Regarding XC6216

My first answer is: look it up on google. You get 747 hits! Question: Do you need any more? The parts do not exist anymore, so why do you need the data sheet?

Peter Alfke, Xil>

Hi Peter,

there are people who would like to have XC2616, I actually tried to buy XC6216 development board at ebay (Ray Andraka was selling) but I was too late :(

as much as I know XC6216 is the only Xilinx device that has full public bitstream documentation (or had at least) also for XC6216 there is free JERC6K JBits system inclusive Java sources still downloadable from xilinx download site :)

so there are historic and educational reasons to have XC6216, if I would have that board would use it for P&R tool testing, as bitstream is known and array is relativly simple.

antti

I still have a bag of XC6216s. I really wish Xilinx would publish the bitstream for the V2Pro. I don't get the logic any more the Pro has DES encryption so if you want to keep your design secret you can.

Maybe it is time to start an open the bitstream project? Anyone got any good ideas?

Steve

Could you use the Virtual FPGA project on a modern device?

Of course that is just pages which have the 2 pieces of text "6216" and "data sheet" (or wven "data" and "sheet" separate) on them. Of course that does not have to be an data sheet for an 6216.

I once got a copy from:

(at least my bookmarks say that, in Jan 2001).

-- Neil Franklin, snipped-for-privacy@franklin.ch.remove

Hacker, Unix Guru, El Eng HTL/BSc, Programmer, Archer, Blacksmith

- hardware runs the world, software controls the hardware code generates the software, have you coded today?

Or even just the standard V2 if they want to keep the top of the line secret (or have some contract with the PPC hard core people).

That reason is now dead. But they still have an whole large bag of others. Such crappy things like "binary compatibility hinders us innovating". Of course any "binary compatible" market would exclude all not compatible competitors, so maximal innovation simply does not matter in that market. And yes this means 2 series (binary and power), so perhaps take the older V1 designs for this (OK, that loses DES)?

Get Xilinx to cooperate with not immediately profit generating (small) user demands?

Sounds unlikely given present corporate un-culture. It is large corporations serving (mainly) large-account large corporations, who could not care for anything other than "what we do today n% more efficient". Damn the small guys who want to do something different. Mass manufacturing industry sucks.

Just look how long it take just to get them to deliver their tools for Linux.

Hmmm, perhaps aim at someone else who wants to expand their lesser market share?

Ask Altera how many design wins they got from their far earlier Linux tool availability exploiting Xilinxes stubbornness. Perhaps they will see public bitstream as an other step in regaining marketshare.

Or perhaps ask the upcoming 3rd guy Lattice if they want an hot "accelerator" feature into the market (gain mind share with all the new entering people). OTOH their "registrate just to see the data sheets" policy does not point to enlightened people working there.

Or perhaps Actel would like more sales for their ProASIC parts? End of their underdog status? (I doubt that write-once Antifuse looks good for the experiment liking sort of people that open bitstreams would attract).

Publicity campaign among open source hackers? Get then to join an petition at Xilinx (or Altera or Lattice or Actel)? Slashdot?

Everyone of them would understand the implication of processor instruction sets being closed (= no Linux). Even just 3D graphic cards and some USB devices being secret is massively annoying. So the promise of "make your own chip" (= get rid of the crappy pc architecture, an future OpenPC) and the point about the parts for that existing but being closed should be easy to get across.

Or some bluesky ideas:

1. Get own (2nd hand?) chip fabrication line and make an OpenFPGA, with its bitstream (and chip design?) public? :-) Or (better?) even just be an fabless house and make the FPGA as open-source portable VHDL or EDIF and then let anyone download and have a batch of FPGAs made at any ASIC house.

Could even fix some annoying features of todays commercial devices, such as making an SRAM FPGA (so least demands on ASIC house), but with separate battery driven VCCCONF, so it needs no config device/boottime (just like Flash or Antifuse based FPGAs but without their process/size/speed problems).

Of course damn patents make it illegal to sell chips implementing some of the newer features (but the basic 4LUT/FF/Mux logic cell and PIP based routing should be unencumbered until such a chip is ready for sales). And just design and "produce for own use" may even make that moot for some users, so make encumbered extra features compile time selectable.

1. Or perhaps better ditch FPGAs (a method to get custom chips out of mass manufacturing an single device), and invent some machine that looks like an laser printer, but prints ASICs? Same as laser printers do the same job as offset printing machines, but a lot smaller and cheaper (but far lower output volume and at lower quality).

Perhaps based on ion implanting? Hmm, I need more semiconductor physics knowledge.

-- Neil Franklin, snipped-for-privacy@franklin.ch.remove

Hacker, Unix Guru, El Eng HTL/BSc, Programmer, Archer, Blacksmith

- hardware runs the world, software controls the hardware code generates the software, have you coded today?

If you have experience with google you know to put quotation marks around : "XC6216 datasheet", and, voila, you get just one hit from a spanish website:

Its already started

the main website is not yet open, but work is in progress :) could you consider donating that bag of XC6216's to openchip project? there are things you could get in return.

please contact me as snipped-for-privacy@openchip.org

antti

PS there isnt much to show yet

this is MPGA SVG viewer, similar could be made with no big effort for XC6216 it would instantly spit out bitstream from the web based editor interface.

sure a fitter tool would be needed for real desing fitting but thats also work in progress :)

Hi Peter, As an experienced 'Googler' ;-) I found this:-

It's a scalable 2 V, 20 GHz FPGA using SiGe HBT BiCMOS technology based on the XC6k stuff. Any idea of lead time, pricing? Just kidding, are you guys involved in this? A small (say a few thousand gates), superfast CML FPGA would be very useful, probably used alongside a much bigger, standard part. cheers, Syms. p.s. I don't really expect an answer, guess this stuff would be company confidential anyway.

Of course I know how to use google. My posing an link (which I found via google) should have pointed that out. :-)

And that may require typing XC6200 (Xilinx names their data sheets by family) or "data sheet" with an blank in between. And adding pdf at the end reduces clutter also. Still gives 8 pages of hits, most of them clutter (PDFed university papers which refer to the paper version of the data sheet), but with above VCC site still as first.

But why about my google experiences, as I evidently ready have already found one?

What I was critisizing was your "You get 747 hits!" line, and its style. That number lookes like the result of an typical uninformed searcher, who doesn't know the difference hits vs usefull links. That there are 747 actual usefull hits is something I would not even take into consideration, seeing how few I found 2.5 years ago[1].

[1] Doing historical research into FPGAs. Managed to find CAL1024 and XC6200. But XC2000, the ancester of all, I still have not found anywhere, grrr. And yes, an "obsolete/EOLed products" page on xilinx.com would be nice. Actel has everything back to ACT1 still on their site. That is nice for tracing product evolution, and so designers experience in what needed changing. I suppose we see here the collision of industrial "bury the past mistakes as fast as possible" and academic "publish and learn from past mistakes" style of thinking.

Deriding people (and making snotty[2] google remarks fits in that category) in not really fitting with the professional image you are trying to (and usually manage to) hold up.

[2] non-snotty is simply making an remark "you can google for that", anything that does not say "I have looked but I am not telling you".

Either be helpfull (cut&paste the first link, after claiming to have found 747 fitting links (or is the 747 just a random invented number?). Or simply not search and make no remark (costs you the least time) and leave googling or tips or lecturing on using google to others.

I should add, that above criticism in not Peter Alfke specific, nor c.a.f specific, but rather a general observation on reactions to "find something" threads on Usenet. Either be helpfull and search, or preach "go searching yourself" and then don't do (or pretend to do) an search just to spite the original poster.

If you intended the "747" as an joke, then it definitely did not come accross. Lacking intonation, Usenet gets interpreted according to the readers estimates of the senders views. And with you known "anti old chips" and "anti open chips" attitide, such an remark comes accross as snotty.

Only one? Hmm, you searched XC6216, while I looked for XC6200 (which I expected the data sheet to have as title (it did), so I found a few more. But nowhere near 747.

Oh, the 1.10 version (april 1997), I only had 1.07 (october 1996). So at least some good came out of giving you an slap. :-)

-- Neil Franklin, snipped-for-privacy@franklin.ch.remove

Hacker, Unix Guru, El Eng HTL/BSc, Programmer, Archer, Blacksmith

- hardware runs the world, software controls the hardware code generates the software, have you coded today?

Hi Neil, Congratulations on the funniest post I've seen on CAF for ages!! Your parody of the precious "Hacker, Unix Guru" getting upset because someone dared offer advice on how to google was priceless! Thanks for brightening my day, and keep up the good work! Meant in good humour, Syms.