How will this change the world of mining?? GTX 1080 / 1070

[restless]

oh finally a good news, so 27 at 130w, do you have proof of the wattmeter? 27MH at 130w is very good, but apparently the nano is doing somehting like 20-22 at 80w, so it's not worse at all...

still they will be pretty expensive, so i would avoid buying right now, you can buy a 970 for cheap(230 euro, used) that does 20.5MH/140w

Makes no sense... one can get 2 7950 for 230e that do 36MH/300-350W stock

[Amph]

oh finally a good news, so 27 at 130w, do you have proof of the wattmeter? 27MH at 130w is very good, but apparently the nano is doing somehting like 20-22 at 80w, so it's not worse at all...

still they will be pretty expensive, so i would avoid buying right now, you can buy a 970 for cheap(230 euro, used) that does 20.5MH/140w

Makes no sense... one can get 2 7950 for 230e that do 36MH/300-350W stock

7950 is bad for resell, no one will buy it after you are done mining ethereum

also the consumption is higher for sure and hash lower, as you mentioned 36 for 350w(i assume the highest since you didn't provide a precise measure) is far worse than 41 at 280w

also there is not only etehruem which is profitable right now, there is at least another one which has the same profit, and amd is bad at mining it

[vaulter]

smi-tool should work on linux as well.

Found where you can set it in linux but dont have a card anymore - maybe next week - but im sure you can lower power draw with reasonable loss in perf

[Amph]

Try finding a store with one IN STOCK.

 8-(



 Someone had a review unit they posted some ETH hashrates on - it was..... VERY VERY underwhelming, a 970 outhashes it.

 Not sure how much of that is "new drivers not optimised for OpenCL or CUDA compute yet" and how much of it is "archetecture changes made it work WORSE on Ethereum than NVidia cards already do"....

they perform surely better on any other algo, due too low consumption and better hash, so i think it may be true that it is just a matter of bad optimization, or lacking of something necessary

genoil said that their potential could be around 40MH, for the 1080, which would be an instant buy.....but anyway, still too pricy for my taste

The maximum of the 1080 is 40MH, just like the max of 290 is 40MH, because of the bandwidth of 320GB/s (simply divide by 8KB/hash). I've heard through the grapvine that Pascal will be able (but does not yet) support (TLB) page sizes of 2MB and 8MB. When this would be enabled, we might finally see hashrates up to copyspeed (40MH) on NVidia. What we've seen with current drivers is much less; 650KH/s on Win7, 6MH/s on Win10 and 24MH/s on Linux (OpenCL).

so theoretically a 970 should have been around 24.5MH that's a pity

[antantti]

Makes no sense... one can get 2 7950 for 230e that do 36MH/300-350W stock

I have two asus 7970 monsters sitting there doing nothing. 100€ each. Interested?

And sorry for OT

[Liquid71]

oh finally a good news, so 27 at 130w, do you have proof of the wattmeter? 27MH at 130w is very good, but apparently the nano is doing somehting like 20-22 at 80w, so it's not worse at all...

still they will be pretty expensive, so i would avoid buying right now, you can buy a 970 for cheap(230 euro, used) that does 20.5MH/140w

Makes no sense... one can get 2 7950 for 230e that do 36MH/300-350W stock

I'd wait for June 29th AMD's Radeon RX 480 $200 14nm should be efficient and good resell value when Ethereum goes PoS

http://www.pcworld.com/article/3077432/components-graphics/polaris-confirmed-amds-200-radeon-card-will-bring-high-end-graphics-to-the-masses.html

I just started a build with 380x rig thinking polaris was months away from hitting the shelves and seeing the $/hash the 1080 is offering I pulled the trigger on some low end AMD GPU's..
The 14nm 480 might make them efficient and good for mining other algos besides Ether. Kinda wish I waited, but then again announcing June 29th availability and them actually being available are two different things. Nvidia has much more control over their supply chain than AMD. I remember when there were no AMD GPU's available and gamers were mad at all the "coin whores" buying up all the GPU's during the Litecoin boom when Nvida just sucked for mining.
I'm not a gamer but maxwell did make me a fan of Nvidia, hopefully they'll have more compelling models like a 1060 and ether mining is on par with AMD otherwise my next rig is 480. Either way I have 6 380x gpu being delivered tomorrow because I got sick of waiting  

[sanas]

The RX480 has just 256 bit memory bus, so it might be slower than the 390 with 512 bit memory bus.

[restless]

Theoretically, 480 should be somewhere between 380 and 390.
25MH maybe more (2GHz gddr5 chips)
It really depends if algo is pure bandwidth(then ~25) or latency bound(maybe close to 30 )

[Amph]

well in the end i'm a bit disappointed about the consumption, all those new gpu should run at a level lower than 100w, not still around 120-130w which is the same of some old gpu if undervolted

i hope that with the correct undervolt they might run at 60-80w

[Next BillG]

well in the end i'm a bit disappointed about the consumption, all those new gpu should run at a lever lower than 100w, not still around 120-130w which is the same of some old gpu if undervolted

i hope that with the correct undervolt they might run at 60-80w

The are supposed to have x2.5 performance/watt. But that might be for the gaming, for mining, the performance/watt is not so great.

[vaulter]

well in the end i'm a bit disappointed about the consumption, all those new gpu should run at a level lower than 100w, not still around 120-130w which is the same of some old gpu if undervolted

i hope that with the correct undervolt they might run at 60-80w

Had pre production board tested just now - at 100w in nvismi (it won't go lower in linux with that pre production board) it does.. the same 26.7mhs
So there you go - hopefully we'll see windows driver at release with support of 2mb tlb fix and will be able to underclock lower - or utilize the perf for dual miner

[Amph]

well in the end i'm a bit disappointed about the consumption, all those new gpu should run at a level lower than 100w, not still around 120-130w which is the same of some old gpu if undervolted

i hope that with the correct undervolt they might run at 60-80w

Had pre production board tested just now - at 100w in nvismi (it won't go lower in linux with that pre production board) it does.. the same 26.7mhs
So there you go - hopefully we'll see windows driver at release with support of 2mb tlb fix and will be able to underclock lower - or utilize the perf for dual miner

nice, i think it can go even lower around 80-90, based on the test i did with the 970, which maintain its hashrate at 33% lower tdp

so if 130w is the start, 90w should be achieved with the 1070

[Bokelka]

So the target for the 1070 mining is not to increase its hash rate, but to reduce the power usage and higher efficiency.

[Amph]

So the target for the 1070 mining is not to increase its hash rate, but to reduce the power usage and higher efficiency.

well 27MH is higher than 22MH from 970 with much lower wattage plus potential for more higher hash

[vaulter]

So the target for the 1070 mining is not to increase its hash rate, but to reduce the power usage and higher efficiency.

The problem here is the memory bandwidth - you are going to see the same for 480
So hitting this limit you can ether go dual miner way and utilize the board to its max or drop GPU voltage and/or clocks to see if you have same hashrate
Basically you can hash at the same speed with less power used

[antantti]

The problem here is the memory bandwidth - you are going to see the same for 480
So hitting this limit you can ether go dual miner way and utilize the board to its max or drop GPU voltage and/or clocks to see if you have same hashrate
Basically you can hash at the same speed with less power used

So it was all about that 512 against 256 mem band? I guess there will be no cheap 290's anymore...

[vaulter]

The problem here is the memory bandwidth - you are going to see the same for 480
So hitting this limit you can ether go dual miner way and utilize the board to its max or drop GPU voltage and/or clocks to see if you have same hashrate
Basically you can hash at the same speed with less power used

So it was all about that 512 against 256 mem band? I guess there will be no cheap 290's anymore...

Not exactly - of course raw bus make an impact but its also dependent on algo - I think dual miner will be better for nv vs amd - we'll see if it happens in near future

[greaterninja]

I am quite skeptical if this review is legitimate.

Please post pics + screenshots of the 1070 &&  27mh/s

[antantti]

The problem here is the memory bandwidth - you are going to see the same for 480
So hitting this limit you can ether go dual miner way and utilize the board to its max or drop GPU voltage and/or clocks to see if you have same hashrate
Basically you can hash at the same speed with less power used

So it was all about that 512 against 256 mem band? I guess there will be no cheap 290's anymore...

Not exactly - of course raw bus make an impact but its also dependent on algo - I think dual miner will be better for nv vs amd - we'll see if it happens in near future

Hmm.. for me it looks like dcr pow mining days are gone...

[Marvell1]

The problem here is the memory bandwidth - you are going to see the same for 480
So hitting this limit you can ether go dual miner way and utilize the board to its max or drop GPU voltage and/or clocks to see if you have same hashrate
Basically you can hash at the same speed with less power used

So it was all about that 512 against 256 mem band? I guess there will be no cheap 290's anymore...

Not exactly - of course raw bus make an impact but its also dependent on algo - I think dual miner will be better for nv vs amd - we'll see if it happens in near future

Hmm.. for me it looks like dcr pow mining days are gone...

I thought dual mining was Amd only