#include "case.h"

#include "ccv_case.h"

#include "ccv_nnc_case.h"

#include <ccv.h>

#include <nnc/ccv_nnc.h>

#include <nnc/ccv_nnc_easy.h>

#include "3rdparty/dsfmt/dSFMT.h"

TEST_SETUP()

{

  ccv_nnc_init();

}

TEST_CASE("simple graph x * 0.3 + 3 compiled with NUMA")

{

  ccv_nnc_symbolic_graph_t* symbolic_graph = ccv_nnc_symbolic_graph_new();

  const ccv_nnc_tensor_symbol_t a = ccv_nnc_tensor_symbol_new(symbolic_graph, CPU_NUMA_TENSOR_NHWC(000, 32F, 1), "a");

  const ccv_nnc_tensor_symbol_t b = ccv_nnc_tensor_symbol_new(symbolic_graph, CPU_NUMA_TENSOR_NHWC(002, 32F, 1), "b");

  const ccv_nnc_tensor_symbol_t x = ccv_nnc_tensor_symbol_new(symbolic_graph, CPU_NUMA_TENSOR_NHWC(000, 32F, 1), "x");

  const ccv_nnc_tensor_symbol_t y = ccv_nnc_tensor_symbol_new(symbolic_graph, CPU_NUMA_TENSOR_NHWC(001, 32F, 1), "y");

  const ccv_nnc_tensor_symbol_t z = ccv_nnc_tensor_symbol_new(symbolic_graph, CPU_NUMA_TENSOR_NHWC(002, 32F, 1), "z");

  ccv_nnc_graph_exec_symbol_new(symbolic_graph, CMD_EWPROD_FORWARD(), TENSOR_SYMBOL_LIST(x, a), TENSOR_SYMBOL_LIST(y), "prod");

  ccv_nnc_graph_exec_symbol_new(symbolic_graph, CMD_EWSUM_FORWARD(), TENSOR_SYMBOL_LIST(y, b), TENSOR_SYMBOL_LIST(z), "sum");

  ccv_nnc_graph_exec_symbol_autogen(symbolic_graph, 0, 0, CCV_NNC_AUTOGEN_ALL_EXECS | CCV_NNC_AUTOGEN_SOURCES_AND_DESTINATIONS);

  SYMBOLIC_GRAPH_GEN(symbolic_graph, CCV_NNC_LONG_DOT_GRAPH);

  ccv_nnc_graph_t* graph = 0;

  ccv_nnc_tensor_arena_t* tensor_arena = 0;

  ccv_nnc_graph_exec_arena_t* graph_exec_arena = 0;

  ccv_nnc_symbolic_graph_compile(symbolic_graph, ccv_nnc_default_compile_params, 0, 0, 0, 0, SYMBOLIC_GRAPH_SOURCES(symbolic_graph), SYMBOLIC_GRAPH_DESTINATIONS(symbolic_graph), &graph, &tensor_arena, &graph_exec_arena);

  GRAPH_GEN(graph, CCV_NNC_LONG_DOT_GRAPH);

  ccv_nnc_tensor_t* const x_tensor = ccv_nnc_tensor_from_symbol(tensor_arena, x);

  x_tensor->data.f32[0] = 1.21;

  ccv_nnc_tensor_t* const a_tensor = ccv_nnc_tensor_from_symbol(tensor_arena, a);

  a_tensor->data.f32[0] = 0.3;

  ccv_nnc_tensor_t* const b_tensor = ccv_nnc_tensor_from_symbol(tensor_arena, b);

  b_tensor->data.f32[0] = 3;

  ccv_nnc_tensor_t* const y_tensor = ccv_nnc_tensor_from_symbol(tensor_arena, y);

  ccv_nnc_tensor_t* const z_tensor = ccv_nnc_tensor_from_symbol(tensor_arena, z);

  REQUIRE(x_tensor->data.f32 != y_tensor->data.f32, "y cannot be assigned to the same place of x due to NUMA constraint");

  REQUIRE(z_tensor->data.f32 != x_tensor->data.f32, "z cannot be assigned to the same place of x due to NUMA constraint");

  REQUIRE(z_tensor->data.f32 != y_tensor->data.f32, "z cannot be assigned to the same place of y due to NUMA constraint");

  ccv_nnc_graph_exec_t source = ccv_nnc_graph_exec_source(graph_exec_arena);

  ccv_nnc_graph_exec_t destination = ccv_nnc_graph_exec_destination(graph_exec_arena);

  ccv_nnc_graph_run(graph, 0, &source, 1, &destination, 1, 0, 0);

  REQUIRE_EQ_WITH_TOLERANCE(z_tensor->data.f32[0], 1.21 * 0.3 + 3, 1e-6, "should be equal to 1.21 * 0.3 + 3");

  ccv_nnc_graph_free(graph);

  ccv_nnc_tensor_arena_free(tensor_arena);

  ccv_nnc_graph_exec_arena_free(graph_exec_arena);

  ccv_nnc_symbolic_graph_free(symbolic_graph);

}

#include "case_main.h"