Forum Discussion
New ContributorHow to solve DDR_core_clock and DDR_phy_clock setup time violate ?
I am platform designer, not so familiar with timing.
I use QSYS to have a FPGA design, however, I always have some timing violation between DDR core clock and phy clock.
but I can't add pipeline bridge between them, because it's all EMIFIP generated logic.
Can anyone give me a hint?
3 Replies
Deshi_IntelRegular Contributor
Hi HHo2,
DDR core clock is still running on FPGA soft logic. Therefore, you may still be able to add pipeline to it.
You can open QSYS interconnect and check the timing failing path whether able to add pipeline or not. Example as below per attachment.
Thanks.
Regards,
dlim
HHo2Hi @dlim thanks for answer
I have check what you suggest, however, my situation is, the falling path is {from emif|ddr_inst|hmc.amm.data to emif|ddr_inst|phy_register}
this path is generated during compile, so I can't see it in qsys, all I can see is just a bus to EMIF IP.
It's from ddr_inst to IO pad phy_register.
I have try over constraint
if {$::quartus(nameofexecutable) != "quartus_sta"} {set_max_delay -to {board_inst|acl_ddr4?_s10|acl_ddr4?_s10_emif_s10_0|acl_ddr4?_s10_emif_s10_0|arch|arch_inst|io_tiles_wrap_inst|io_tiles_inst|tile_gen[?].lane_gen[?].lane_inst|lane_inst~phy_reg1} 3.000ns }
but still won't help.
- Deshi_Intel
Hi HHo2,
Can you check below EMIF spec estimator link to ensure your DDR4 IP configuration is within the spec ?
Once you confirmed you are using valid DDR4 IP setting then you can easily generate DDR4 example design from DDR4 IP using same exact setting. After that, you refer to the example design to get better clue on how to close timing.
Thanks.
Regards,
dlim
