MUXCY-based multiplexers

[Jo Pletinckx]

Hi,

In the XST manuals, it is mentioned rather cryptically that these type
of muxes can be more efficient in some cases than the MUXFX versions,
especially for very large muxes. What is actually meant: very large
input muxes, or muxes with a limited number of very wide inputs? I fail
to see how they are more efficient... I've searched the web/news groups
to no avail.


Thanks for your answer,

with regards,

Jo

 

[M.Randelzhofer]

"Jo Pletinckx" <jo_pletinckx@yahoo.com> schrieb im Newsbeitrag
news:MPG.1b60cc26cf245af7989680@News.Individual.NET...

Hi,

In the XST manuals, it is mentioned rather cryptically that these type
of muxes can be more efficient in some cases than the MUXFX versions,
especially for very large muxes. What is actually meant: very large
input muxes, or muxes with a limited number of very wide inputs? I fail
to see how they are more efficient... I've searched the web/news groups
to no avail.


Thanks for your answer,

with regards,

Jo

see Xilinx techxclusives multiplexer selection from ken chapman:

http://www.xilinx.com/xlnx/xweb/xil_tx_display.jsp?sGlobalNavPick=&sSecondaryNavPick=&category=&iLanguageID=1&multPartNum=1&sTechX_ID=kc_multiplexer

MIKE

 

[Jo Pletinckx]

see Xilinx techxclusives multiplexer selection from ken chapman:

http://www.xilinx.com/xlnx/xweb/xil_tx_display.jsp?sGlobalNavPick=&sSecondaryNavPick=&category=&iLanguageID=1&multPartNum=1&sTechX_ID=kc_multiplexer

MIKE

Thanks your quick answer Mike. Unfortunately, it does not really answer
my question. In fact, the TechExclusive from Ken Chapman uses a
different approach: he only has one input per LUT, whereas the XST-
generated muxes have 2 two inputs and two selectors and function much in
the same way as Ken's multiplexer, with one selector en or dis-abling
the LUT and the other selecting one of both inputs. At two inputs per
LUT, this in roughly the same ratio as the MUXFX based multiplexers. So
when does it become more efficient the use the carry-chain?

Or is the gain located only in the implementation, as it reduces the
probability of routing congestion by spreading the inputs over a taller
column than the MUXFX multiplexer does?