Topic: How will amd 7000 series effect mining? (Read 1747 times)
The biggest unknown is the switch from LVIW to MIMD instructions. I've heard arguments both ways that this could both help or hurt mining speeds since they're monkeying with the instruction set.
Don't forget price drops for the 6xxx cards. It might give us some new options now that the 5xxx cards are almost all dried up.
This is a great point people dismiss casually.
You can't get good, cheap mining cards lke 5830 anywhere these days. Your only options are scavenging auction sites.
At current prices 6xxx cards are not efficient for mining at all (except for the 6770 which at $110 and ~210mhash is a good investment).
When the 7-series comes out, 6xxx cards should drop in price significantly, making cards like 6870/6970/50 cost-effective.
This won't happen with GPUs though, because of the difficulty of Programming, our task is trivial compared to the effort being made in graphic engines.
I resent that; I assure you that there is nothing trivial about our cause.
Also, it seems like what is bing said about the 7000 series die shrink indicates that an increase in mining performance will come in one of two ways:
1. Lower power draw yields higher hash rate per Kwh, or
2. Higher shader count yields higher hash rate compared to previous generation GPUs
One thing to add though is that there are ticks and tocks in chip manufactoring.
This means after the decrease in feature size the manufacturing process is perfected so that the wavers have less errors.
This means after the decrease in die area it usually increases again - not to the extent to the fully compensate for the loss but some so the 8000 generation could be about 1.8-1.9 as efficent as the previous one. But as haploid23 said there could be optimizations for gaming without actual increase in processing power though.
The issue is GPUs have LOTS of stuff on them we don't need, optimizations for a particular task can make a huge differnce. If for example all the unnecessary stuff is left out and the chips would use different architecture instead we would have a performance advantage of the factor of 6-20 already.
This won't happen with GPUs though, because of the difficulty of Programming, our task is trivial compared to the effort being made in graphic engines.
This is one of the negative implications of Moores law (and it's interpretation) as the costs for a transistor goes down the time to design and program the chips increases exponentially. The implications of this already begin to manifest. So while we would could have 32 times the processing power by 2021 the manpower required would be 32 times as high as well, since there aren't that many hardware developers and programmers each transistor is used less efficiently.
The way out of this dilemma is starting the process all over again, meaning use a architecture more suited for our purpose and using increased volume to get it to a smaller feature size.
Preparations are already on the way
This means after the decrease in feature size the manufacturing process is perfected so that the wavers have less errors.
This means after the decrease in die area it usually increases again - not to the extent to the fully compensate for the loss but some so the 8000 generation could be about 1.8-1.9 as efficent as the previous one. But as haploid23 said there could be optimizations for gaming without actual increase in processing power though.
The issue is GPUs have LOTS of stuff on them we don't need, optimizations for a particular task can make a huge differnce. If for example all the unnecessary stuff is left out and the chips would use different architecture instead we would have a performance advantage of the factor of 6-20 already.
This won't happen with GPUs though, because of the difficulty of Programming, our task is trivial compared to the effort being made in graphic engines.
This is one of the negative implications of Moores law (and it's interpretation) as the costs for a transistor goes down the time to design and program the chips increases exponentially. The implications of this already begin to manifest. So while we would could have 32 times the processing power by 2021 the manpower required would be 32 times as high as well, since there aren't that many hardware developers and programmers each transistor is used less efficiently.
The way out of this dilemma is starting the process all over again, meaning use a architecture more suited for our purpose and using increased volume to get it to a smaller feature size.
Preparations are already on the way
ElectricMucus pretty much summarized it. the 7xxx will have a die shrink to 28nm. these new chips will probably also have architecture improvement, although that doesn't affect mining, as seen in the 6xxx series. 6xxx is better for gaming per stream processor because of the architecture improvements from the 5xxx series, but mining does not discriminate from architecture efficiencies. what mining does take advantage will be the power efficiencies that will come with the 7xxx series. it would only makes sense that there will be a lower power consumption at the same performance level because of the die shrink, unless AMD decides to keep power requirements the same and add more shaders instead.
The theoretical maximum improvement is (40/28)^2=2.04081633
That goes for the assumption that die area will be the same. However with each feature size decrease there also comes a slightly smaller decrease in the die size. The reason for this that the errors in the silicon do not shrink at the same pace so chip manufacturers decrease the die size along with the feature size every time to compensate for increase costs and to get the least failure rate.
The die area decrease is said to be close to the golden ratio.
This leaves us with an improvement of only 1.2-1.3.
There could be architectural improvements though, we'll see but just don't be disappointed when it comes out
That goes for the assumption that die area will be the same. However with each feature size decrease there also comes a slightly smaller decrease in the die size. The reason for this that the errors in the silicon do not shrink at the same pace so chip manufacturers decrease the die size along with the feature size every time to compensate for increase costs and to get the least failure rate.
The die area decrease is said to be close to the golden ratio.
This leaves us with an improvement of only 1.2-1.3.
There could be architectural improvements though, we'll see but just don't be disappointed when it comes out
28nm mfg process indicates that these new GPUs will be superior to previous generations in terms of thermal dissipation, higher count and/or better quality shaders, overall processing speed and computational capabilities. While we cant be sure until they arrive, i would say that it is a safe bet that the 7000 series will prove to be significantly superior to previous generations as far as mining goes. Like everyone else here, i'll be very interested to see what the initial results are whenever these cards hit the market. I have a xfire system standing by already!
Don't forget price drops for the 6xxx cards. It might give us some new options now that the 5xxx cards are almost all dried up.
*affect
My guess is, drivers will be crappy enough to start out with that they won't be any better than the 5xxx series, but software improvements will eventually make them the better miner.
My guess is, drivers will be crappy enough to start out with that they won't be any better than the 5xxx series, but software improvements will eventually make them the better miner.
Do you guys expect the card to pump out more mhz than our current cards? Will it be like nvidia cards as the new card architecture will make it worse for mining?
Jump to: