Int8 calibration on tensorrt 5 is much slower than tensorrt 4

Hi, after upgrading from tensorrt4 to tensorrt5, we notice int8 calibration on the same model is an order of magnitude slower than before. Is it generally expected that tensorrt 5 has slower calibration process?

can you share a small repro that demonstrates teh slow calibration between TRT4 vs TRT5 you are seeing? thanks.