Topic: GPU vs FPGA help me out.. (Read 5773 times)

As far as I'm concerned, in practical terms of realistic usefulness, FPGAs are nothing but door stops if you stop bitcoin mining with them. So in my opinion, betting on FPGAs is betting 100% on bitcoin mining.

Are you quite sure? My understanding was that a key was flashed into a non-readable and non-reversable memory area on the device, which could be consulted as necessary if the bitstream was encrypted. Otherwise, the key would be ignored.

Quote from: http://forums.xilinx.com/t5/Spartan-Family-FPGAs/Spartan-6-BRAM-AES-Key-Programming/m-p/194462/highlight/true#M14434

FPGA can only be reprogrammed if the FPGA has bitstream encryption disabled, or if you have the encryption key.

A certain vendor has locked down FPGA which require an encrypted bitstream.  That makes non-Bitcoin resale value essentially $0.00.

FPGA can be used for other cryptographic purposes. Cracking TrueCrypt containers, RAR files, WEP / WPA. Or do protein folding, prime numbers.

The CIA / NSA might also want to buy your hardware if you sell it to them.

I am an Electrical and Computer Engineer, for example .....

Edit: Found more information here: https://bitcointalksearch.org/topic/m.801269
sounds like souped up breadboard, will be very useful for super geeks but it's going to be hard to make it "profitable"

FPGAs are immensely powerful devices.  Only problem is, unless you're an Electrical or Computer Engineer they're not going to do you much good.

It's more like hashcat can be reprogrammed to work with FPGAs, not the opposite.

The firmware can be reprogrammed to work with hashcat to crack passwords.

...for example...

in any case based on what I have read around the forums a lot of people have the same thoughts I do and that makes me not want to go the fpga route any time soon.  thanks again for the input guys!

I'm seeing a lot of bullish mining sentiment like I did when I first started with Bitcoin back in February of 2011. So much so that I think we might see a price pop as all these BFLs start coming in and piling on the difficulty.

That said, gigavps is right, when GPU miners talk about "at least I can sell my GPUs", keep in mind that they will be selling their GPUs far sooner than the FPGA miners will be needing to sell their FPGAs. It boggles my mind that a GPU miner can even say "If Bitcoin fails," because it goes to show that their incompetence regarding FPGA miners extends to their knowledge of Bitcoin in general. Bitcoin *cannot* go to $.01 as long as it is still a scarce resource, in fact, we'll likely never see $2 again as the difficulty rises to 2 million and higher once BFL gets all their backorders completed. It can go lower, obviously, but it will be because FPGA miners are profiting at those levels.

If trends are any indication, GPUs started in Summer 2010, were standardized as miners in Winter 2011, were heavily invested in by mining operators in Spring 2011, and have been steady sources of income from Summer 2011-Winter 2012 to even now/today if you live somewhere where power isn't extra costly.

Similarly, FPGAs were first really tested in Fall 2011, and are now being heavily invested in here in Spring 2012. My guess is that as these miners all come online, the price is going to have to rise as the GPU miners realize they need to stop selling and consider large investments in FPGA operations, this means they may start hoarding coin and selling cards. The early GPU miners that realize this are going to do better than the later ones, obviously. I think by the halving of the block reward we will see GPU miners whining about how it is no longer profitable to GPU mine. And at that point it never will be profitable again.

The argument does hold water with FPGA. But instead of selling them, you would need to have other profitable things to do with them.

If you were going to go all in whole hog with FPGA's and basically have no "fallback" resale value anyway, why wouldn't you just buy the physical coins? The argument to mine rather than just speculate was always "well, if Bitcoin goes to $0, at least I can still  sell my GPUs as GPUs". That argument doesn't hold any water with FPGA's, so why not just skip the middleman and just buy btc?

Thos eof us mining sure looked silly vs the people who just bought bitcoin at < $1 when the price shot up to $30, didn't we?

+1 gonna have to agree with you on that..

thanks for the input guys it gave me a lot more to think about in the long term

For the time being I'm sticking with GPUs in my farm.

The way I see it, FGPA's are very power efficient and could be less prone to hardware issues (FPGA's are quite new so we don't know how they will run over a couple of years.  Certainly, you probably won't be scouring eBay to find replacement fans for non-reference coolers, etc.  They might also be easier to manage from a software perspective.  You don't need one rig for every 4-5 GPUs.

Giga is probably right that any collapse will take time and GPU miners selling off their equipment will likely depress prices for a while.  For example, look at last fall when btc prices fell below $3.  It was looking bleak for many people, so they sold their hardware.  I was able to snatch up a lot of 5830's for $65-70.  Now, it's hard to get the same card for under $100 in any decent quantities.

That being said, GPUs can be easily sold/traded/repurposed and will likely retain a high residual value per dollar than FPGA's.  There's been a lot of discussion about whether bitcoin FPGA's could be repurposed for other algorithms, but it seems that they would be pretty bad at it.  Unless someone has a specific example, it seems that if mining stops being profitable those FPGA's are probably headed for an e-waste facility.

I think you need to weigh quite a few more variables if you're looking which route to go.  If you want my opinion on the sole idea that bitcoin fails, as in < 0.01$/BTC then obviously the route is GPUs.  Your target demographic for GPU resale is orders of magnitude larger than your demographic for FPGAs.  Even then, you really need to understand what FPGA's are used for and how the current (and future offerings) are designed.  For instance, the BFL single is a specific use product that contains an encrypted key.  Assuming BTC fails, you're going to have a hard time trying finding anyone to buy one.  If you look at the Icarus (which has a boat-load of GPIOs available) it's a viable development board.  That means people may actually be interesting in picking one up, but that's only if BTC fails soon enough that the Icarus hardware isn't considered legacy. 

I see someone else above has mentioned this, but my approach was to start with the best bang-for-the-buck GPUs.  Once they started earning I expanded with more GPUs.  Now that I have those operational, I'm moving onto FPGAs.  I'll expand at a rate that maintains a payoff period of 3 months or less.

I just invested quite a bit in new GPUs and as part of the end goal I ordered BFL singles. The plan is to phase out the GPUs with the singles as they arrive.

I went with GPUs to start. They are readily available, provide great MH/$ and acceptable MH/w. I'll continue to mine and expand on GPUs until we start to see the 20-28 nm based FPGAs hit the market. I'm banking on the new FPGAs being faster, even more efficient and hopefully approaching GPU MH/$ ratings. At that point I'll run the GPUs until I replace their MH/s with FPGAs and then liquidate the GPUs for a tidy profit and a FPGA expansion. I should then be able to mine quite effectively after the reward drop even if the BTC price doesn't increase much.

thanks gigavps!

EXACTLY the type of opinion I was looking for thanks.. keep em rollin!