Author

Topic: Fundamental error in coin-profitability sites/calculations? (Read 557 times)

member
Activity: 60
Merit: 10
I hate GPU mining now too, scrypt might be worth playing with once FPGAs for it come out but for example I bought a 7950 from NewEgg, put it into a machine, something went flash when I actually tried to mine with it even before messing with clocks etc, and that machine now no longer works.

I put it into a second machine, mined for maybe a week or two, then there was a frying circuits smell (ozone probably I guess) and the machine went dead.

Each time I put that card in, the ozone smell happens and the machine will not boot. Luckily the machine still does at least still work without that card in it.

Now having to look into how to RMA the thing to NewEgg.

Even the GPUs that work none of them produce the same number of KHashes on scrypt as they do MHashes for SHA256, which the profitability sites seem to imagine they should. I get about 255 Khashes on 5870's that give me 405 MHashes on SHA256. I suspect scrypt is just a less efficient system than SHA256 intrinsically.

FPGAs aren't necessarily all that great either, out of a stack of twelve it was only weeks before two of them died, and the cost of sending them in to be fixed cannot really be justified as they are so close to obsolete already and were already even when I bought them. They, and the block eruptors I bought since, are/were mostly just a stopgap measure to be able to grab up some merged mined coins before they go onto major pools and thence onto exchanges whereupon like the merged mined coins you can merged mine at bitparking they would skyrocket in difficulty beyond what a couple of 5870's could really hope to rake in easily.

So yeah, just buying coins is looking pretty good, though I still hope a KNC Saturn might arrive in time to pay for itself even if K16s don't.

-MarkM-


You're forgetting that Scrypt is VERY memory intensive, and for GPU BTC mining the most efficient way is to downclock your memory, which should never happen with Scrypt, so it's a config problem not a algo efficiency issue.
sr. member
Activity: 280
Merit: 250

Can only imagine the number of laptops that have been fried by mining.

We are seeing the beginnings of a New Wave of CPU mining...
With increasingly better algos that will be GPU/BotNet/Cloud resistant...
And there are non-hardware ways to make distributions more democratic...
As Satoshi had intended until the Mining Oligarchies hijacked the entire space.


legendary
Activity: 2940
Merit: 1090
I hate GPU mining now too, scrypt might be worth playing with once FPGAs for it come out but for example I bought a 7950 from NewEgg, put it into a machine, something went flash when I actually tried to mine with it even before messing with clocks etc, and that machine now no longer works.

I put it into a second machine, mined for maybe a week or two, then there was a frying circuits smell (ozone probably I guess) and the machine went dead.

Each time I put that card in, the ozone smell happens and the machine will not boot. Luckily the machine still does at least still work without that card in it.

Now having to look into how to RMA the thing to NewEgg.

Even the GPUs that work none of them produce the same number of KHashes on scrypt as they do MHashes for SHA256, which the profitability sites seem to imagine they should. I get about 255 Khashes on 5870's that give me 405 MHashes on SHA256. I suspect scrypt is just a less efficient system than SHA256 intrinsically.

FPGAs aren't necessarily all that great either, out of a stack of twelve it was only weeks before two of them died, and the cost of sending them in to be fixed cannot really be justified as they are so close to obsolete already and were already even when I bought them. They, and the block eruptors I bought since, are/were mostly just a stopgap measure to be able to grab up some merged mined coins before they go onto major pools and thence onto exchanges whereupon like the merged mined coins you can merged mine at bitparking they would skyrocket in difficulty beyond what a couple of 5870's could really hope to rake in easily.

So yeah, just buying coins is looking pretty good, though I still hope a KNC Saturn might arrive in time to pay for itself even if K16s don't.

-MarkM-
sr. member
Activity: 280
Merit: 250
Isn't my profitability in real life the amount of fiat a given number of watt-hours brings in, minus the fiat cost of those watt-hours and amortised hardware?

This... less the Opportunity Cost of time spent mining.

That ranges from $10/hour working at 7/11 if you are unskilled...
To about $40-50/hour + benefits that a minimally talented Corporate Engineer makes.

PLUS all crypto-mining ASSUMES Bitcoin continues to go up forever...
Which means it's not a Business Model at all... it's a pure gamble on Bitcoin.

Boiling it down = just buy BTC...
Mining is just a complicated, indirect way of buying BTC.



sr. member
Activity: 364
Merit: 250
Looking at various sites about the relative profitability of the various coins, aren't they all making a fundamental error?

Isn't my profitability in real life the amount of fiat a given number of watt-hours brings in, minus the fiat cost of those watt-hours and amortised hardware?

It seems to me very likely that for any, many, or most coins that I would have to use a GPU or CPU to mine my profit per watt-hour is way the heck less than if I threw away or sold my GPUs and bought ASICs instead, maybe even block eruptors would be more profitable?

So that I would be better off mining all the merged mined SHA256 coins and using them to buy any other kinds of coins than I would buying those other kinds of coins directly?

-MarkM-


You're about to get a ton of hate mail from the ASIC owners. Wink

-Merc
legendary
Activity: 2940
Merit: 1090
Looking at various sites about the relative profitability of the various coins, aren't they all making a fundamental error?

Isn't my profitability in real life the amount of fiat a given number of watt-hours brings in, minus the fiat cost of those watt-hours and amortised hardware?

It seems to me very likely that for any, many, or most coins that I would have to use a GPU or CPU to mine my profit per watt-hour is way the heck less than if I threw away or sold my GPUs and bought ASICs instead, maybe even block eruptors would be more profitable?

So that I would be better off mining all the merged mined SHA256 coins and using them to buy any other kinds of coins than I would mining those other kinds of coins directly?

-MarkM-
Jump to: