Merge branch 'develop' into gemm_layernorm_instance

644df335 · rocking · d99640ab · 7494c1c6 · 644df335 · 644df335
Commit 644df335 authored Jan 30, 2023 by rocking
20 changed files
--- a/include/ck/utility/reduction_operator.hpp
+++ b/include/ck/utility/reduction_operator.hpp
@@ -251,27 +251,27 @@ constexpr T GetIdentityValueForInMemoryDataOperation(InMemoryDataOperationEnum o
 };
 template <InMemoryDataOperationEnum Operation, typename DataType>
-struct InMemoryDataOperatonSupportedOnDataType
+struct InMemoryDataOperationSupportedOnDataType
 {
    static constexpr bool value = false;
 };
 template <typename DataType>
-struct InMemoryDataOperatonSupportedOnDataType<InMemoryDataOperationEnum::AtomicAdd, DataType>
+struct InMemoryDataOperationSupportedOnDataType<InMemoryDataOperationEnum::AtomicAdd, DataType>
 {
    static constexpr bool value =
        is_same<DataType, float>::value || is_same<DataType, double>::value;
 };
 template <typename DataType>
-struct InMemoryDataOperatonSupportedOnDataType<InMemoryDataOperationEnum::AtomicMax, DataType>
+struct InMemoryDataOperationSupportedOnDataType<InMemoryDataOperationEnum::AtomicMax, DataType>
 {
    static constexpr bool value =
        is_same<DataType, float>::value || is_same<DataType, double>::value;
 };
 template <typename DataType>
-struct InMemoryDataOperatonSupportedOnDataType<InMemoryDataOperationEnum::Set, DataType>
+struct InMemoryDataOperationSupportedOnDataType<InMemoryDataOperationEnum::Set, DataType>
 {
    static constexpr bool value =
        is_same<DataType, float>::value || is_same<DataType, double>::value ||
@@ -280,7 +280,7 @@ struct InMemoryDataOperatonSupportedOnDataType<InMemoryDataOperationEnum::Set, D
 };
 template <typename DataType>
-struct InMemoryDataOperatonSupportedOnDataType<InMemoryDataOperationEnum::Add, DataType>
+struct InMemoryDataOperationSupportedOnDataType<InMemoryDataOperationEnum::Add, DataType>
 {
    static constexpr bool value =
        is_same<DataType, float>::value || is_same<DataType, double>::value ||

--- a/include/ck/utility/synchronization.hpp
+++ b/include/ck/utility/synchronization.hpp
@@ -18,6 +18,7 @@ __device__ void block_sync_lds()
    __syncthreads();
 #endif
 }
 __device__ void s_nop()
 {
 #if 1

--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_reduce.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_reduce.hpp
--- a/library/include/ck/library/reference_tensor_operation/cpu/reference_softmax.hpp
+++ b/library/include/ck/library/reference_tensor_operation/cpu/reference_softmax.hpp
--- a/library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp
+++ b/library/include/ck/library/tensor_operation_instance/device_operation_instance_factory.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/batched_gemm_bias_permute.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/batched_gemm_bias_permute.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/batchnorm_infer.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/batchnorm_infer.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/device_elementwise_instance.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/device_elementwise_instance.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/gemm_add_multiply.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/gemm_add_multiply.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_add.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_add.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_amax.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_amax.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_avg.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_avg.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_max.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_max.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_min.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_min.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_norm2.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_norm2.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_amax.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_amax.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_max.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_max.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_min.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_min.hpp
--- a/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16_add.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16_add.hpp