

12875 of 21074 relevant lines covered (61.09%)

89.41 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

0.0

/src/data_flow/library_nodes/math/tensor/elementwise_ops/minimum_node.cpp

#include "sdfg/data_flow/library_nodes/math/tensor/elementwise_ops/minimum_node.h"

#include "sdfg/analysis/analysis.h"
#include "sdfg/builder/structured_sdfg_builder.h"
#include "sdfg/data_flow/library_nodes/math/cmath/cmath_node.h"
#include "sdfg/data_flow/library_nodes/math/tensor/tensor_node.h"
#include "sdfg/types/type.h"

namespace sdfg {
namespace math {
namespace tensor {

MinimumNode::MinimumNode(
    size_t element_id,
    const DebugInfo& debug_info,
    const graph::Vertex vertex,
    data_flow::DataFlowGraph& parent,
    const std::vector<symbolic::Expression>& shape
)
    : ElementWiseBinaryNode(element_id, debug_info, vertex, parent, LibraryNodeType_Minimum, shape) {}

bool MinimumNode::expand_operation(
    builder::StructuredSDFGBuilder& builder,
    analysis::AnalysisManager& analysis_manager,
    structured_control_flow::Sequence& body,
    const std::string& input_name_a,
    const std::string& input_name_b,
    const std::string& output_name,
    const types::IType& input_type_a,
    const types::IType& input_type_b,
    const types::IType& output_type,
    const data_flow::Subset& subset
) {
    auto& code_block = builder.add_block(body);
    data_flow::AccessNode* input_node_a;
    if (builder.subject().exists(input_name_a)) {
        input_node_a = &builder.add_access(code_block, input_name_a);
    } else {
        input_node_a = &builder.add_constant(code_block, input_name_a, input_type_a);
    }
    data_flow::AccessNode* input_node_b;
    if (builder.subject().exists(input_name_b)) {
        input_node_b = &builder.add_access(code_block, input_name_b);
    } else {
        input_node_b = &builder.add_constant(code_block, input_name_b, input_type_b);
    }
    auto& output_node = builder.add_access(code_block, output_name);

    bool is_int = types::is_integer(input_type_a.primitive_type());

    if (is_int) {
        // Use tasklets for integer types - distinguish between signed and unsigned
        auto tasklet_code = TensorNode::get_integer_minmax_tasklet(input_type_a.primitive_type(), false);
        auto& tasklet = builder.add_tasklet(code_block, tasklet_code, "_out", {"_in1", "_in2"});

        if (input_type_a.type_id() == types::TypeID::Scalar) {
            builder.add_computational_memlet(code_block, *input_node_a, tasklet, "_in1", {}, input_type_a);
        } else {
            builder.add_computational_memlet(code_block, *input_node_a, tasklet, "_in1", subset, input_type_a);
        }
        if (input_type_b.type_id() == types::TypeID::Scalar) {
            builder.add_computational_memlet(code_block, *input_node_b, tasklet, "_in2", {}, input_type_b);
        } else {
            builder.add_computational_memlet(code_block, *input_node_b, tasklet, "_in2", subset, input_type_b);
        }
        builder.add_computational_memlet(code_block, tasklet, "_out", output_node, subset, output_type);
    } else {
        // Use intrinsics for floating-point types with correct suffix
        auto& node = builder.add_library_node<
            cmath::CMathNode>(code_block, this->debug_info(), cmath::CMathFunction::fmin, input_type_a.primitive_type());

        if (input_type_a.type_id() == types::TypeID::Scalar) {
            builder.add_computational_memlet(code_block, *input_node_a, node, "_in1", {}, input_type_a, DebugInfo());
        } else {
            builder.add_computational_memlet(code_block, *input_node_a, node, "_in1", subset, input_type_a, DebugInfo());
        }
        if (input_type_b.type_id() == types::TypeID::Scalar) {
            builder.add_computational_memlet(code_block, *input_node_b, node, "_in2", {}, input_type_b, DebugInfo());
        } else {
            builder.add_computational_memlet(code_block, *input_node_b, node, "_in2", subset, input_type_b, DebugInfo());
        }
        builder.add_computational_memlet(code_block, node, "_out", output_node, subset, output_type, DebugInfo());
    }

    return true;
}

std::unique_ptr<data_flow::DataFlowNode> MinimumNode::
    clone(size_t element_id, const graph::Vertex vertex, data_flow::DataFlowGraph& parent) const {
    return std::unique_ptr<
        data_flow::DataFlowNode>(new MinimumNode(element_id, this->debug_info(), vertex, parent, this->shape_));
}

} // namespace tensor
} // namespace math
} // namespace sdfg

1	#include "sdfg/data_flow/library_nodes/math/tensor/elementwise_ops/minimum_node.h"
2
3	#include "sdfg/analysis/analysis.h"
4	#include "sdfg/builder/structured_sdfg_builder.h"
5	#include "sdfg/data_flow/library_nodes/math/cmath/cmath_node.h"
6	#include "sdfg/data_flow/library_nodes/math/tensor/tensor_node.h"
7	#include "sdfg/types/type.h"
8
9	namespace sdfg {
10	namespace math {
11	namespace tensor {
12
13	MinimumNode::MinimumNode(	×
14	size_t element_id,
15	const DebugInfo& debug_info,
16	const graph::Vertex vertex,
17	data_flow::DataFlowGraph& parent,
18	const std::vector<symbolic::Expression>& shape
19	)
20	: ElementWiseBinaryNode(element_id, debug_info, vertex, parent, LibraryNodeType_Minimum, shape) {}	×
21
22	bool MinimumNode::expand_operation(	×
23	builder::StructuredSDFGBuilder& builder,
24	analysis::AnalysisManager& analysis_manager,
25	structured_control_flow::Sequence& body,
26	const std::string& input_name_a,
27	const std::string& input_name_b,
28	const std::string& output_name,
29	const types::IType& input_type_a,
30	const types::IType& input_type_b,
31	const types::IType& output_type,
32	const data_flow::Subset& subset
33	) {
34	auto& code_block = builder.add_block(body);	×
35	data_flow::AccessNode* input_node_a;
36	if (builder.subject().exists(input_name_a)) {	×
37	input_node_a = &builder.add_access(code_block, input_name_a);	×
38	} else {	×
39	input_node_a = &builder.add_constant(code_block, input_name_a, input_type_a);	×
40	}
41	data_flow::AccessNode* input_node_b;
42	if (builder.subject().exists(input_name_b)) {	×
43	input_node_b = &builder.add_access(code_block, input_name_b);	×
44	} else {	×
45	input_node_b = &builder.add_constant(code_block, input_name_b, input_type_b);	×
46	}
47	auto& output_node = builder.add_access(code_block, output_name);	×
48
49	bool is_int = types::is_integer(input_type_a.primitive_type());	×
50
51	if (is_int) {	×
52	// Use tasklets for integer types - distinguish between signed and unsigned
NEW 53	auto tasklet_code = TensorNode::get_integer_minmax_tasklet(input_type_a.primitive_type(), false);	×
NEW 54	auto& tasklet = builder.add_tasklet(code_block, tasklet_code, "_out", {"_in1", "_in2"});	×
55
56	if (input_type_a.type_id() == types::TypeID::Scalar) {	×
57	builder.add_computational_memlet(code_block, *input_node_a, tasklet, "_in1", {}, input_type_a);	×
58	} else {	×
59	builder.add_computational_memlet(code_block, *input_node_a, tasklet, "_in1", subset, input_type_a);	×
60	}
61	if (input_type_b.type_id() == types::TypeID::Scalar) {	×
62	builder.add_computational_memlet(code_block, *input_node_b, tasklet, "_in2", {}, input_type_b);	×
63	} else {	×
64	builder.add_computational_memlet(code_block, *input_node_b, tasklet, "_in2", subset, input_type_b);	×
65	}
66	builder.add_computational_memlet(code_block, tasklet, "_out", output_node, subset, output_type);	×
67	} else {	×
68	// Use intrinsics for floating-point types with correct suffix
NEW 69	auto& node = builder.add_library_node<	×
NEW 70	cmath::CMathNode>(code_block, this->debug_info(), cmath::CMathFunction::fmin, input_type_a.primitive_type());	×
71
72	if (input_type_a.type_id() == types::TypeID::Scalar) {	×
73	builder.add_computational_memlet(code_block, *input_node_a, node, "_in1", {}, input_type_a, DebugInfo());	×
74	} else {	×
75	builder.add_computational_memlet(code_block, *input_node_a, node, "_in1", subset, input_type_a, DebugInfo());	×
76	}
77	if (input_type_b.type_id() == types::TypeID::Scalar) {	×
78	builder.add_computational_memlet(code_block, *input_node_b, node, "_in2", {}, input_type_b, DebugInfo());	×
79	} else {	×
80	builder.add_computational_memlet(code_block, *input_node_b, node, "_in2", subset, input_type_b, DebugInfo());	×
81	}
82	builder.add_computational_memlet(code_block, node, "_out", output_node, subset, output_type, DebugInfo());	×
83	}
84
85	return true;	×
86	}	×
87
88	std::unique_ptr<data_flow::DataFlowNode> MinimumNode::
89	clone(size_t element_id, const graph::Vertex vertex, data_flow::DataFlowGraph& parent) const {	×
90	return std::unique_ptr<	×
91	data_flow::DataFlowNode>(new MinimumNode(element_id, this->debug_info(), vertex, parent, this->shape_));	×
92	}	×
93
94	} // namespace tensor
95	} // namespace math
96	} // namespace sdfg

daisytuner / sdfglib / 20623330469

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous