Xilinx memories

[Marc Jeambrun]

Hello,

Does anyone know if there exists a Xilinx block memory IP core for FPGA
synthesis that includes an acknowlegment signal ACK ???

Thanks in advance.

Marc

 

[Mark Norton]

Marc Jeambrun wrote:

Hello,

Does anyone know if there exists a Xilinx block memory IP core for FPGA
synthesis that includes an acknowlegment signal ACK ???

Thanks in advance.

Marc

Just fired up coregen. The block memory FIFO has write acknowledge.
The single port block RAM has an option for "handshaking signals". I
haven't looked up the datasheet for the SPBRAM but it sounds like what
you are looking for.

Best regards,
Mark

--
==============================
Mark Norton <markn@cdvinc.com>
Concept Development, Inc.
Irvine, CA, USA

 

[Marc Jeambrun]

Thanks Mark for the write-ACK signal concerning the FIFO block memory.

In fact, I've already checked out "handshaking signals" on Xilinx single
port memories and it seems quite insufficient for acknowledging a read
or write access. For instance, according to the official support, "Ready
for DATA (RFD) is always true, except when EN is inactive" !!!!

I'm on my way to find out an alternative to this lack in the Xilinx model.

Marc



Mark Norton a écrit :

Marc Jeambrun wrote:
Hello,

Does anyone know if there exists a Xilinx block memory IP core for FPGA
synthesis that includes an acknowlegment signal ACK ???

Thanks in advance.

Marc

Just fired up coregen. The block memory FIFO has write acknowledge.
The single port block RAM has an option for "handshaking signals". I
haven't looked up the datasheet for the SPBRAM but it sounds like what
you are looking for.

Best regards,
Mark

 

[Eric Smith]

Marc Jeambrun <mjeambrun@gmail.com> writes:

Thanks Mark for the write-ACK signal concerning the FIFO block memory.

In fact, I've already checked out "handshaking signals" on Xilinx
single port memories and it seems quite insufficient for acknowledging
a read or write access. For instance, according to the official
support, "Ready for DATA (RFD) is always true, except when EN is
inactive" !!!!

I'm on my way to find out an alternative to this lack in the Xilinx model.

It's a single-cycle synchronous RAM. Every cycle that it is enabled,
it performs a read or write, and completes it that cycle. So the behavior
they spec for RFD is the only behavior that makes much sense.

If you want to pretend that the RAM is slower than single-clock, you could
add a register to delay the ACK by a cycle.