Altera_Forum
Honored Contributor
13 years agoprocess flow for generating NIOS + TSE
Hi,
I need to generate an TSE solution to Cyclone IV GX, where the datapath is: Memory (internal in the beginning, DDR2 later) => SGDMA => TSE => PCS => PMA (GX tranceiver) => SFP module (copper and/or fibre). And the reverse. I have used several days trying to put this up, with little success. So far I have done the following: Generated QSYS system with minimalistic setup: NIOS+internalMEM+2*SGDMA+TSE+PLL+sysid+jtaguart. The QSYS generation works fine, with no errors. Then I have instantiated the system into a VHDL top-level file (FPGA pin names differ from QSYS pin names). Then I tried to understand the TSE UG for running TCL for the system, with failure (it tries to assign IO standard to default pins:TXP & RXP). I manually reassigned the IO standards to correct pins (fpga top-level pins). Then I included the QSYS generated SDC file into Quartus project (Assignments => settings => TimeQuest Timing Analyzer => File is added here). Then I created a small SDC file describing my clocks (50 MHz and 125 MHz), and deriving pll clocks. Ten I ran the Quartus "Start Compilation". Synthesis seems ok, but Timing Analysis shows lots of errors (Slack starting from -3.211 ns), tens of them. The TSE UG (June 2012) eg says in page 141 (8-1): "# Name the clocks that will be coming into the tse core named chaned from top level". I do not understand what Altera is trying to say with this. Also the SDC file gets rewritten every time I generate a new QSYS system. If I need to edit it (I don't know) I need to re-edit it after every QSYS generate command, right? *** So does anybody know what is the correct process flow to do this? Would be highly appreciated. BR, -Topi