Spartan III place fails

- B
- B?rge Strand
  
  Contact options for registered users
posted
19 years ago

Wed, Jan 26, 2005 8:29 AM

Trying to port my latest design from a SpartanIIe-300 to a SpartanIII-200 I get the following warning:

"WARNING: Place 119 - Unable to find location. SLICEM component not placed".

This leads to : "ERROR: Place 120 - There were not enough sites to place all selected components."

Looking at the device utilization summary, it reports that 678 of 960 SLICEMs are used.

What can I do to make this problem go away? I'm using an FPGA development board, so it is a little hard to assign pins differently. Most of the SLICEM warnings are in regard to an internal FIFO. I have regenerated this core without things getting any better.

Greetings,

Børge Strand

- J
- John Adair
  
  Contact options for registered users
Vote on answer
posted
19 years ago

Wed, Jan 26, 2005 11:13 AM

One major difference between Spartan-IIE and Spartan-3 is that Spartan-3 only supports 1 out of the 2 LUTs within a Slice as a memory element. Your problem is probably due to either using a macro element that has a fixed relational structure or constraints that relatively lock memory elements. I would suggest rebuilding the macro for Spartan-3, or if using constraints for this structure, remove the Slice location constraints.

-- John Adair Enterpoint Ltd. - Home of Broaddown2. The Ultimate Spartan3 Development Board.

formatting link

- A
- Ansgar Bambynek
  
  Contact options for registered users
Vote on answer
posted
19 years ago

Wed, Jan 26, 2005 11:49 AM

Hi

I had a similar problem when using a black box (ROM) during synthesis and an edif netlist for the black box during P&R. The ROM was build for a different target device (virtex II instead of Spartan III)

When I parsed the edif netlist I could see placement directives which could not be solved for the spartan III

So my suggestion is to either recreate this black box for the required device or remove all placement directives from the edif.

HTH

Ansgar

-- Attention please, reply address is invalid, please remove "_xxx_" ro reply "B?rge Strand" schrieb im Newsbeitrag news: snipped-for-privacy@posting.google.com...