- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi all,
I have a question about the bank usage for DDR3 memory in the Arria 10. I'm using the following hardware: Arria 10: 10AX115R3F40E2SG DDR3 memory: Micron MT41J128M16HA-187E:D Quartus Prime 17.0.0 The present design of my board (new board) tries to use bank 3G and 3H but when I tried to do "Fitter" with these banks by using the example code generated by the Quartus, the Fitter was failed. Then, when I did the same thing with bank 2L and 2K, it worked well. In the readme file (Section 6 Pin location), one bank for address and one of DQ/DQS/DM (x8), where most of pins are in this bank, and the other bank just for the remaining DQ/DQS/DM (x8). So I used 3H (48 pins) for the former and 3G for the latter because 3G has only 47 pins (there is no 0 index pin in this bank.) My question is cannot we use 3G and 3H for DDR3 in the above Arria 10 with DDR3? If there is a way to avoid this issue, I'd like to know it. regards, JunichiLink Copied
4 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Tried to remove all location assignment for the DQ/DQS/DM (x8) in bank 3G.
The only force one DQS pin with constraint as below see if you can fit the DQ/DQS/DM (x8) in bank 3G. set_location_assignment IOBANK_3G -to <dqs pin name> (This message was posted on behalf of Intel Corporation)- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thank for your answer.
Concerning your suggestion, what does happen for the removed pins? BTW, after some iterations, I managed it. What I did is to use bank 3H for both DQ but when I used 1-11 and 12-23, the Fitter did not pass but when I used 12-23 and 36-47, the Fitter finished w/o error. regards, Junichi- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
For those remove pin, quartus will auto fit it to a location that meeting the fitting requirement.
Yes, your observation is expected. You have to use I/O lane that adjacent with each other. See Figure 93. I/O Bank Sharing of EMIF handbook https://www.altera.com/documentation/hco1416493470528.html#hco1416492629263 (This message was posted on behalf of Intel Corporation)- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thank for your answer.
Concerning the automatic pin assignment, I tried it and found that this is very useful. Concerning Fig 93, I read it and what I said, that is, [1] 1-11 and 12-23 (bad case) [2] 12-23 and 36-47 (good case) both of them may satisfy this condition, I think. If you are talking the "adjacent" lane, [2] may not satisfy the condition but it's OK in the Fitter. BTW, concerning the address (a[13:0]), we have to use only the specific pins. I mean that can't we swap them among address pins? regards, Junichi
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page