By: David Kanter (dkanter.delete@this.realworldtech.com), August 16, 2011 4:31 pm
Room: Moderated Discussions
Robert Daivid Graham (bigrobg@gmail.com) on 8/16/11 wrote:
---------------------------
>>For bit coin mining, you should just send money to RWT
>>directly, since it's a giant scam.
>
>Lol.
>
>I use bitcoin mining because no other benchmark is as optimized for each platform,
>from CPU to GPU to FPGA.
There are a lot of well tuned benchmarks. I would think Linpack for starters, or FFTs, etc.
>That's because it turns "integer addition" into $$$$ (as
>long as you sell the bitcoins you've mined as soon as >possible). The more you can
>optimize it, the more money you can generate.
Unlike dollars, pesos, rubles, etc. bitcoins are not legal tender *anywhere*. There is nobody standing behind them who will guarantee an exchange for a currency which can be used to purchase goods and services.
One of the fundamental differences between a currency and a bitcoin is that there are no financial instruments denominated in bitcoins. Every nation collects taxes in their local currency, creating a liquidity demand. Who needs liquidity for bitcoins?
While ostensibly you can exchange bitcoins for $, that was also true of mortgage backed securities up until recently. There is no reason for anyone to want to exchange a hard currency for bitcoins, since you cannot buy anything with them. In fact, the only possible motivation is that you want to hedge against risk in your own currency - at which point, I'd suggest equities, gold, TIPS or commodities.
So you have 0 demand for bit coins. On the other hand, the supply is very large, since not only are people generating bitcoins...whoever created the idea probably has a huge pile sitting around. When supply exceeds demand, prices will fall, and in this case, probably to 0.
Moreover, there are all sorts of problems with the fact that technical changes are fairly discontinuous. Bitcoins must have a way to deal with arbitrage at any point in time around the introduction of new hardware.
Back to actual technical discussions, what real workload does bitcoin mining resemble? To be of any value, a benchmark must resemble some real world workload.
>But the future is in instructions like "vpcmov" that >allows you to express parallel
>integer program flow without branching. Of course, it's >not universal, but a lot
>of interesting problems can be solved this way.
Predication is critical for wide SIMD, that is certainly true.
David
---------------------------
>>For bit coin mining, you should just send money to RWT
>>directly, since it's a giant scam.
>
>Lol.
>
>I use bitcoin mining because no other benchmark is as optimized for each platform,
>from CPU to GPU to FPGA.
There are a lot of well tuned benchmarks. I would think Linpack for starters, or FFTs, etc.
>That's because it turns "integer addition" into $$$$ (as
>long as you sell the bitcoins you've mined as soon as >possible). The more you can
>optimize it, the more money you can generate.
Unlike dollars, pesos, rubles, etc. bitcoins are not legal tender *anywhere*. There is nobody standing behind them who will guarantee an exchange for a currency which can be used to purchase goods and services.
One of the fundamental differences between a currency and a bitcoin is that there are no financial instruments denominated in bitcoins. Every nation collects taxes in their local currency, creating a liquidity demand. Who needs liquidity for bitcoins?
While ostensibly you can exchange bitcoins for $, that was also true of mortgage backed securities up until recently. There is no reason for anyone to want to exchange a hard currency for bitcoins, since you cannot buy anything with them. In fact, the only possible motivation is that you want to hedge against risk in your own currency - at which point, I'd suggest equities, gold, TIPS or commodities.
So you have 0 demand for bit coins. On the other hand, the supply is very large, since not only are people generating bitcoins...whoever created the idea probably has a huge pile sitting around. When supply exceeds demand, prices will fall, and in this case, probably to 0.
Moreover, there are all sorts of problems with the fact that technical changes are fairly discontinuous. Bitcoins must have a way to deal with arbitrage at any point in time around the introduction of new hardware.
Back to actual technical discussions, what real workload does bitcoin mining resemble? To be of any value, a benchmark must resemble some real world workload.
>But the future is in instructions like "vpcmov" that >allows you to express parallel
>integer program flow without branching. Of course, it's >not universal, but a lot
>of interesting problems can be solved this way.
Predication is critical for wide SIMD, that is certainly true.
David