Hi,
I am testing the peer to peer stream bandwidth of PXIe-1082 by using Labview 2018 with an FPGA FlexRIO system. According to Labview 2018 help(https://zone.ni.com/reference/en-XX/help/371599P-01/lvfpgahelp/fpga_creating_fifos/), I use PXIe-5785 as writer and PXIe-7976R as reader, representation is I16 and write eight data to FIFO each time.
From the PXIe-1082's User Manual, 1082 has 2 GB/s per slot dedicated bandwidth. And according to the introduction of peer to peer streaming(http://www.ni.com/product-documentation/10801/zht/), if modules which connect to the same PXIe switch, the streaming path will through the switch only and the modules seems connect to the switch by a x4 PCI Express link. So I plug 5785 and 7976 into slot 3,4 (same switch) and expect bandwidth reach 2 GB/s.
However, it only reach about 902 MB/s. Does any chassis backplane setting missing or PXIe switch limitation reaching? Moreover, the bandwidth reduce to about 743 MB/s if I change to slot 2 and 4. But they seemed both stream through x4 PCI Express link. What is difference of them?