Topic: How will this change the world of mining?? GTX 1080 / 1070 - page 42. (Read 134091 times)

Must be nice to be an insider, loaner or keeper?

The 1070 results don't seem to scale the same as the 1080. Based on the 1080 rate you posted I was
estimating 26 MH/s on the 1070. But the power usage on the 1070 is unexpectedly low compared with
the 970/980 ratio.

Neoscrypt is a curious beast. The original neoscrypt kernel (DJM34) performs better on kepler (780ti specifically)
than the improved Pallas neoscrypt kernel, although pallas' works better on Maxwell, both compiled with cuda 6.5.
Then the Pallas neoscrypt took a big hit when compiled with cuda 7.5. DJM34 took a crack at it and restored much
of the lost hash. Now it appears it's taking another hit on Pascal.

I would suggest trying the original DJM34 neoscrypt (SP_MOD 58) compiled with cuda 6.5, the Pallas kernel compiled with
6.5 & 7.5 and the improved DJM34 (I think that is what you already tested).

As I mentioned in other topic
1080 at
Quark - 32Mhs at around ~170W
Ethereum - crashes on Genoil 1.0.7 with "device bit not recognizes" message (smth like that)
With Ocl ethereum miner - 12.5mhs at.. 30w
Neoscrypt is not optimized - 0.450
What do you suggest to test next?

1070
Quark was at 24mhs at 110W
the same for ethereum - 12.5 mhs at 30W

If the price depreciation is low after you spend lots of money to buy the high end price, it is still worth it if the power consumption is low.

well it's not fair to compare price at launch, it is known that price after the launch is very high, if you wait some weeks or 1/2 months it will be lower around 600

still worth the lower consumption for those like me that have high electricity cost

The 1080 will cost $700, have 2500 shaders and clocked @ 1600mhz.
A used 980ti can be picked up for $400, 2760 shaders and can be oveclocked to 1500mhz stable. (Gigabyte G1 windforce) Quark will draw around 240Watt

The shadercount of the gtx 1070 is unknown.

To benchmark the pascal is easy

compile this sourcecode

https://github.com/tpruvot/ccminer


run it with

ccminer -a quark --benchmark


And then compare it to my modded Maxwell kernal.

Here are the results on the 980ti: (31,8MHASH) (132.5KHASH /Watt)

no, real consumption target is 150 for 1070 and it was 180 for 970, in fact i said 160, because that what i'm consuming now with my 970, and for 1070 would be around 120w, by derivation

i think the difference in the end will be like going  from 770 to 970

 Actual numbers on both sides are likely lower (well, will be for the 10x0) - you're quoting TDP figures but real world Ethereum mining seems to run somewhat under max TDP on my current 9x0 cards (more so on the 960s, which don't seem to be very efficient at Ethereum than on the 950s that hash almost as fast).

 Still should be a noticeable power usage drop AND should be at least a small speed increase.

 I suspect there might be a small "premium" on the pricing initially, but that should ramp down fairly quickly. Unfortunately, 9xx pricing doesn't seem to have moved yet - guess there's too much stuff in the channel still, and possibly some folks haven't HEARD about the 10x0 being due so soon yet.



 PoS date seems to still be up in the air, the issue at this point is the difficulty rampup, apparently at least in part due to more and more folks finding out about Ethereum and pointing rigs at it. I hesitate to wonder what that would be like if it had a 32-bit version, though the limit on required RAM on your vidcard provides SOME limitations (I'd love to point my 7750s at Ethereum, even if they only managed 3-5 MH/s they're CHEAP cards to run).

The PoW to PoS could happen this time next year or even later. That is a delay to the original plan, which is early next year.

but it seems that the new gpu will have the same 970 price so the price for 970 will be very low after those come out, which mean losing money on each gpu that i sell, and i don't know if the profit of all my six card will match their future loss, when selling them on ebay

Will they be marked at 50% more price. I read that the price will be around $379. That is similar to 970.

Keep.

 As far as I know, there is no hard deadline for the transition away from PoW. Also takes into account any popups from driver issues that might need to be solved.

this does not mean that they will be not faster by 50% in other algo, there is not only ethereum out there

edit, some research lead to a 42% on the access memory, so this is more than 30MH/s versus a 970(22MH/s) with a consumption of 120w versus 160w

so 40% increase in hash and 25% decrese in consumption, very good for me, they will be surely worth 50% more the price

If the memory access speed is not 50% faster, the Ethereum mining will not be 50% faster. I think it is just the power efficiency. We will find out next month.

it's not about power efficient only, they are also much faster up to 50%

When those new graphics comes out, the price will be higher than the MRSP. So you still have time to sell your old cards.

They will be too expensive for ETH mining. Not be worth the power efficiency in the end.

I would keep mining untill the new cards are released

my dilemma now, is do i need to sell my rig now and wait for the new gpu and so no mining for two month, until new ccminer come out, or better to keep mining until they come out, but risking to losing some % on my gpu sales?

My current planning is to build a new GTX970 rig (triple or quad, not sure yet, finding a GOOD triple-card motherboard is proving impossible as NOBODY wants to put slots in SANE places to allow for all 3 cards to get cooling air).

 I'll try it on Ethereum just to see, but it's not INTENDED to be an Ethereum miner except perhaps as a short-term "make back a little of the cost to build it" idea before I put it into it's REAL intended usage (similar to my existing triple GTX950 and double GTX960 rigs, but they'll have a couple months longer of mining before I switch them).


What I'm going to be REAL curious to see is what AMD comes up with on their top-end ZEN APUs for an on-chip GPU. Given the amount of shrink 14nm FinFet allows over their existing process, I would be supprised at less than 1024 stream units, and wouldn't be shocked at 1500ish...

 ... DESPITE announced plans to make the first Zen APUs at least octocore.


 With the clock rate increases the new process allows, I expect AMD to at least match NVidia's 30-40% clock rate jump on their next-gen gear - GPUs as well as CPUs/APUs - pretty much across the board, even on the first gen at the new node.



 Picture this for AMD's next "flagship" product. 6144 Stream units. Base clock around 1400 Mhz, boost more like 1600. Minor increases in IPU efficiency from already-announced "tweeks" to the architecture adding another 5-10% performance. TDP 20-40% lower than a Nano.

 Then they build the "dual" version, having to downclock 5% or so to avoid overheating.

 Remember, you read it here first!



 Yes, 2016 and 2017 look to be VERY interesting times for new hardware in the cryptocoin mining arena.

AMD Polaris 10 or the 480 just has 256 bit memory bus. So it might not be faster than 390 in the Etheruem mining.