03-02-2020 07:35 AM
My company has an old project that is currently working on HW-V4-ML405-UNI-G board (ML405/406). Now this board is long obsolete. From where we can buy this board? Old or new , doesn't matter(but should be working). Quantities from 1 to 20 will be useful (we will buy even more , if available).
Thanks in advance and really looking forward for some good suggestions.
03-18-2020 03:47 AM
Hello @tahirsengine1 ,
Please note that kit is no longer manufactured, and based on the many Xilinx distributors websites (you can find these by simply doing a google search for 'ml405 evaluation board'), these are no longer available to be purchased.
For the alternatives, please contact your local Xilinx FAE/Authorized Distributor and they will be able to advice on this then.
Hope this helps.
03-18-2020 03:57 AM
@tahirsengine1 How much are you willing to spend? The schematics (and even the PCB layout drawings) seem to be available, and the Virtex 4 is also available, so technically I guess you could make your own. If you get in touch with your Xilinx FAE, they might even be willing to supply the original PCB project (on the basis that it has no value to Xilinx any more, but if they give it to you then you'll buy a bunch of Virtex 4s from them).
03-18-2020 04:20 AM
My understanding is that a dev board is to keep nervous developers busy tinkering with things while the final hardware is developed. Buying them in "production" numbers even if long ago obsolete doesn't make any sense to me... but last time I saw a bit of that was in the 90s... nothing since then.
03-18-2020 04:52 AM
@u4223374 Nice suggestion. Actually we need 15 at max. This board will be used as the a prototyping product for some initial research on an ASIC. But customer seems to be in hurry and also nobody will pay for cost of board manufacturing, components and other things.
So, here goes another dream unfulfilled.
03-18-2020 05:20 AM
@tahirsengine1 Do you need the PowerPC blocks? If not, Altium's DB36 Virtex-4 board seems to be not-technically-obsolete and probably represents the minimum-work (and minimum-time) path to getting the old code running (probably just changing pins and re-running implementation).
With that said, if the code isn't heavily optimized for that specific chip, you can keep the same toolchain and use devices up to a Kintex 7 - which might be a shorter lead time option if you can spare the manpower to update the project (the DB36 seems to generally be a 6 - 10 week lead time; Kintex 7 boards can be had in a day or two).
03-18-2020 05:39 AM
Research ASIC with Virtex-4 ????????????
Why? Why not a beautiful Virtex-7? Actually, it's quite probable than a Kintex-7 outperforms a Virtex-4, and I wouldn't be surprised if an Artix-7 speed -3 does. Now I know why do people look for intelligence in faraway galaxies...
03-18-2020 07:03 AM
Things to know about HW-V4-ML405-UNI-G
It is a University board (The UNI in the part number tells you this). It was never available to the general public because it was a price supported board that Xilinx allowed universities to purchase at a discount to support advanced design classes. Because of its complexity (Power PC cores) it could never be considered a board that was sold to intro students, rather this was senior project, masters degree research domain. Because of that the sales were small so there will be no vast supplies of them.
The board disappeared from the mainstream development after Zynq appeared, so any stocks will be at least that old. (2012)
You might find a couple on ebay, maybe, but that is about it. No doubt there are a few sitting in closets of old electronics labs but how to get them to appear?
The talk in this discussions about V7s or Artix7 or what ever misses the point of Power PCs. They would be a difficult beast to make work in even a large V7. IBM did release synthesize source code for PPC but that is a lot of work to bring back a chip. If you must run the V4FX parts the fastest path would be to spin a board.
However I would thing a quicker and more supportable path to a completed project would be to get an Ultra96 and port the project to a modern part that will run circles around a V4FX.
03-18-2020 07:20 AM
We are in 2020. PowerPC was created in 1991. A good thing back in its glorious days, but nowadays highly superseded by ARM. If we are talking about research and making ASICs, and insisting on the dinosauric PowerPC, is really ignoring what R&D is about.
03-18-2020 09:29 AM
Research in University is to get to know what is yet unknown. Research in the industry is to produce what is not produced yet (in the wide sense, like producing the same at a lower cost, etc).
It's the same. To achieve what others haven't achieved you need to use what others haven't used, otherwise, you will probably achieve something already existing (assuming you don't fail on the way)
If PowerPC is not in the latest FPGA families, couldn't you suggest your customer stop and consider there might be a powerful reason for that?