Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
11/11/06 18:48
Modified:
  11/11/06 18:49

Read: times


 
#127787 - What\'s important is the entry effort
Responding to: ???'s previous message
A typical ~100-component one-sheet schematic takes hundreds and hundreds of pages of HDL just to describe it. Saving trees is a sufficient justifcation for using the schematic. That doesn't mean the design is idiotically simple.

The fact that anyone who went to engineering school when one had to be able to read, write, and spell, can read and correctly interpret schematics, while that it takes a roomful of HDL experts weeks just to agree on what a page of HDL says, is also sufficient justification. If the typical HDL user knew how much more efficient his design entry would be if he simply put in a symbol for the element he had to type up for half a day, he'd go back to schematics.

My main interest in continuing to use schematic entry is in making the review process accessible to the people who pay for it. Any engineer, even mechanical or civil engineers, can easily understand and interpret schematic diagrams, while HDL experts have difficulty even in interpreting their own work once they've been away from it a week or longer.

Reviews of HDL-entered designs take weeks and cost hundreds of thousands of dollars, while schematic reviews of the same circuitry (we've tested this notion) take half a day and cost a couple of thousand if one includes the price of lunch. It's unacceptable to enter a design in HDL and subsequently translate it to understandable schematics for review, as it means the client reviews/accepts one thing and pays for another.

If there were sufficiently reliable and useful tools to automate the translation, perhaps that would make the process acceptable, but, for now, it isn't.

I sometimes use VHDL to enter what becomes a block symbol in a schematic. However, that requires that I provide an extensive device-specific simulation of the HDL-entered logic and additional documentation that specifies and verifies what the functional block is designed to do. Even when the blocks are defined in HDL, in order to present a reviewable block diagram, one has to have a schematic entry package that enables one to document what's "in there" so the senior engineering managers can see what they're buying.

RE



List of 81 messages in thread
TopicAuthorDate
Getting Started With FPGAs            01/01/70 00:00      
   A Book            01/01/70 00:00      
   Proceed with caution!            01/01/70 00:00      
   This is quite a nice deveolpment board            01/01/70 00:00      
   I like this board            01/01/70 00:00      
      I\'ve got one of these and can\'t recommend it            01/01/70 00:00      
         What are you talking about            01/01/70 00:00      
            Let me explain ...            01/01/70 00:00      
   generaly speaking            01/01/70 00:00      
      Damn, Jez, you need a spell-checker!            01/01/70 00:00      
         HDL-based design needs TEST BENCHES            01/01/70 00:00      
            I use 'em all the time, but ...            01/01/70 00:00      
               If your design is so simple            01/01/70 00:00      
                  What\'s important is the entry effort            01/01/70 00:00      
                     I dunno where you get these ideas from Richard            01/01/70 00:00      
                        Can you say ModelSim?            01/01/70 00:00      
                     Wrong            01/01/70 00:00      
                        Well, the schematic needs a little work            01/01/70 00:00      
                           more work?            01/01/70 00:00      
                              a few points ...            01/01/70 00:00      
                           ...            01/01/70 00:00      
                     HDL vs Schematics, take 1E6            01/01/70 00:00      
                        Thta may be great comfort to you ...            01/01/70 00:00      
                           Schematics? You're kidding!            01/01/70 00:00      
                              No, nor is my customer.            01/01/70 00:00      
                              both are a 'representation of Boole'            01/01/70 00:00      
                                 Yep, you're right...            01/01/70 00:00      
                                    To Clarify            01/01/70 00:00      
                                       schematic as equivalence check            01/01/70 00:00      
                                          Manual verification impossible because ...            01/01/70 00:00      
                                             schematic verification            01/01/70 00:00      
                                       Some more clarification            01/01/70 00:00      
                                          hanging problem            01/01/70 00:00      
                                             so will your post be            01/01/70 00:00      
                                    it's not fear of the unknown, but fear of its cost            01/01/70 00:00      
                           Static Timing Analysis and those "young engineers"            01/01/70 00:00      
                              It's a sign of the times, I suppose            01/01/70 00:00      
                                 And times change            01/01/70 00:00      
                                 senior?            01/01/70 00:00      
                                    That's because you've been avoiding the subject            01/01/70 00:00      
            I agree entirely. test benches are a pain but            01/01/70 00:00      
   Update from the OP            01/01/70 00:00      
      It's true ... we see things differently            01/01/70 00:00      
      the first buuk for any such venture            01/01/70 00:00      
      FPGA boards and that Cypress book            01/01/70 00:00      
         This may answer your question(s)            01/01/70 00:00      
   some of the references in the fpga faq            01/01/70 00:00      
      Good attitude            01/01/70 00:00      
   Also remeber to look at webistes like www.xilinx.c            01/01/70 00:00      
   My recommendation and opinions            01/01/70 00:00      
      I would stay away from Virtex-II            01/01/70 00:00      
         If you're going to fiddle with the 805x core ...            01/01/70 00:00      
            Something about life-cycles            01/01/70 00:00      
               No doubt about it.            01/01/70 00:00      
                  useful work            01/01/70 00:00      
                     Well, I beg to differ.            01/01/70 00:00      
                        salient point            01/01/70 00:00      
                           There's an area where that's not necessarily true            01/01/70 00:00      
                           FPGA Editor            01/01/70 00:00      
            Not necessarily            01/01/70 00:00      
               It's still too costly            01/01/70 00:00      
   Thanks to all            01/01/70 00:00      
   Update #2 from the OP            01/01/70 00:00      
   A testbench is...            01/01/70 00:00      
      I C            01/01/70 00:00      
         test bench            01/01/70 00:00      
            Makes sense            01/01/70 00:00      
   Yeah but...            01/01/70 00:00      
   Tristate buffer with propagation delays            01/01/70 00:00      
   Schematics vs. HDL            01/01/70 00:00      
      In a nutshell            01/01/70 00:00      
         it\'s been a while            01/01/70 00:00      
      Yes, but ... and there's always a but ...            01/01/70 00:00      
   Yep you can its called gate level design            01/01/70 00:00      
   Its really nothing to do with showing off            01/01/70 00:00      
      did you paraphrase this?            01/01/70 00:00      
      Jez, It\'s not \"sea of gates\" design, you know ...            01/01/70 00:00      
         Richards, it is there philosophy            01/01/70 00:00      
         Viewlogic            01/01/70 00:00      
   it is true            01/01/70 00:00      
   Here you go Richard graphical design exploration            01/01/70 00:00      

Back to Subject List