Bitcoin Forum

I just wanted to push an idea out there that occurred to me following the GPU discussion.

The idea is a cryptocurrency that is optimized for mining with specialized hardware, but specialized hardware that already exists and is widely distributed. I'm thinking consoles, like the Xbox360 or the PS3.

You'll remember the problem I was interested in solving was allowing for widespread mining at a reasonable rate to help adoption of the currency.

This would aid adoption because a lot of people have an unused console about the house. People could pop a disk in the drive, switch it on and convert electricity to cryptocurrency. (At a reasonable rate, and the exactly the same rate as others)

There wouldn't be a botnet problem - consoles booting from a disk would be unlikely to be infected with anything. (Probably a more secure way to operate a cryptocurrency wallet than from a PC too)

It would probably be possible to design a work function that takes advantage of particular features of the hardware so that using the console was already the most efficient way to mine the coin, and making FPGA/ASIC wouldn't offer much saving or improvement over the existing hardware.

Commercial mining wouldn't be viable because a commercial operation would need to invest in the hardware, but a home user would already have that as a sunk cost.

I like this idea, do it.   Should this be in project development?  I'm sure at this point people have to be busting out some form of PS3 mining.

Great idea but make sure to premine 10 billion coins before it is publicly released so you get LOADED my friend :)

The issue is that xbox360 is essentially a computer.

The xbox 360 CPU is a 3.2Ghz 3core design.  While it runs on the power PC architecture this doesn't provide any "protection".  Sure the machine code written for xbox would be incompatible with x86 architecture but anyone analyzing the source code could likely come up with equivalent code for x86 that executes equally fast (probably faster) and most likely any code would be dominated by GPUs.

Worse the GPU in the xbox 360 is not GPGPU capable as it uses non-unified shaders.  Maybe someone could hack something together to use xbox GPU but it would be simply outmatched by modern GPUs.  

The problem is that consoles are essentially computers and today (even more so than 2 generators ago) use rather standard off the shelf parts making it it difficult to design an algorithm that would be "console specific".

/thread

Another, perhaps even larger, problem is that you can't run unsigned code on the 360 unless you have a JTAG hacked box. If you're that big of a nerd though, you probably have better hardware to hash with anyway  ;)

DeathAndTaxes's post sums it up. While it's 100% doable technically the amount of effort involved in coding, and the hashrate per watt make the whole thing useless.

It anyone is up for the challenge it should be quite easy to get an XNA port of https://bitcointalk.org/index.php?topic=22867.0 (the managed cpu-only miner), using the 100% managed code cpu-based algorithm, talking to a PC on the LAN that acts as a proxy.

You make an excellent point - the PS3 appears to have much better support for homebrew so it might make a better candidate. PS3Coin.

I agree that creating a hashing function optimized for specific hardware is a difficult task. I'm wondering if it is possible. I don't have the requisite knowledge to answer this question - I'm here to learn.

I've already learned that the PS3 might be a better candidate - it's more homebrew friendly currently. Here's the specs -

-----
The PS3 uses the Cell microprocessor, which is made up of one 3.2 GHz PowerPC-based "Power Processing Element" (PPE) and six accessible Synergistic Processing Elements (SPEs). A seventh runs in a special mode and is dedicated to aspects of the OS and security, and an eighth is a spare to improve production yields. PlayStation 3's Cell CPU achieves a maximum of 230.4 GFLOPS in single precision floating point operations and 100 GFLOPS double precision.[3] The PS3 has 256MB of Rambus XDR DRAM, clocked at CPU die speed. [4]

The graphics processing unit, according to Nvidia, is based on the NVIDIA G70 (previously known as NV47) architecture. The GPU is clocked at 550 MHz and makes use of 256 MB GDDR3 RAM clocked at 700 MHz with an effective transmission rate of 1.4 GHz.
-----

So I wonder if it is possible to make a hashing function that is optimized for this hardware.

  I can verify 60MH on a PS3, model CH'K'.  But back to the problem of one using an app written for said console on their computer. It is obviously easy to do on PS, but is extremely difficult to do on Xbox360. There is still lacking a functioning emulator that will run 360 software on the PC several years after release. And one would assume it is certainly not due to lack of desire from the gaming community...

I think you are confusing the concept of a PORT with an EMULATOR.

In an emulator the goal is to run Xbox360 native code inside a virtual machine which represents the xbox360 computing environment.  This is useful for pirating games it is utterly useless for hashing xbox coins on a PC.

Reverse engineering the code to produce similar output (i.e. hashes) would be analogous to PORTING a game.  This is done routinely.  Often xbox game and PC game are launched at the same time.  Granted the PC game code can't run on the xbox and the xbox code can't run on the PC but you have the same game running on both platforms.  This is a trivial task. 

So nobody would look to run xbox coins in a emulator they would take the source code figure out how xbox coin hashing works and then write the code for native PC execution.  The end result is that given more powerful hardware it likely would run much faster.  Quickly non-xboxes would dominate the "xboxcoin" network.

To OP:  I used xbox as an example.  All of the issues w/ xbox apply equally to PS3 and any future console.  The hardware in all consoles is essentially PC hardware, PC ram, CPU, GPU, hard drives, flash, etc.  It would be very difficult to find an algorithm which is efficient on xbox (or PS3 or any console) and inefficient on PC.  Even if you did given PC hardware is continually getting faster due to Moore's law any algorithm superiority wouldn't last.  Even less efficient PC hardware would eventually be able to brute force their way into superior hashing power.

I'm not so sure. The hard work doesn't necessarily have to be in the hashing.

One idea - imagine a proof of work function that requires a FPS PS3 game to run for 1 second taking a sequence of inputs, and uses the colour of the pixel that appears at position (0,0) on the display. A port might be able to reproduce this - but it would have to be an absolutely perfect port to reproduce such an exact detail. An emulator might do it - but it'll be a long time before it can do it more efficiently that the PS3 hardware.


Understood - but please use PS3 for future examples to reflect the evolution of the idea - I think the Xbox360 is a non-starter because it's difficult to run homebrew on. However, a next-gen console might prove a better platform if it can be jailbroken.


Remember that it doesn't need to more efficient forever . . . . a 5 or 10 year headstart would be sufficient to allow for mass adoption. Bitcoin only had 1 year or so before the GPU's took over - it wasn't enough time.

Moore's law is a monster.  There is no 5-10 year headstart .... ever.

I don't see GPU as a problem to mass adoption of bitcoin.  99.99999999999999999999999999999999999999999999999999% of the world has no employment as a VISA card processor yet they use VISA card (or paypal, or Dwolla, or etc).  Consumers don't need to be part of the hashing network to use the product.  The hashing network needs to be distributed but it doesn't need to have universal distribution.

Today bitcoin block generation is worth about $36,000 a day.  Now imagine the algorithm perfectly split that among all participants (lets ignore how unfair that would be given hardware and electrical costs).

With 36,000 participants it is $1 per day ea.
With 360,000 participants it is $0.10 per day ea.
With 3.6 million participants it is $0.01 per day ea.

How many people are going to hash 24/7/365 for a penny a day?

Lets look about an end game scenario.
I believe Paypal has 40 million users and generates about $2.23 billion in revenue.  Now lets say someday bitcoin gets that large however the goal would be LOWER transaction costs.  Say bitcoin someday generates 1/4th that ~$600M in gross revenue.  Split among 40 million hashing nodes/users that is $1.50 per person per year (minus hardware and electrical cost).

There is no reason for hashing network to be as distributed as the user network.  It simply needs to be distributed "enough" and everything so far indicates it is.

Actually, this thread brings up an interesting question.

Has anyone actually tried implementing a test miner using an XBox 360 and the XNA Framework?

Anything that gets people involved and invested in the network would be positive. I'm thinking of college kids hashing on their PS3s using "free" university provided power. Pretty soon their roommates, neighbors, friends, professors, etc. would find out. It would help spread the meme.
As a miner, it would make me sad in the short term, though.

https://en.bitcoin.it/wiki/Mining_hardware_comparison

PS3 - 21 MHash/s

I dont think Xbox has significantly greater performance. 21 Mhash/s is nothing.

Setting aside feasibility, is this a good idea? How long is the life cycle of any specific device? What happens to, say, Xbox360Coin when people start moving to the next generation of consoles?

Interesting point. Console manufacturers bring out new console generations every 5 to 10 years, but existing ones remain in use.

If there were a PS3Coin - I guess there are two things to think about with the release of PS4.

One - does it have PS3 emulation, and is that emulation more efficient (in terms of KW) that the original hardware.

Two - how long before the PS4 is jailbroken and the PS3Coin software could be run on it?

Probably the PS4 would be released, but PS3 would still be used for mining until both those issues were addressed.

Another point is that manufacturers also bring out more energy efficient versions of existing consoles - so the PS3Slim is 33% more efficient than the PS3 original, so this would be the preferred hardware to mine on. PS3 original wouldn't be prohibitive, but wouldn't be recommended.

The problem with your idea, as I see it, is that you'd have to design for a very specific piece of hardware to actually keep mining locked into a specific device. Over any appreciable time two things will happen:
-The original device will be phased out and attrition will slowly bring the number of functional units down.
-People will make specialized hardware to do the mining.

I don't see how this scheme is workable or desirable.

That might be a problem 10 years down the road but as you say . . . .


The challenge is to create a proof of work algorithm that makes this expensive and difficult for the next 5/10 years.

It's easy to say 'that's impossible' or 'it's not workable'. It takes more knowledge to say, 'Well here's one way you might think about going about that . . . '  or 'It's not possible, but the closest you could get to that is . . . . '

Currently, there is no way to access the video chip on a PS3. It looks like they JUST (http://twitter.com/#!/gitbrew) got 2d acceleration working in Linux. The miner that is out there uses the 8 (I think there are 8 ) cell processors in the ps3 to mine with.

Hasn't Sony been working to lock out anything homebrew on the PS3?