Quote from: zzzzzzzzzz on February 02, 2017, 03:35:17 AM

Quote from: blockops on February 02, 2017, 03:16:36 AM

Quote from: zzzzzzzzzz on February 02, 2017, 03:07:57 AM

Quote from: blockops on February 02, 2017, 02:29:08 AM

Quote from: tiennou on February 02, 2017, 01:46:23 AM

Quote from: blockops on February 02, 2017, 01:43:09 AM

Just upgraded my linux boxes to 1.6.0. Looks like a solid and reliable build so far. Getting an increase across the board. At -i 7 cards are doing:
RX470 8G - 288
Rx480 8G -301
Fury X - 452
Fury Pro - 470
R9 390 - 365

How long have you been using your -i 7 ? mostly out of curiosity, i'm wondering whether that's just as stable as -i 6 on stock cards.
But more importantly, have you also compared your Watt/Sol ratio between -i 6 and -i7 ?

I've been using -i 7 for about a week now.

Good question on Watt/Sol ratio. I had not thought of that! I have wattmeters and will be at the warehouse tomorrow so I will run a test.

Hey blackops! Just chatted with you on the other forum. I've looked at power vs intensity, and at least at the "auto" intensity and higher, there is virtually no difference in power consumption, at least on my Nanos. I see a small rate gain with higher intensity, but I favor stability over a couple sols/s. With 120+ GPUs, it doesn't take very many GPUs hanging to wipe out any hash rate gain, for me anyway.

Yes, that's a lot of GPU's, I see why you are set up for stability.

-i 7 is pretty stable so far, but you're right on the downside of a card hanging, especially if it is at night. It's fun to chase a few extra solutions per sec, but I'm in it for making money, not for bragging rights.

I just counted, I'm only at 60 GPU's....it might be time to go shopping! But, I want to wait just a little longer until the Vega cards come out - either they will be much better, or the Nano's will come down in price.

Humble minds think alike!

Been thinking along the same lines. However, from what I've seen from the AMD tease at CES, HBM2 does *not* have higher memory bandwidth - they doubled the memory clock, but cut the bus width in half. That means the global memory reads done in Equihash will probably not improve *at all*, although the higher mem clock should help the slot writes (relatively small writes) a good bit, but there are much fewer of those. Anyway, only a real mining run with them will tell.

So, I checked the difference between "-i 6" and "-i 7" on my rx 470 (xfx true oc). TL;DR : nothing changed (sols/s, watt, ..). So I reverted back to the default "-i 6".
This might have a better outcome with different GPUs but I guess optiminer did a pretty good choice at picking the intensity parameters.