Skip to content

Commit

Permalink
test fix
Browse files Browse the repository at this point in the history
  • Loading branch information
@isanghao
isanghao committed Sep 12, 2024
1 parent 299575e commit 16ae143
Show file tree
Hide file tree
Showing 2 changed files with 25 additions and 25 deletions.
2 changes: 1 addition & 1 deletion ...ns/intel_gpu/src/kernel_selector/kernels/dynamic_quantize/dynamic_quantize_kernel_opt.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -139,7 +139,7 @@ KernelsData DynamicQuantizeKernelOpt::GetKernelsData(const Params& params) const
}

KernelsPriority DynamicQuantizeKernelOpt::GetKernelsPriority(const Params& /*params*/) const {
return FORCE_PRIORITY_1;
return FORCE_PRIORITY_2;
}

bool DynamicQuantizeKernelOpt::Validate(const Params& params) const {
Expand Down
48 changes: 24 additions & 24 deletions src/plugins/intel_gpu/tests/unit/test_cases/dynamic_quantize_gpu_test.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -26,17 +26,15 @@ using namespace ::tests;
class dynamic_quantization_gpu_tests: public ::testing::Test {
public:

void test_dynamic_quantization(bool is_caching_test, bool is_dynamic, int batch = 1, int ifm = 1024) {
void test_dynamic_quantization(bool is_caching_test, bool is_dynamic, int batch = 1, int ifm = 1024, uint64_t group_size = 1024) {
tests::random_generator rg(GET_SUITE_NAME);
auto& engine = get_test_engine();

long int batch_num = batch;
long int ifm_num = ifm;

bool is_4d = true;

auto input_ps = is_4d ? ov::PartialShape{ batch_num, 1, 1, ifm_num } : ov::PartialShape{ batch_num, ifm_num};
auto dyn_input_ps = is_4d ? ov::PartialShape{ -1, 1, 1, ifm_num } : ov::PartialShape{ -1, ifm_num};
auto input_ps = ov::PartialShape{ batch_num, 1, ifm_num };
auto dyn_input_ps = ov::PartialShape{ -1, 1, ifm_num };
auto input_mem = engine.allocate_memory({ input_ps, data_types::f32, format::bfyx });

auto input_data = rg.generate_random_1d<float>(batch_num * ifm_num, -16.0f, 16.0f);
Expand All @@ -49,7 +47,7 @@ class dynamic_quantization_gpu_tests: public ::testing::Test {
: layout{ input_ps, data_types::f16, format::bfyx };

auto reorder_1 = reorder("reorder_1", input_info("input"), layout{ input_ps, data_types::f16, format::bfyx });
auto dyn_quan_prim = dynamic_quantize("dyn_quan_prim", input_info("reorder_1"), 32, {data_types::f16, data_types::i8});
auto dyn_quan_prim = dynamic_quantize("dyn_quan_prim", input_info("reorder_1"), group_size, {data_types::f16, data_types::i8});
auto reorder_2 = reorder("reorder_2", input_info("dyn_quan_prim"), layout{ input_ps, data_types::f16, format::bfyx });

// Implemented dynamic quantize kernel
Expand Down Expand Up @@ -111,44 +109,46 @@ class dynamic_quantization_gpu_tests: public ::testing::Test {
max_diff = abs_diff;
avg += abs_diff;
count++;
if (abs_diff > 1 && i < 10)
std::cout << "idx " << i << " ref " << (int)output_ptr_ref[i] << " - " << (int)output_ptr[i] << std::endl;
OPENVINO_ASSERT(abs_diff < 1);
}
GPU_DEBUG_LOG << "---> count: " << count << ", max_diff:" << max_diff << ", avg_diff: " << (avg/count) << std::endl;
}
};

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_large_size) {
this->test_dynamic_quantization(false, false, 2048, 4096);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_per_token_large_size) {
this->test_dynamic_quantization(false, false, 2048, 4096, 4096);
}

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_large_size_dynamic) {
this->test_dynamic_quantization(false, true, 2048, 4096);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_per_token_large_size_dynamic) {
this->test_dynamic_quantization(false, true, 2048, 4096, 4096);
}

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_small_size) {
this->test_dynamic_quantization(false, false, 64, 4096);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_group_32) {
this->test_dynamic_quantization(false, false, 2048, 4096, 32);
}

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_single_batch) {
this->test_dynamic_quantization(false, false, 1, 4096);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_group_128) {
this->test_dynamic_quantization(false, false, 2048, 4096, 128);
}

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_ref_only) {
this->test_dynamic_quantization(false, false, 16, 33);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_per_token_small_size) {
this->test_dynamic_quantization(false, false, 64, 4096, 4096);
}

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_ref_only_dynamic) {
this->test_dynamic_quantization(false, true, 16, 33);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_per_token_single_batch) {
this->test_dynamic_quantization(false, false, 1, 4096, 4096);
}

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_invalid) {
this->test_dynamic_quantization(false, false, 16, 7);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_ref_only) {
this->test_dynamic_quantization(false, false, 16, 33, 33);
}

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_unaligned) {
this->test_dynamic_quantization(false, false, 16, 32);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_ref_only_dynamic) {
this->test_dynamic_quantization(false, true, 16, 33, 33);
}

TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_unaligned_dynamic) {
this->test_dynamic_quantization(false, true, 16, 32);
TEST_F(dynamic_quantization_gpu_tests, simple_quantizing_invalid) {
this->test_dynamic_quantization(false, false, 16, 7, 7);
}

0 comments on commit 16ae143

Please sign in to comment.