it can not be done for win 7, win 7 is doomed, new pascal does not work there, on win 10 it work already so dunno why you are complaining there, same for lyra2v2 it work good(41MH with a 1070 is almost 90% more than a 970)
the only problem is neos, which i can not expain since ethereum is working, and neo is still there broken(win 10 also) when it is even less memory intensity
You are one hypocritical PoS. You LITERALLY were asking for the same thing in the ethminer thread. Not only that, I don't know why you'd be against bug fixes in the first place. Do you think members in this thread don't read Genoil's Ethminer thread or something?
I explained all the bugs and as bugs they should be fixed. You are the first and probably only person I'm going to put on ignore on here (I don't even have Wolf0 on ignore for how much we disagree), your posts not only have no informational value, they often times are just completely nonsensical, you try to take control of a situation for no real reason, or you just go completely back on what you're saying.
And in case you didn't get this weeks ago... Or maybe today... They're bugs and the miners are broken, which is why they need to be fixed. Hopefully you got it that time. You can't just expect a introduction of a major new GPU architecture to magically work for exactly reasons Pallas was mentioning. Microsoft or Nix won't just magically fix miners for the mining community.
it can not be done for win 7, win 7 is doomed, new pascal does not work there, on win 10 it work already so dunno why you are complaining there, same for lyra2v2 it work good(41MH with a 1070 is almost 90% more than a 970)
the only problem is neos, which i can not expain since ethereum is working, and neo is still there broken(win 10 also) when it is even less memory intensity
the neoscrypt game is all about how the memory is accessed.
latest djm34 version is highly optimised in that regard.
the 1070 must have some differences in memory access which is triggering the slower behavior.
a 1070 should come to my door tomorrow or monday, then I will make some tests on linux.
Interesting, I suppose the same is also correct for 1080. Specifically because it uses GDDR5X, instead of just 5. Considering the amount of memory bandwidth available, it's entirely possible for the 5x to be faster in tasks like NeoS and Dagger... but no one has developed for its 'nuances' yet. Although that being said, HBM didn't end up that much faster on Fury for Dagger regardless of the size of the memory bus, so maybe it's not true at all.
We still need someone to develop a dual miner for Nvidia. It's been about five months since the introduction of Claymores and considering the relative performance of the 1070, there is definitely a lot of performance to be tapped there while mining Dagger.