I quoted just 7force who quoted a bunch of people in 2012 saying that mining isn't profitable any more so why bother. The thing that struck me is had the original poster ignored this advice setup mining and paid for the electricity and just mined bitcoins (so essentially converting electricity to bitcoin at the time with no profit.) They could have had a substantial amount of coins to sell at $200 each or $900+ now.
Sure, the only argument is if instead of buying the equipment and electricity and just bought BTC then he would probably be even better off.
Yeah, if they had wasted a shitload of money on hardware and electricity to mine some coins that were worth nothing at the time, they would be rich now. They would've just had to hold on to the worthless coins for a year and half since their crystal ball would've been able to predict the rise in value. It's all just a load of pointless speculation.
This is true. In my case I invest in PC hardware so when I'm done I have nice Graphics cards and extra modern systems available to me. Or if you believe that the coins have value are here to stay then naturally you would believe that the value would go up.
It's all risk as any investment. Though if you bought hardware specifically for mining, you should check if the amount of coins you could've bought with he hardware investment at the time of purchase would have been worth more or less now than the coins you mined since. Even if it's rofitable in dollars, maybe it'd have been better to just buy coins. Though if you have fun building rigs and learning computer stuff, it ain't lost. btw, I just tried for fun to mine some LTC with a tower we got at school, 90kH/s cpu only Though this would never be profitable in any way, it eats something like a kilowatt when doing so...
The original thinking was that if it all crashed then I would have a badass rig. I enjoy working with all the machines and building new ones. So far I've paid for a lot of cool stuff with my mining. I mined with a CPU for a little bit. IF everything stayed the same you would mine one coin in a month. (IF the difficulty doesn't change which it will) Graphics cards are much better if you have any laying around. Even my old 5770 does 200 khash.
Well, I got access to 4 of those beasts (2 cpua,6 cores each, hyperthreaded) but the electricity would be a killer. Also, I'd get in trouble running that at 100% CPU for days for personal stuff. I wonder if it could be possible for a supercomputer to mine bitcoins with a profit now. You know, like, they mine when cores are idles, between jobs and stuff. Or simply stop for one week, mine like crazy, and buy more units with the profits (if any).
Profit? Probably not considering most supercomputers use retail processors which are an order of magnitude or more less efficient because they're bottlenecked by their fetch-decode-execute operation. Mining is an embarrassingly parallel workload, so to my understanding the mining ASICs are just a huge-ass array of pipelined SHA256 units which will push data close to the theoretical limit for the chip's process/feature size and die size. IMO ASICs ruined everything because they pushed everyone else out of the game. At least high-end GPU were still competitive with FPGA due to their speed and area limitations. I doubt there will be many generations of mining ASICs because of the extremely high cost to fab and diminishing profit that will come from increased difficulty. To make any profit mining at today's difficulty you have to start ASAP and with the best ASIC hardware you can get ahold of. Then you'll probably have to reinvest all profit into more hardware to stay competitive with tomorrow's difficulty. There is no sweet spot with hardware, you need the very best since it's an arms race... At some point it'll be time to quit and sell off your gear before diminishing returns due to increasing difficulty, power cost and falling hardware value.
I get that cpu are inefficient at it compared to gpus and ofc ASICS (which could theoretically be as efficient as can be [as in no lost instructions/useless computation]), but supercomputers are a shitload and a half of cpus. The one we got has ~3300 x 12 cores HT cpus... That can't be bad at mining ain't it? It's also very well cooled, "kinda" power efficient and always on anyway... I'm not saying it sure is profitable, but it's not easy to settle out for such huge networks of systems. EDIT: We got pretty cheap electricity too in here.
The age of gpu mining is over as well, custom chips are the only way now. The power of a cpu cluster for mining is going to erode all profits.
It's not really that CPUs are inefficient - it's just that for the specific purpose of generating these hashes custom hardware is far more efficient - and because of the way that the Bitcoin system dynamically adjusts the difficulty depending on how often blocks are found then you reach the situation where the availability of ASICs has made everything else effectively useless.
To put it in perspective an ASIC can pipeline the SHA units down to one hash/tick/unit == likely 1 billion+ hashes/second/unit with a recent process. This is possible because many streams can be worked on in parallel, maybe 100 with a big ASIC. So even if a processor has a SHA friendly SIMD instruction set, it will still work out to multiple ticks/hash. So even with a better fab process and many times more CPUs than ASICs in a mining box, the supercomputer's will lose out due to *inefficiency. If there is a profit to be made with a supercomputer today, there won't be for long. I'd guess the profit today is negligible/less than what supercomputer operators charge for crunching a scientific data set. If however the supercomputers had unutilized crypto accelerators (I don't think so) and it cost them very little in terms of CPU time and power to keep the accelerators fed, maybe it would be profitable to put them to use. *yes relative
Funny how the value of bitcoins began to skyrocket dramatically right after the Silk Road was shut down by the FBI last month. The media began giving the concept of an online drug black market a lot of attention and... Within a month they were worth $900! It's absurd, in 2012 they were worth $5-$10 at the most. They crashed a couple days ago and now the price is back where it was again. I hope it stabilizes.. I like the concept behind bitcoins more than, say, the one behind the Federal Reserve Note, but precious metals (gold, silver) still seem to be the best route for making serious investments. I want to learn more about that sort of thing.
For bitcoins and other sha256 currencies it is but for Litecoin and other Scrypt currencies it's still the way to go. There are rumors of specialized Scrypt miners on the way though.
It's going to be a lot harder to build a parallel engine for Scrypt, though - it was originally intended for password hashing, and making it difficult to implement efficiently in hardware was a specific design aim. It mostly does this by having a key schedule that requires large amounts of RAM to implement, since this is a resource that large logic arrays are generally fairly deficient in. This isn't to say it can't be done, though - just that you will end up being able to pack far fewer units on the same chip than you would with something like SHA1.
I was raaaaather upset when I found out they'd hit $1000USD. Had something like 80-90 at one point when they were under $10. Oh well!
I would imagine we will see a 14nm chip with a lot of on die memory come out sometime in the near future. It's going to be expensive as hell though. I wouldn't be surprised if they skip the whole FPGA market and go straight to ASIC as i can't think of any FPGA's that would fit the proper needs.
I was wondering about that, googled it and came across the unreleased “Horus” 10MH/s FPGA Based scrypt Miner. Correct me if I'm mistaken as I don't have any background in ASIC development, but wouldn't you start off with an FPGA implementation first in any case, to make sure everything works? And if it's a working byproduct of your development cycle, it seems reasonable to me to try and make a standalone product out of it, so you can make some profit while you're moving the now-proven design over to an actual ASIC.
