Awesome release! Really good results with Navi 5700:

5700XT:
 - KawPOW: 21-22Mh/s using 1275/900/750mV pulling 150W at the wall (beats RTX 2070 SUPER in H/W which does about 23Mh/s at 190W)
 - ETH: 51.1Mh/s using 1275/900/750mV (stock timings) pulling 130W at the wall

(Win10 1903, drivers: 19.11.3)

Great work as always!

Thanks for the release guys!

I still get with my vega56 a


Is it because I am on the latest drivers?

Perfect ^ thanks buddy!

Impressive miner for both Vega and Navi cards!

KawPoW

Old Miners:

Vega: 940/920/812 Clocks/Volts

21 mhs

Navi: 1500/1900/800

19 mhs

TRM

Vega: 940/920/812 Clocks/Volts

23 mhs

Navi: 1500/1900/800

26 mhs

ETHASH

Navi: 1500/1900/800
57.9

I didn't understand, why all start to mine kawpow. It uses more power. Eth algo gives more profit.

Up to a couple days ago, rvn was more profitable.  Lately, eth does better on NH (at least at my rates of >.10/kwh), but not necessarily on normal pools.

Also, it's good to be able to run multiple algos on a GPU in case profit shifts or you do want to mine multiple coins. After all, that is one of the reasons to have GPU rigs instead of just getting an ASIC. Variety being the spice of life and whatnot.

It’s not the 1st for AMD - there are at least 2 more before theirs!
Also, don’t forget 2% TRM fee for their fresh algo addition compare to their ETH fee - probably more unless we actually see their source code!!

Yawn, you have nowhere else to FUD at the moment? My statement was the "first and only public AMD miner so far that can _compete with Team Green_", not the first working miner.

The 2% fee is the same as all other AMD miners for the same algo, of course with the exception of the open source reference miner. For this algo, we wrote our own machine-code generating compiler from scratch for the random math in progpow, we don't rely on any OpenCL compiler like everyone else. All code is written from scratch, not a single line "borrowed" from any reference implementation. That goes for both host- and gpu-side. We then chased every single f-ing cycle we could find in the algo. The results speak for themselves. We then set the same dev fee to the same level as everyone else despite probably spending 5x more time.

Regarding fake or hidden super hashrates, we have never lied about anything, there are other miners for that which we have shown already. There always seem to be a number of random posters with some hidden agenda appearing in our threads though claiming the same FUD bullshit about hidden fees. I wonder why.

All in all, you should run the miner you feel comfortable running. I truly understand that many people have issues with closed source software. That said, it's not very difficult to isolate your mining rigs from any sensitive data or network access to mitigate any worry about rogue "features". What you should stop doing is posting in every closed source miner thread saying they should give away what took them 100s of hours to produce at the price of your liking rather than what they decide justifies their effort. If you don't like closed source, don't use it. The dev fee is not your call to make, and I don't think you've ever ran any numbers to justify your whining about expensive fees, you're just an egocentric dude that think you're entitled to everything for free, otherwise you wouldn't be here posting in the first place.

Hi!

Well it's not that simple, there's just no point really. We don't run mining the same way as the traditional switching miners do, we always mine user and dev in parallel, all the time, for all algos and situations where it's possible which is almost all. 99 hashes to the user, 1 hash to the dev, 99 hashes to the user, 1 to the dev, and so on. The user pool is always connected, the dev pool is always connected. We can print "User and dev fee now mining" at the start and that would be it .

We built it this way for a much smoother user experience: continuous hashrate for the user with no drops in pool avg, much lower risk for pools being angry that you stop submitting shares for a while, the pool diff stays constant and doesn't fluctuate with dev fee sessions, absolutely the same continuous gpu load all the time with no short interruptions when switching, etc. Now, if you'd be evil you could argue we built it this way to hide our super massive dev fee hashrate that really is 500% more than we're saying, but yeah, simply not the case...

There are a few notable exceptions though, MTP is one of them. We still try to switch different gpus at different times to keep things smooth for the user and it would honestly just (1) litter the logs with and (2) invite even more hackers that try to pick things apart by logging dev fee start/stop. Moreover, I can count the nr of users we have for MTP on one hand and none of them has complained .

For miners that don't fully trust the displayed hashrate or that the communicated dev fee is correct, I would argue that the best way to verify poolside/dev fee hashrate is with fake pool hacks, much like the one we built ourselves for ethash, available on my github. It sets a low diff, the miner you want to test submits many many more shares compared to a normal pool, and after maybe 12h you have a very accurate picture of the "poolside" hashrate of the miner you're testing because you'll have submitted 500,000 shares by then and get a much much more accurate data point than looking at a 6h or 24h average on some pool for 1000 submitted shares (which really means nothing tbh).