P
Panic
Guest
I have a question about how I best should implement my I/O
in an FPGA (Altera EPXA1F484C1) . Being a student, and
not having a whole lot of people to ask for advice, I have no
idea how I best should solve my problem.
I initially wanted to have 128 + a few inputs and 128 + a few
output pins, so that I could work in parallell on 128 bit blocks.
This didn't work, apparently because I had used to many pins.
Quartus (II 3.0) gave said in it's report that I had only used 258
of in all 292 available pins, but I suppose I had used all available
general purpos I/O-pins or whatever. I scaled down my design,
until I got to compile it. At that time I was using 186 pins. And
I when looked at the floorplan, every row and column I/O-pin
was in use. So I guess 186 pins is the maximum limit.
So my first question is: Why can I only use 186 pins, when Quartus
tells me that there are 292 pins available?
So I had to decise how I would send and recieve my data. And this
is my main question, my design question:
Would I be best off using say 64 pins for input and 64 pins for output,
and solve my problem that way, or could I use a bidirectional solution,
where I alternate on recieving and sending data over a 128 bits bus?
I put together a test-design wich utilizes this solution, and it compiled
fine.
I put an image of the schematics on a website, so you can take a look:
http://www.battlefield.no/bilder/EPXA1_PinTest3.gif
But is this an sound solution?! Or should I rather go for separat in- and
out-pins? Since I have no clue, I have to ask someone. And you guys
drew the shortest straw ;-)
-"Panic"
in an FPGA (Altera EPXA1F484C1) . Being a student, and
not having a whole lot of people to ask for advice, I have no
idea how I best should solve my problem.
I initially wanted to have 128 + a few inputs and 128 + a few
output pins, so that I could work in parallell on 128 bit blocks.
This didn't work, apparently because I had used to many pins.
Quartus (II 3.0) gave said in it's report that I had only used 258
of in all 292 available pins, but I suppose I had used all available
general purpos I/O-pins or whatever. I scaled down my design,
until I got to compile it. At that time I was using 186 pins. And
I when looked at the floorplan, every row and column I/O-pin
was in use. So I guess 186 pins is the maximum limit.
So my first question is: Why can I only use 186 pins, when Quartus
tells me that there are 292 pins available?
So I had to decise how I would send and recieve my data. And this
is my main question, my design question:
Would I be best off using say 64 pins for input and 64 pins for output,
and solve my problem that way, or could I use a bidirectional solution,
where I alternate on recieving and sending data over a 128 bits bus?
I put together a test-design wich utilizes this solution, and it compiled
fine.
I put an image of the schematics on a website, so you can take a look:
http://www.battlefield.no/bilder/EPXA1_PinTest3.gif
But is this an sound solution?! Or should I rather go for separat in- and
out-pins? Since I have no clue, I have to ask someone. And you guys
drew the shortest straw ;-)
-"Panic"