Hello,
I am sending a small burst of data (32 Bytes) from the uP to the FPGA Virtex7 through PCIe GEN3 x8.
With an AMD processor, I see that the FPGA is receiving 1 TLP of 2x32bytes word (4 ns between each word, the first word has the address + 16 Bytes and the second only the last 16 Bytes).
With a XEON processor, I see 2 TLPs with a latency of 90 ns between them (the first TLPs has address+16 data and the second Address+16 data).
AMD processor is quite old (4 years) and XEON is very new.
Does someone know why I have this poor performances with XEON processor, and does I miss any FPGA IP configuration?
I am on LINUX and I use exactly the same driver for the 2 cases.
Thanks