Could mining come back to CPUs?

[molecular]

I've been thinking wether or not a cpu manufacturer could come up with the idea of assigning some die space to putting a bitcoin mining unit on one of their products.

What would be the cost for them and could it be a good-enough feature to increase sales by enough to make it viable?

[jmoon]

It is not going to happen.

But CPU's with built-in hashing algorithms have long be thought, however the huge ecosystem of algorithms makes it economically infeasible and pointless.

It simply makes more sense to do it by software.

However, some low-power hardware, such as MCUs with energy-harvesting, will adopt that in midterm. But with low-power usage in mind, and not high-performance.

[molecular]

It is not going to happen.

That doesn't convince me. Why is it not going to happen.

You go on to argue that putting implementations of all kinds of hash-functions into cpus doesn't make sense economically. That's not what I suggested.

[molecular]

Very doubtful because GPU is must more beneficial, I could be wrong who knows?

?. GPU is being killed by ASIC. Couldn't whatever is implemented a ASIC chip be put as an add-on into a cpu?

[grue]

so you want a part of the processor that specifically does one thing? we have that, it's called ASIC

[k9quaint]

OP should mine with CPUs.

[molecular]

so you want a part of the processor that specifically does one thing? we have that, it's called ASIC

I'm just thinking it might be a lot cheaper to do it this way. What's done with an ASIC could fit into a corner on a CPU with all the production lines being there already with (maybe) little extra cost.

It would also put mining back into the hands of the masses. <- this is not an argument for CPU makers, though, they will only look at cost/reward.

[odolvlobo]

I've been thinking wether or not a cpu manufacturer could come up with the idea of assigning some die space to putting a bitcoin mining unit on one of their products.

What would be the cost for them and could it be a good-enough feature to increase sales by enough to make it viable?

Individually, nobody would be able to compete with the professional miners and their equipment, so it would only work if there were a reason that everyone would want to mine a tiny amount of bitcoins while they are using their computers. Then there would have to be enough demand to allow manufacturers to charge more to cover the additional cost. There are lots of possible reasons why this might occur. For example, a government decides that it is in the best interest of the country, so it gives a tax credit to anyone that buys one of these CPUs. But, I think the chances are low.

[ebildude123]

Simply put: No.

[Peleus]

In a nice way you're asking this because you don't understand what makes miners fast vs slow. That's not an insult many people don't so I'll give you a quick rundown.

CPU's are really good at doing lots of different types of maths. Hence, it' can do hashing (aka bitcoin mining) at a pretty good rate, and can do any type of hashing out there because it's general purpose.

GPU's are really really good at doing lots of different types of maths, because they have a lot more cores than CPU's do. Because you can mining in a lot more parallel way you get dramatic increases over a CPU. Again, because the cores aren't specifically designed for bitcoin mining they can do any type of math with them, and that's what makes them useful (playing games, etc).

ASICs are application specific integrated circuits. Key words here are application specific. They do their desired purpose REALLY REALLY REALLY well, but the down side is they can't do ANYTHING else. ASICs (Bitcoin mining asics) can, and only ever will SHA2(SHA2()). They can't be used for graphics, other types of hashing, adding 1+1, anything.

So, when you say "add a bitcoin mining chip" to a CPU, what you're really talking about is adding an ASIC to a motherboard. That's the only way something would be an effective bitcoin mining chip over a CPU. Short answer is no I can't see manufacturers adding ASICs to boards as it's way to specific a purpose for such a small market, and they can't be used for anything else.

[meebs]

I've been thinking wether or not a cpu manufacturer could come up with the idea of assigning some die space to putting a bitcoin mining unit on one of their products.

What would be the cost for them and could it be a good-enough feature to increase sales by enough to make it viable?

why on earth would they waste space on 100% of cpu's when only 0.001% of users would utilize that space?

Also.. prepare for HARDCORE botnet abuse if CPU mining was reasonably profitable again.

[GernMiester]

It has its called litecoin. 
cpu for bitcoin is dead and has been..



Hey intel can you add BTC mining to your CPU's? LOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOLOL

[18dew]

IN the reward era of 9 or above mining can come back to CPU's as it would not  be lucarative at that point for any one GPU's or asic's to mine btc.

[jspielberg]

I can't see of a practical reason for it.  With the introduction of the difficulty increases that accompany the efficiency of asic mining, future mining will be ruled by "professional" miners.  Casual miners I don't think will have much incentive.

Maybe in the future this is how the network will be secured if:
1) mining becomes less lucrative due to bit shift reward decrease every four years... say... mid-next decade.  I think 90% of total bitcoins mined ever will have been produced.. with the remaining 10% produced of the next 110 years http://en.wikipedia.org/wiki/File:Total_bitcoins_over_time.png
2) bitcoin becomes an accepted method of facilitating commerce that is critical for global economic function

Short of those two things, I don't think there will be much impetus for something like what you are requesting for the general public.

While adding a bitcoin mining asic to the standard chipsets seems extremely unlikely, I would say bitcoin going to $200+ was also extremely improbable.  So ... who knows?

[6135870]

AMD was expected to release the 8k desktop gpu's this spring but they announced that it will be closer to Jan 2014. Only reason the difficulty is going up so quick for ltc is people are adding more gpus, not faster ones.

[molecular]

In a nice way you're asking this because you don't understand what makes miners fast vs slow. That's not an insult many people don't so I'll give you a quick rundown.

CPU's are really good at doing lots of different types of maths. Hence, it' can do hashing (aka bitcoin mining) at a pretty good rate, and can do any type of hashing out there because it's general purpose.

GPU's are really really good at doing lots of different types of maths, because they have a lot more cores than CPU's do. Because you can mining in a lot more parallel way you get dramatic increases over a CPU. Again, because the cores aren't specifically designed for bitcoin mining they can do any type of math with them, and that's what makes them useful (playing games, etc).

ASICs are application specific integrated circuits. Key words here are application specific. They do their desired purpose REALLY REALLY REALLY well, but the down side is they can't do ANYTHING else. ASICs (Bitcoin mining asics) can, and only ever will SHA2(SHA2()). They can't be used for graphics, other types of hashing, adding 1+1, anything.

So, when you say "add a bitcoin mining chip" to a CPU, what you're really talking about is adding an ASIC to a motherboard. That's the only way something would be an effective bitcoin mining chip over a CPU. Short answer is no I can't see manufacturers adding ASICs to boards as it's way to specific a purpose for such a small market, and they can't be used for anything else.

No I'm not, I'm talking about putting it on the die. They're putting GPUs into the CPUs, too, you know.

[molecular]

Ivy Bridge Intel CPU has 1.4 billion transistors. Fully loop-unrolled bitcoin miner unit should use around 10 million transistors (rough gues, who knows a better number?).

So for the cost of 0.7% of the die space, Intel could include a 3 GH/s bitcoin miner on-die.

This is not ASIC, it's "pure silicon", 22nm process (bfl ASIC uses 65nm), so it would be extremely power-efficient compared to the ASICs we can buy today.

That's what I'm talking about, not - as some seem to think - using the general purpose arithmetic unit of a CPU for mining.

[Viceroy]

Intel can do anything they want... it cost them billions to design that chip.  What is your point?

[Signus]

Even if Intel or AMD have the resources to dedicate CPU power to Bitcoin, they wouldn't.

The problem is that this would involve the creation of a new architecture, meaning a new datapath, a redesign of the ALU's, control logic, and they would have to write  and implement the new instructions for the architecture.

Even if they did ALL of that to support the necessary instructions, they could still not do it at even close to the speed of a GPU because a CPU only has so many transistors that they can pack into it, and we're reaching the physical limit of how many transistors we can fit onto a die. Just compare the size of a CPU to a GPU.

And assuming that all of the above doesn't even matter, a manufacturer like Intel or AMD wouldn't waste their time developing hardware for a cryptocurrency. It's not in their interests.

So regardless GPU >>>>>>>>> CPU.

[Malawi]

I am thinking that FPGA has not been explored enough.

Maybee there are some that can be used to mine different crypto-currencies at a decent rate with low power-consumption and low unit-price compared to CPU based mining riggs.