Quote from: heavyarms1912 on April 11, 2018, 05:30:18 PM

Quote from: inv1s1bl3 on April 11, 2018, 04:22:32 PM

Quote from: heavyarms1912 on April 11, 2018, 01:56:58 PM

Quote from: inv1s1bl3 on April 11, 2018, 01:39:00 PM

Okay, so in comparison with xmrig's speed 3,5kh, with this SBRMiner my RX580s are hashing 820h/s on cryptonight heavy!!! Incredible performance! I am currently with intensity 32, because if I get the intensity higher, I have noticed that after some time some cards start hashing around 550h/s(dropping by 300h) .
Running stable with 4150h/s on my 5 cards @ 550W from the wall. Great work man, hands down!

4gb or 8gb? I read 8gb go over 1k on CN-Heavy.

It may give such hash, but my cards are modded for the ethash algo and as far as I read, the timings for cryptonight should be a bit different. I am now looking for a guide or a site from which I will be able to get my hands on such a BIOS mod and I will try then Cheesy

Otherwise still pretty amazing performance, cannot deny that!

P.S - Cards are 8GB Sapphire Nitro+ Special Editions.

Which memory type? Yup, cryptonight works better with different timing straps than ETH.
For 8gb cards you should try the same intensities that you've used on cryptonight.

Also a generic tip. If you are not actively monitoring GPU parameters close apps like HWInfo64, gpu-z or afterburner or change the polling rate to 3sec,5sec.

I'm using the "over and over" utilized Ubermix 3.1, intensity 57 + 2 threads and i'm getting 1100~1150h/s in my Asus RX580 8GB (Samsung memory) at 1280/2010. Try pushing intensity to the limit and then reduce to find the perfect sweet spot. Remember: Cryptonight-heavy is a really different algorithm comparing to Cryptonight V7. The same board I can reach 1100H/s+ in Heavy can do only 1053H/s in V7 (with the best intensity for both).

Unafortunately I don't have specific timings to cryptonight, so I'm using the same I did to Ethash. For Cryptonight V7 the great surprise was RX570 4GB Elpida, getting 1000H/s at 1230/2030. It's quite impressive for me.