ROCm TF 1.3 resnet batch 256 hang inside MIOpen bwd BN

MIOpen(HIP): miopenStatus_t miopenBatchNormalizationBackward(miopenHandle_t, miopenBatchNormMode_t, const void *, const void *, const void *, const void *, const miopenTensorDescriptor_t, const void *, const miopenTensorDescriptor_t, const void *, const miopenTensorDescriptor_t, void *, const miopenTensorDescriptor_t, const void *, void *, void *, double, const void *, const void *){
bn_mode = 1
xDesc = 256, 16, 32, 32
x = 0x6231a2200
dyDesc = 256, 16, 32, 32
dy = 0x6151a2200
dxDesc = 256, 16, 32, 32
dx = 0x60d1a2200
bnScaleBiasDiffDesc = 1, 16, 1, 1
bnScale = 0x60980db00
resultBnScaleDiff = 0x609b9ef00
resultBnBiasDiff = 0x60a1a1a00
epsilon = 1e-05
savedMean = 0x609ba0700
savedInvVariance = 0x609ba0800
}
<<hip-api tid:5.3403 5.3403 hipLaunchKernel 'BatchNormBwdSpatial' gridDim:{16,1024,1} groupDim:{1,1024,1} sharedMem:+0 stream:0.1 @328820053308573