GPU or FPGA? BTC future?

[legolouman]

In relation to my other current thread: https://bitcointalk.org/index.php?topic=69250.0

I'm thinking about investing a great deal more into this all. I've been around for like 9 months. I learned about BTC just before the MyBTC ordeal (it was like mid June). I'm worried about the future of Bitcoin, from an investment standpoint. I'm still going to be here regardless of how profitable this is, but I won't throw money into a falling knife. Enough backstory blah.

With the Block Reward halving in December. Do people think GPU's are the way to go, or are the currently less profitable FPGA's going to dominate 2013?

Will the market readjust to the halved reward? What about the affect of the crazy solominer 88.6.216.9?

Thanks all

Lou

[1713578551]

1713578551

[kliffen]

I've been thinking alot about the same, and there are different things to keep in mind when you argue GPU vs FPGA.

The pros of GPU mining is
 - Resale value. The GPU market is always there, and GPUs can be used for lots of things besides mining
 - Lower entry price. The $/mhash is significantly lower on a GPU than with a FPGA. The BFL single is closing this gap somewhat though.
 - Hardware flexibility. Find a old computer at a recycling center, change the PSU and you can use it for mining. GPU mining needs little besides a good GPU!

The cons
 - Heat. GPUs produce alot of heat
 - Noise. Constant tornadoes in your room. I haven't slept properly since about March 2011
 - Electricity. GPU mining uses alot of juice! Living in countries where the $/KW price is high can effectively ruin mining
 - Space. GPU mining takes up alot of space, especially when  you start your multirig 9001 Ghash/s rigs.

The pros of FPGA mining
 - Less heat than GPU mining
 - Less noise than GPU mining
 - Uses considerably less electricity than GPU mining
 - Takes up less space than GPU mining
 - No tornadoes

The cons
 - Higher $/mhash price
 - Resalve value. Now i'm not entirely sure about this one, but to my knowledge FPGAs can be used for passwordcracking or renderingfarms, and probably other things that i haven't even heard of. I'm not sure how effective specialized BTC FPGAs will be at either of these, but i'm sure they can be tweaked to work for these purposes. Renderingfarms are needed/used at schools and other places, passwordcracking is abit more shady.


These are the only things i can think off. I'm sure someone out there has more to add to this list, or something that is wrong with that list. This is just what i've picked up in the FPGA/GPU debate. I myself will be GPU mining for aslong as i can. I currently pay nothing for electricity, and if i have to one day in the future, the Kw price here is $0.06, so nothing major. If i had $30k and had the patience to wait 15 weeks for the BFL rigbox i would probably invest in one. But that's a different story Hope this helped answer atleast some of your questions!

[Dhomochevsky]

I'm lazy so I'll just copy-paste here what I posted on that other thread of yours:

"1 - If you worry about the mystery miner from Spain flooding the market with coins, then you should be terrified about the early adopters such as Artforz pushing the market to zero with their current accumulated wealth. Hell, between MM and Artforz, I think only the 2 of them have about 30-30something percent of the total hashing power. Unless they aren't in this for the profit and are spending millions of dollars just to fuck around and take down Bitcoin, dumping everything in an uncontrolled manner would be a silly thing to do. So don't worry about it. I think the Bitcoin network is following a Pareto distribution mode, where 20% of the miners own 80% of the hashing power, or something like that.


2 and 3 - Based on some back-of-the-envelope calculations of the FPGA designs sold here on this forum, I would say that the percentage of "public" FPGA's currently covers about 1.5% of the network's hashing power (optimistic estimate, it should actually be a little lower). By the end of May I expect it to reach about 3%. This is based on a tally of currently delivering FPGA units and an estimate of future sales. For example - the first couple of batches of Icarus boards numbered in the tens. x6500 sold 100 units so far. Ztex sold somewhere around that amount too and BFL just started delivering singles. The third batch of Icarus boards I estimate numbered almost 100 boards, maybe more and x6500 is just starting to deliver another 100 boards. So the growth rate is slightly accelerating, that's why I estimate a doubling in market share in the next 2 months. Beyond that predictions become rather murky though. There are FPGA design projects floating around, there are high odds of optimized bitstreams appearing in the wild, pushing efficiency an extra 50%, there are large mining turnkey solutions being discussed as starting delivery this summer... All in all the market is becoming more and more dynamic and it's realistic to expect a geometric progression in FPGA hashing power until december, when the reward will halve. Therefore it's safe to say that GPU mining will become increasingly unprofitable to mine with, provided the current $/MH ratio stays constant. It might increase, but still, it will be much more profitable to mine with a FPGA than with a GPU.

When the reward halves, I believe the market will slowly adjust to reflect the change, meaning price per BTC will go up, but it will be a rather slow climb, enough to put unprofitable miners out of the game. That is, of course, barring some crazy swings in price or another bubble like the one that popped last summer.

All in all, don't worry about major dumps of coins or people gaining too much hashing power. That would affect everyone, not just you and sooner or later a solution will be found. And if you want to get in the game for a long time, go FPGA, gather capital and wait for the first large scale/high efficiency mining units."

Also, what Kliffen said.