Topic: List your potential leads for scrounging cheap or free FPGAs (Read 4357 times)
That was more to give an idea of where to look. Maybe there are lesser FPGAs in current products for those markets.
Just taking a look at the kintex-7 page.
http://www.xilinx.com/products/silicon-devices/fpga/kintex-7/index.htm
It says they are used in LTE wireless networks, 3D/HD displays, broadcast video on demand, ultra sound, optic fibre networks.
Look I'm sorry to be an ass, but you don't get it.http://www.xilinx.com/products/silicon-devices/fpga/kintex-7/index.htm
It says they are used in LTE wireless networks, 3D/HD displays, broadcast video on demand, ultra sound, optic fibre networks.
First of all, Kintex-7's aren't even available yet (AFAIK). In order for FPGAs to be recoverable, you have to find them in end of life products. Kintex-7s are so new that you will never, ever find them for recovery. Hell, most companies can't even order them yet. It's like you are arguing for looking for Ferrari Enzos in the local junkyard (e.g. waste of time).
The only thing you will find that is cheap is either Virtex-2 or Virtex-4 parts, and that's if you get really lucky. I have found some that are cheap, but again you have to pay the chip recovery fees ($250 per part), fab a PCB to mount them, populate the PCB (all surface mount parts, which is $$$), and use a JTAG uploader to program them.
All of this is well outside the path of feasibility for average people, and it's only a reasonable proposition for electronics nerds like me.
Just taking a look at the kintex-7 page.
http://www.xilinx.com/products/silicon-devices/fpga/kintex-7/index.htm
It says they are used in LTE wireless networks, 3D/HD displays, broadcast video on demand, ultra sound, optic fibre networks.
http://www.xilinx.com/products/silicon-devices/fpga/kintex-7/index.htm
It says they are used in LTE wireless networks, 3D/HD displays, broadcast video on demand, ultra sound, optic fibre networks.
A factory that makes HD Tv controller boards would be nice. They scrap shit. There's an LCD/plasma graveyard nearby but I'm guessing it's upto 50 mhash maybe at most per fpga. Those guys want 50 a board, could maybe get them down to $35 - $25. I can reball if someone can print pcb pci cards. Maybe we can scour craigslist for free or cheap broken tv's, get the model numbers and look them up. I'd chipin for membership at a website that has schematics and repair manuals.
$50 / board for 50MH would be a pretty great starting price for FPGAs considering the power savings. It might only be a 1/1 ratio, but if each chip used, say, 10W you can do 1GH/s for ~200W - I'd save nearly $450/yr/rig over what i've got
Yeah but you could try negotiating with them. Maybe $50 is the once in a blue moon customers price. The friendly dude that comes in a few times a week and has a friendly chat could get a whole pile for much lower unit cost.
You'd need 20 chips for that. How many chips could you fit on a board? I'm assuming you'd want them on a PCI board?
A factory that makes HD Tv controller boards would be nice. They scrap shit. There's an LCD/plasma graveyard nearby but I'm guessing it's upto 50 mhash maybe at most per fpga. Those guys want 50 a board, could maybe get them down to $35 - $25. I can reball if someone can print pcb pci cards. Maybe we can scour craigslist for free or cheap broken tv's, get the model numbers and look them up. I'd chipin for membership at a website that has schematics and repair manuals.
$50 / board for 50MH would be a pretty great starting price for FPGAs considering the power savings. It might only be a 1/1 ratio, but if each chip used, say, 10W you can do 1GH/s for ~200W - I'd save nearly $450/yr/rig over what i've got
Hey nema, can you try taking some pictures of the parts? I can tell you the speed in MH and we can estimate from there.
A factory that makes HD Tv controller boards would be nice. They scrap shit. There's an LCD/plasma graveyard nearby but I'm guessing it's upto 50 mhash maybe at most per fpga. Those guys want 50 a board, could maybe get them down to $35 - $25. I can reball if someone can print pcb pci cards. Maybe we can scour craigslist for free or cheap broken tv's, get the model numbers and look them up. I'd chipin for membership at a website that has schematics and repair manuals.
Well, in order to acquire them cheaply the best way to do that is consumer products. Expensive and esoteric applications like bioinformatics is almost hopeless, because there are maybe 10 institutions in the nation who would buy such equipment. The focus needs to be on relatively cheap commercial off-the-shelf products for recovery.
The best candidates IMO is networking equipment from 2000 onward. They frequently use FPGAs. However, the problem is like I stated above: they use small ones which are essentially a waste of time. It's also really hit or miss on which brands and which models contain them, unless someone is willing to compile a big list.
The best candidates IMO is networking equipment from 2000 onward. They frequently use FPGAs. However, the problem is like I stated above: they use small ones which are essentially a waste of time. It's also really hit or miss on which brands and which models contain them, unless someone is willing to compile a big list.
Sorry its also posting an offtopic post in my own thread.
Maybe we could start building a list of devices that use FPGAs. Even if they are fairly low powered ones that could still be good for learning with or even setting up armies of minion workers.
From the wikipedia article:
Applications of FPGAs include digital signal processing, software-defined radio, aerospace and defense systems, ASIC prototyping, medical imaging, computer vision, speech recognition, cryptography, bioinformatics, computer hardware emulation, radio astronomy, metal detection and a growing range of other areas.
http://en.wikipedia.org/wiki/Field-programmable_gate_array#Applications
Bioinformatics would be one interesting area. They may use FPGA's to sequence genomes or something like that but they are always coming up with a faster way of doing it. I wonder what would happen to their old FPGA's?
Maybe we could start building a list of devices that use FPGAs. Even if they are fairly low powered ones that could still be good for learning with or even setting up armies of minion workers.
From the wikipedia article:
Applications of FPGAs include digital signal processing, software-defined radio, aerospace and defense systems, ASIC prototyping, medical imaging, computer vision, speech recognition, cryptography, bioinformatics, computer hardware emulation, radio astronomy, metal detection and a growing range of other areas.
http://en.wikipedia.org/wiki/Field-programmable_gate_array#Applications
Bioinformatics would be one interesting area. They may use FPGA's to sequence genomes or something like that but they are always coming up with a faster way of doing it. I wonder what would happen to their old FPGA's?
Umm, this doesn't help us. That paper is about algorithms for routing closure, which is not our concern. It's also a shitty paper, because they don't present any experimental results.
Just a suggestion but what about DNA computing?
http://www.waset.org/journals/waset/v36/v36-43.pdf
http://www.waset.org/journals/waset/v36/v36-43.pdf
Oh one more thing.
The only FPGAs worth using at all for mining are the really big ones. This is problematic, because they are much rarer to find in commercial products than the smaller ones. Small FPGAs can only do 5-25 MH each (depending on a lot of factors), so even if you had lots of them its a PITA.
I live in Silicon Valley, and I was considering dumpster diving for some of these parts. If anyone lives nearby and is interested in this, send me PM.
The only FPGAs worth using at all for mining are the really big ones. This is problematic, because they are much rarer to find in commercial products than the smaller ones. Small FPGAs can only do 5-25 MH each (depending on a lot of factors), so even if you had lots of them its a PITA.
I live in Silicon Valley, and I was considering dumpster diving for some of these parts. If anyone lives nearby and is interested in this, send me PM.
Let me first say I'm glad someone made this thread. I've been working on this idea for quite awhile, but I've had no success. Here are my problems:
1. Most devices which use FPGAs still have high intrinsic value. There are a number of high speed PCIe cards which use FPGAs, but they still can cost upwards of $200/ea.
2. Often times PCB recyclers have boards with parts on them, but the pricing is far from "free".
3. Even if you can get a board for close to nothing, you still have to mount the FPGAs somewhere. There is no off the shelf PCB mounting options I'm aware of for FPGAs. The ones I am aware of are expensive as well. Even the open-source FPGA design on this forum is going to cost significant $$$ when it's done, and today it's designed only for a Spartan-6 LX150.
4. The other problem is BGA removal. Removing BGA parts and remounting them is also $$$. A local place quoted me around $250 per part.
If anyone has ideas on how to overcome these problems, I'd love to know.
1. Most devices which use FPGAs still have high intrinsic value. There are a number of high speed PCIe cards which use FPGAs, but they still can cost upwards of $200/ea.
2. Often times PCB recyclers have boards with parts on them, but the pricing is far from "free".
3. Even if you can get a board for close to nothing, you still have to mount the FPGAs somewhere. There is no off the shelf PCB mounting options I'm aware of for FPGAs. The ones I am aware of are expensive as well. Even the open-source FPGA design on this forum is going to cost significant $$$ when it's done, and today it's designed only for a Spartan-6 LX150.
4. The other problem is BGA removal. Removing BGA parts and remounting them is also $$$. A local place quoted me around $250 per part.
If anyone has ideas on how to overcome these problems, I'd love to know.
FPGAs are used in many products and sometimes those devices get EOL'd or thrown own because of similar problems and that make them no longer fit for their purpose possibly in medical or scientific equipment, military industrial use or even video transforming.
I found this interesting site where back in 2007 this guy was cracking SHA with a custom rig "built out of surplus Grass Valley HD video transform boards, scrapped by GV because of defects."
He writes that "The boards contain 15 Virtex-II Pro (XC2VP20) FPGAs in 3 identical sets of 5 (here called "channels"). Each channel also owns a Spartan-II (XC2S50) FPGA that was originally used as a control chip, and a DSP (ADSP21160M) which probably calculated transform parameters. There is also a shared XC2S50 chip, which is not used in this application, just like the DSPs. The clock distribution tree unfortunately contains 2 domains, which means the 39MHz channel clock had to be distributed from chip to chip, using internal Virtex-II DCMs to clean it up."
http://nsa.unaligned.org/index.php
I could imagine that there would be a lot of this stuff gathering dust in Hollywood.
So I know that the BTC profitability is getting miserably low now and its not worth buying new GPUs and electrictiy prices are going up so fast it might not even be cost effecitve to run your current gear. So if you wanted to build a zero budget rig that used FPGAs where would you start?
I found this interesting site where back in 2007 this guy was cracking SHA with a custom rig "built out of surplus Grass Valley HD video transform boards, scrapped by GV because of defects."
He writes that "The boards contain 15 Virtex-II Pro (XC2VP20) FPGAs in 3 identical sets of 5 (here called "channels"). Each channel also owns a Spartan-II (XC2S50) FPGA that was originally used as a control chip, and a DSP (ADSP21160M) which probably calculated transform parameters. There is also a shared XC2S50 chip, which is not used in this application, just like the DSPs. The clock distribution tree unfortunately contains 2 domains, which means the 39MHz channel clock had to be distributed from chip to chip, using internal Virtex-II DCMs to clean it up."
http://nsa.unaligned.org/index.php
I could imagine that there would be a lot of this stuff gathering dust in Hollywood.
So I know that the BTC profitability is getting miserably low now and its not worth buying new GPUs and electrictiy prices are going up so fast it might not even be cost effecitve to run your current gear. So if you wanted to build a zero budget rig that used FPGAs where would you start?
Jump to: