Xuejun/snippet remove inference engine (#21293)
* [CPU Plugin][Snippet] remove InferenceEngine, replace with ov Signed-off-by: Zhai, Xuejun <xuejun.zhai@intel.com> * [CPU Plugin][Snippet] remove ngraph namespace Signed-off-by: Zhai, Xuejun <xuejun.zhai@intel.com> --------- Signed-off-by: Zhai, Xuejun <xuejun.zhai@intel.com>
This commit is contained in:
parent
815980f290
commit
4fdbb2d4e8
@ -4,7 +4,6 @@
|
|||||||
|
|
||||||
#include "snippets/enforce_precision.hpp"
|
#include "snippets/enforce_precision.hpp"
|
||||||
#include <gtest/gtest.h>
|
#include <gtest/gtest.h>
|
||||||
#include <ngraph/ngraph.hpp>
|
|
||||||
|
|
||||||
namespace ov {
|
namespace ov {
|
||||||
namespace test {
|
namespace test {
|
||||||
|
@ -8,7 +8,6 @@
|
|||||||
#include "snippets/fake_quantize_decomposition_test.hpp"
|
#include "snippets/fake_quantize_decomposition_test.hpp"
|
||||||
|
|
||||||
using namespace LayerTestsDefinitions;
|
using namespace LayerTestsDefinitions;
|
||||||
using namespace ngraph;
|
|
||||||
|
|
||||||
namespace {
|
namespace {
|
||||||
|
|
||||||
@ -16,7 +15,7 @@ namespace decompositionInSubgraph {
|
|||||||
const std::vector<TestValues> testValuesDecompositionScalars = {
|
const std::vector<TestValues> testValuesDecompositionScalars = {
|
||||||
{
|
{
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
ngraph::Shape{1, 3, 16, 16},
|
ov::Shape{1, 3, 16, 16},
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
1.f,
|
1.f,
|
||||||
{{}, {}, {}, {}},
|
{{}, {}, {}, {}},
|
||||||
@ -25,23 +24,23 @@ const std::vector<TestValues> testValuesDecompositionScalars = {
|
|||||||
const std::vector<TestValues> testValuesDecompositionPerChannel = {
|
const std::vector<TestValues> testValuesDecompositionPerChannel = {
|
||||||
{
|
{
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
ngraph::Shape{1, 3, 16, 16},
|
ov::Shape{1, 3, 16, 16},
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
1.f,
|
1.f,
|
||||||
{{1, 3, 1, 1}, {1, 3, 1, 1}, {1, 3, 1, 1}, {1, 3, 1, 1}},
|
{{1, 3, 1, 1}, {1, 3, 1, 1}, {1, 3, 1, 1}, {1, 3, 1, 1}},
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
ngraph::Shape{1, 3, 16, 16},
|
ov::Shape{1, 3, 16, 16},
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
1.f,
|
1.f,
|
||||||
{{1, 3, 1, 1}, {1, 3, 1, 1}, {}, {}},
|
{{1, 3, 1, 1}, {1, 3, 1, 1}, {}, {}},
|
||||||
},
|
},
|
||||||
};
|
};
|
||||||
|
|
||||||
std::vector<std::pair<std::shared_ptr<Node>, std::pair<std::string, std::string> >> operations = {
|
std::vector<std::pair<std::shared_ptr<ov::Node>, std::pair<std::string, std::string> >> operations = {
|
||||||
{std::make_shared<opset1::Abs>(), {"Subgraph", "Abs,fakeQuantize"}},
|
{std::make_shared<ov::op::v0::Abs>(), {"Subgraph", "Abs,fakeQuantize"}},
|
||||||
{std::make_shared<ngraph::op::v4::Swish>(), {"Subgraph", "Swish,fakeQuantize"}},
|
{std::make_shared<ov::op::v4::Swish>(), {"Subgraph", "Swish,fakeQuantize"}},
|
||||||
};
|
};
|
||||||
|
|
||||||
INSTANTIATE_TEST_SUITE_P(
|
INSTANTIATE_TEST_SUITE_P(
|
||||||
@ -83,36 +82,36 @@ namespace legacyFuse {
|
|||||||
const std::vector<TestValues> testValuesLegacyFuse = {
|
const std::vector<TestValues> testValuesLegacyFuse = {
|
||||||
{
|
{
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
ngraph::Shape{1, 3, 16, 16},
|
ov::Shape{1, 3, 16, 16},
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
1.f,
|
1.f,
|
||||||
{{1, 3, 1, 1}, {1, 3, 1, 1}, {}, {}}
|
{{1, 3, 1, 1}, {1, 3, 1, 1}, {}, {}}
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
ngraph::Shape{1, 3, 16, 16},
|
ov::Shape{1, 3, 16, 16},
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
1.f,
|
1.f,
|
||||||
{{}, {}, {1, 3, 1, 1}, {1, 3, 1, 1}}
|
{{}, {}, {1, 3, 1, 1}, {1, 3, 1, 1}}
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
ngraph::Shape{1, 3, 16, 16},
|
ov::Shape{1, 3, 16, 16},
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
1.f,
|
1.f,
|
||||||
{{}, {}, {}, {}}
|
{{}, {}, {}, {}}
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
ngraph::Shape{1, 3, 16, 16},
|
ov::Shape{1, 3, 16, 16},
|
||||||
ov::element::f32,
|
ov::element::f32,
|
||||||
1.f,
|
1.f,
|
||||||
{{1, 3, 1, 1}, {1, 3, 1, 1}, {1, 3, 1, 1}, {1, 3, 1, 1}}
|
{{1, 3, 1, 1}, {1, 3, 1, 1}, {1, 3, 1, 1}, {1, 3, 1, 1}}
|
||||||
},
|
},
|
||||||
};
|
};
|
||||||
|
|
||||||
std::vector<std::pair<std::shared_ptr<Node>, std::pair<std::string, std::string>>> operations = {
|
std::vector<std::pair<std::shared_ptr<ov::Node>, std::pair<std::string, std::string>>> operations = {
|
||||||
{std::make_shared<opset1::Convolution>(), {"Convolution", "Convolution,fakeQuantize"}},
|
{std::make_shared<ov::op::v1::Convolution>(), {"Convolution", "Convolution,fakeQuantize"}},
|
||||||
};
|
};
|
||||||
|
|
||||||
INSTANTIATE_TEST_SUITE_P(
|
INSTANTIATE_TEST_SUITE_P(
|
||||||
|
@ -25,7 +25,7 @@ std::vector<std::vector<ov::PartialShape>> input_shapes{
|
|||||||
static inline std::vector<std::vector<element::Type>> quantized_precisions() {
|
static inline std::vector<std::vector<element::Type>> quantized_precisions() {
|
||||||
std::vector<std::vector<element::Type>> prc = {};
|
std::vector<std::vector<element::Type>> prc = {};
|
||||||
// In Snippets MatMul INT8 is supported only on VNNI/AMX platforms
|
// In Snippets MatMul INT8 is supported only on VNNI/AMX platforms
|
||||||
if (InferenceEngine::with_cpu_x86_avx512_core_vnni() || InferenceEngine::with_cpu_x86_avx512_core_amx_int8()) {
|
if (ov::with_cpu_x86_avx512_core_vnni() || ov::with_cpu_x86_avx512_core_amx_int8()) {
|
||||||
prc.emplace_back(std::vector<element::Type>{element::i8, element::i8});
|
prc.emplace_back(std::vector<element::Type>{element::i8, element::i8});
|
||||||
prc.emplace_back(std::vector<element::Type>{element::u8, element::i8});
|
prc.emplace_back(std::vector<element::Type>{element::u8, element::i8});
|
||||||
}
|
}
|
||||||
@ -40,7 +40,7 @@ static inline std::vector<std::vector<element::Type>> precisions(bool only_fp32
|
|||||||
auto quant = quantized_precisions();
|
auto quant = quantized_precisions();
|
||||||
std::copy(quant.begin(), quant.end(), std::back_inserter(prc));
|
std::copy(quant.begin(), quant.end(), std::back_inserter(prc));
|
||||||
// In Snippets MatMul BF16 is supported only on bf16/AMX platforms
|
// In Snippets MatMul BF16 is supported only on bf16/AMX platforms
|
||||||
if (InferenceEngine::with_cpu_x86_bfloat16() || InferenceEngine::with_cpu_x86_avx512_core_amx_bf16()) {
|
if (ov::with_cpu_x86_bfloat16() || ov::with_cpu_x86_avx512_core_amx_bf16()) {
|
||||||
prc.emplace_back(std::vector<element::Type>{element::bf16, element::bf16});
|
prc.emplace_back(std::vector<element::Type>{element::bf16, element::bf16});
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
@ -30,7 +30,7 @@ const std::vector<std::vector<ov::PartialShape>> inputShapes_3D = {
|
|||||||
};
|
};
|
||||||
|
|
||||||
static inline bool is_bf16_supported() {
|
static inline bool is_bf16_supported() {
|
||||||
return InferenceEngine::with_cpu_x86_bfloat16() || InferenceEngine::with_cpu_x86_avx512_core_amx_bf16();
|
return ov::with_cpu_x86_bfloat16() || ov::with_cpu_x86_avx512_core_amx_bf16();
|
||||||
}
|
}
|
||||||
|
|
||||||
static inline std::vector<std::vector<element::Type>> precision_f32(size_t count) {
|
static inline std::vector<std::vector<element::Type>> precision_f32(size_t count) {
|
||||||
|
@ -4,7 +4,6 @@
|
|||||||
|
|
||||||
#include "snippets/precision_propagation_convertion.hpp"
|
#include "snippets/precision_propagation_convertion.hpp"
|
||||||
#include <gtest/gtest.h>
|
#include <gtest/gtest.h>
|
||||||
#include <ngraph/ngraph.hpp>
|
|
||||||
|
|
||||||
namespace ov {
|
namespace ov {
|
||||||
namespace test {
|
namespace test {
|
||||||
|
@ -18,12 +18,12 @@ static inline std::vector<std::vector<element::Type>> precisions(bool only_fp32
|
|||||||
};
|
};
|
||||||
if (!only_fp32) {
|
if (!only_fp32) {
|
||||||
// In Snippets MatMul INT8 is supported only on VNNI/AMX platforms
|
// In Snippets MatMul INT8 is supported only on VNNI/AMX platforms
|
||||||
if (InferenceEngine::with_cpu_x86_avx512_core_vnni() || InferenceEngine::with_cpu_x86_avx512_core_amx_int8()) {
|
if (ov::with_cpu_x86_avx512_core_vnni() || ov::with_cpu_x86_avx512_core_amx_int8()) {
|
||||||
prc.emplace_back(std::vector<element::Type>{element::i8, element::i8});
|
prc.emplace_back(std::vector<element::Type>{element::i8, element::i8});
|
||||||
prc.emplace_back(std::vector<element::Type>{element::u8, element::i8});
|
prc.emplace_back(std::vector<element::Type>{element::u8, element::i8});
|
||||||
}
|
}
|
||||||
// In Snippets MatMul BF16 is supported only on bf16/AMX platforms
|
// In Snippets MatMul BF16 is supported only on bf16/AMX platforms
|
||||||
if (InferenceEngine::with_cpu_x86_bfloat16() || InferenceEngine::with_cpu_x86_avx512_core_amx_bf16()) {
|
if (ov::with_cpu_x86_bfloat16() || ov::with_cpu_x86_avx512_core_amx_bf16()) {
|
||||||
prc.emplace_back(std::vector<element::Type>{element::bf16, element::bf16});
|
prc.emplace_back(std::vector<element::Type>{element::bf16, element::bf16});
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
@ -115,12 +115,12 @@ static inline std::vector<std::vector<element::Type>> precisions(bool only_fp32
|
|||||||
};
|
};
|
||||||
if (!only_fp32) {
|
if (!only_fp32) {
|
||||||
// In Snippets MatMul INT8 is supported only on VNNI/AMX platforms
|
// In Snippets MatMul INT8 is supported only on VNNI/AMX platforms
|
||||||
if (InferenceEngine::with_cpu_x86_avx512_core_vnni() || InferenceEngine::with_cpu_x86_avx512_core_amx_int8()) {
|
if (ov::with_cpu_x86_avx512_core_vnni() || ov::with_cpu_x86_avx512_core_amx_int8()) {
|
||||||
prc.emplace_back(std::vector<element::Type>{element::i8, element::i8});
|
prc.emplace_back(std::vector<element::Type>{element::i8, element::i8});
|
||||||
prc.emplace_back(std::vector<element::Type>{element::u8, element::i8});
|
prc.emplace_back(std::vector<element::Type>{element::u8, element::i8});
|
||||||
}
|
}
|
||||||
// In Snippets MatMul BF16 is supported only on bf16/AMX platforms
|
// In Snippets MatMul BF16 is supported only on bf16/AMX platforms
|
||||||
if (InferenceEngine::with_cpu_x86_bfloat16() || InferenceEngine::with_cpu_x86_avx512_core_amx_bf16()) {
|
if (ov::with_cpu_x86_bfloat16() || ov::with_cpu_x86_avx512_core_amx_bf16()) {
|
||||||
prc.emplace_back(std::vector<element::Type>{element::bf16, element::bf16});
|
prc.emplace_back(std::vector<element::Type>{element::bf16, element::bf16});
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
Loading…
Reference in New Issue
Block a user