dag is not loading to the gpu memory. It is calculated by gpu in it originally

V-dimension is best mining option now with gtx1080ti. Check uupool for details

I've got 4.2mhs at 220 watt (gtx1080ti). Nice!

with latest updates Claymore has the same speed with nVidia as Phoenix.
Phoenix's AMD kernels are more "tweak-able" and are faster than Claymore's.
But Phoenix doesn't work as expected with latest 19.5.1 + drivers. We need an update

my micron 5x8 cards (sapphire nitro, se and normal) are doing 34.3mhs 1240/2250 @ 865mv
no mem errors nor incorrect shares

3100 will consume 180-220watt depending on silicon lottery. cn-r takes some advantage from memory overclock and timings on radeon 7 but this impact is small.
eth is 80-82mhs at 200-220watt and no boost from memory clocks above 940mhz and memory timings.
Only core clock matters

2 Eliovp

Hi! Seems I found one more bug in beta 5 and beta 6.
Memory frequency for Radeon VII is not saved/loaded correctly from profiles.
In beta 4 all was ok.
Also would like a feature to have an option to apply only freq/voltage/fans without memory timings loaded from profile.
Thanks for your great work!

Thanks, but 13*13 or 14*13 still better for me. Now I have  1560/1100@865mv setup and 2.88khs cn-r.

Thank you very much!

Didn't you miss my post? Loading fan speeds for Polaris cards from saved profile seems to be broken in beta 5.

Is it only for me that loading profile with latest V5 results in fan speeds blank table at the right bottom ?
Polaris cards. V4 was and is OK.

check mining calculators and you will see what is better from pair 2.77khs xmr @ 150watt and 80mhs eth with 200watt
As for heavy algos we need some new drivers to overcome speed regression when using > 8gb ram. Current drivers can't deliver it for VII

2 kerney66

Now got my hands on radeon 7 (power color with samsung mem).
1540/940@850mv + some timings (similar with my vega64) + 13*13 mode and I have 2.77khs.
Is it ok or should I expect higher results with some tuning? I am not considering to increase voltage and gpu clock. Don't want extra heat.
So, is there a way to get boost without increasing gpu clock?

I've got 82mhs with lowest 850mv voltage I was able to set. (~200watt)
Do you have lower voltage so your clocks are lower and hash is only 75?  What is the wattage?

I'm keeping the strategy "take the max clock from the min voltage", so I never run the card with lower gpu clock that it can handle with given voltage.

latest nvidia kernel boost (not straps but kernel!) made it on par with phoenix.
Now we need the same for polaris. Phoenix has faster kernel 1 and kernel 3, and the speed now is the same due to straps in Claymore's
If you (some way) bring those straps to phoenix it will be even faster.
For example, rx570 8gb micron 1220/2250@855mv (83watt gpu-z) is capable of 34.4mhs with such combination.

My experience with Radeon VII is that it is highly compute bounded. It has so much free mem bandwidth that neither timings nor memory clock matters

Any recommendation for heavy algo and Radeon VII? I want to use extra memory to bump hash rate but it crashes immediately if I set 64 or higher intensity ((

Get Anyone managed to run this beast lower than 850mv gpu voltage? It has 850mv mvddi so I think it getting me from going down the voltage. Now I have 1540/940@850mv with my new Powercolor Radeon VII (Samsung mem) and 81mhs eth. Have no from the wall measurement but I think it is around 200watt (150watt gpu asic power according to hwinfo)

2 Claymore

One more thing. Think you have a bug with NVIDIA timings. When I  try to apply strap 1 to one card and strap 2 to another. But they both have the strap 2 as result.

Not true