Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
13 years ago

a very nooby question

i was reading altera architecture; ok i may get something wrong but.. there are LAB-s right, and inside them are LE-s. inside LE-s are 2 main things LUT(lookup table) and Register. and all the stuff happens there. but i don't understand where actually are the transistor themselves. these CMOS guys. are they inside LUT? and how many are they? or one LE may transform into one of the Logic Gates...? or LE has many logic gates...? where is the analog world itself...?

PS don't get bully on me i'm a cub but soon i'll become a lion :))

4 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    13 years ago

    I think even to most FPGA lions, it's still pixie dust inside the LUTs and Registers. So yes, there are gates that make up these structures, but we don't know how many or exactly how they're put together, and to a certain degree don't care since there's nothing you can do with that information. The LUT/reg is really the DNA building block that you can target/manipulate. (And more in more designs don't look at that, since designs are getting so big, it's like assembly when writing a large program)

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    13 years ago

    Ilik,

    Think of it like this.

    There are Protons, Neutrons, and Electrons - and they are what make up Atoms. (transistors -> cmos P-channel, n-channel, diordes, etc).

    These Atoms can then be grouped together to make Compounds (Water - brass - etc)

    (LE's, F/F, Adders, RAM's, etc). From that we get all sorts of useful things like Cars, Boats, Houses - but even that is tiny compared to the solor system and the Galaxy...

    How do we get from here to there?

    The designers at Altera use 'engineering tools' to create "pictures" of layers of silicon to stack up upon one another to build the die that we know called an FPGA. That is where the analog stuff (in saturation mode) exists. We do not get to work directly with that level of stuff. It is what "makes" the FPGA internal structures. These designers give that stuff (called Masks) to a company they have been using for a very long time called TSMC who makes the parts (in a place called a FAB). These are the parts that we get to use - sort of. First they have to be tested, then inserted into packages by another company so we can easily solder them to the boards we use when designing systems (the PCB).

    Where do 'we' come in?

    What we get to play with is inserting a programming file (called a .sof or .pof file) which sets these internal "transistors" to open or closed such that the LE's, the F/F's, etc are realized & interconnected to make our design (on the fabric).

    I hope this helps you see how things are made from the bottom up.

    There are many "layers" in the electronic world where one can learn the craft, choose a profession, and earn a living, be it down deep in the making of hte chips, all the way up to the using of them to build large systems. Keep asking questions, and keep learning.

    Cheers,

    Avatar
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    13 years ago

    thanks for answers. yes i know that there is one big register that keeps information how to set up circuit inside FPGA.and this information is provided by programmer file.but i am still having a feeling that all this microchip stuff looks like something massonic you know? :D in a sense that only few men have grabbed this huge industry in their hands, and all the "know how". and they do not have a will to give it out.it's like an oil industry they give you books and university study but they never tell you how the hell to spot oil and set up a oil mining on it :) of course it has huge connection to national security, you must always be sure, that enemy will never have high and compact processing strength on it's war machine(missile,tank targeting system etc.all these hardware languages do give us superior abilities to command massive amount of elements inside circuitry;but these languages also serve as a shield, to keep an engineer as far as possible from "insight technology"of an FPGA. i mean, command "reg" for example. you just type reg. and thats it. job is done; no one knows how :) . so i guess these reasons said above are the main reasons why they will NEVER tell us the truth about where the transistors are :D and it always will rise a question of questions: how big is an FPGA productivity? i mean, what will you answer - 100 LAB-s?? 100 LE-s? what does these names mean after all; without knowing how many transistors are in it...

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    13 years ago

    --- Quote Start ---

    looks like something massonic you know? :D .

    --- Quote End ---

    LOL:D , wel there is some useful books explains the fabric of the fpga you should have studied them if you are an electronics engineer, however as Rysc said you will never need to know anything about this layer while writing HDL code, HOWEVER there is something important which is the LE it self can operate in two modes : Normal mode and Arithmetic mode, the second is superior for adder,counter .... purposes while the Normal is better for combinational circuits.

    search for them and you will find more about it