| ??? 12/28/07 01:27 Modified: 12/28/07 01:33 Read: times |
#148817 - it\'s just a documentation issue Responding to: ???'s previous message |
Andy Peters said:
You said (or perhaps you quoted Xilinx), "The SP3 might perform better, but the SP2 would work where the SP3 probably (it might work at 3 volts ... but ...) wouldn't."
There was no context for this, which is why I said I didn't understand it. Then I said, "Of course the Spartan 3 family works at 3.3V. I use them with 3.3V I/O all the time" because I don't understand the context of the above statement. Just to clear things up, I was explaining why I use the SP2 rather than the "better" "faster" "cheaper" SP3. It's just so it can be verified. If I were in the cellphone or other video toy industry, which is a big and growing industry, I'd be more interested in small packages and small power supplies. I prefer to work with the multi-KW supplies that roll around on the floor and provide outputs with under 5 mV of noise, even under full load. They're hard to lift, but I don't have to do that often. To be honest, I really don't care about replacing a 5V 40-pin DIP version of an 8x51 with an FPGA version. With the exception of actually performing long-term testing on a onesie, just to verify that it actually works as you expect, it would have no practical application. However, it a "thing" can't survive long-term testing, and I doubt that there's a cellphone+network combination that can, it's just a toy. But you seem to want the sun, the moon and the stars: you want 5V-compatible outputs but you want the thing to run at blazing speeds. We've learned that 100 MHz operation of S2 parts is pretty difficult, but S3E parts do it easily. But, again, we don't use anything that requires 5V I/O. It's difficult, but the XILINX marketeers claim it's an everyday thing. It would be interesting to see what they claim a 48-bit loadable synchronous up/down counter can do, frequency-wise, without any hand place-and-route effort. I doubt it would even to go much above 40 MHz, if that high. "I'm looking for some way to find out what one can do with a logic cell aside from the obvious. They describe ways of using a single LUT as a 5-input function generator. They describe ways of doing other things, too, but most of the time they simply telly you that there's a way, but don't tell you how to use it or how to make it happen. If you don't know you can do a specific thing with their hardware, you won't attempt it, will you?"
That's when you have to experiment with the tools and see what happens. You shouldn't have to do that. The capabilities and behavior of their products, and the software is a product, should be documented such that one could read it and know, confidently, exactly what the result would be. It is a product, in part, consisting of software, however, and my experience tells me that you can seldom tell what the outcome will be when there's software involved. "Moreover, without endless hours of poring over their reports, you really won't know how it did this or that."
What other choice do you have, really? Well, there isn't an option, but there should be. What the synthesizer does should be precisely documented so it's understandable and predictable prior to being "tried." I already mentioned that you should try to see what happens when you instantiate the MUXCY.
If it was easy, everyone would do it, right? The problem is that anyone who thinks he/she can type can "do" it, but most of them do it badly. Anyways ... I give up. Masturbation, whether mental or otherwise, is a job for only one person.
-a True enough! RE |
