diff --git a/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_xdl_cshuffle_v3.hpp b/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_xdl_cshuffle_v3.hpp index 00adc1fa07..f6ea23a1e7 100644 --- a/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_xdl_cshuffle_v3.hpp +++ b/include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_xdl_cshuffle_v3.hpp @@ -53,8 +53,7 @@ __global__ void const ComputePtrOffsetOfBatch compute_ptr_offset_of_batch, const index_t num_k_per_block) { -#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx908__) || defined(__gfx90a__) || \ - defined(__gfx94__)) +#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx9__)) const index_t g_idx = __builtin_amdgcn_readfirstlane(blockIdx.z); const index_t k_idx = __builtin_amdgcn_readfirstlane(blockIdx.y * num_k_per_block); @@ -82,7 +81,7 @@ __global__ void k_idx); #else ignore = karg; -#endif // end of if (defined(__gfx908__) || defined(__gfx90a__)) +#endif // end of if (defined(__gfx9__) } template