Bitcoin Forum

Alternate cryptocurrencies => Mining (Altcoins) => Topic started by: chrisk76 on December 17, 2017, 10:15:52 PM



Title: Difference in hash rate between cards (Ethash)
Post by: chrisk76 on December 17, 2017, 10:15:52 PM
Hi all.

I have two Radeon RX570 4GB cards in my PC, running Claymore's dual miner for Ethereum only.

I have used PolarisBiosEditor to alter the memory timings for the BIOS for both cards (used one-click feature), and use MSI Afterburner to alter the memory clock once mining.

What I am confused about is that despite running both cards at the same CPU Core frequency and same RAM frequency, one card (Powercolor) is generating 28.818Mh/s, whilst the other (XFX) is generating 23.233Mh/s. What causes such a huge difference between the two? The Powercolor card has Samsung and Elpida RAM, and the XFX has Elpida and Hynix RAM. I know from reading about that the quality of the RAM differs, but I'm missing something, as I'd have thought if both were running at the same clock speed, they should perform roughly the same. (The Powercolor is in an x16 PCIe slot, and the XFX in an x4 PCIe slot).

What is responsible for the massive difference between them?


Title: Re: Difference in hash rate between cards (Ethash)
Post by: smoolae on December 17, 2017, 10:26:02 PM
As far as I know, a GPU can only have a memory from one manufacturer (I might be wrong though)
Maybe bad drivers?


Title: Re: Difference in hash rate between cards (Ethash)
Post by: chrisk76 on December 17, 2017, 10:31:54 PM
I did wonder about the 'two manufacturers' thing. Maybe the BIOS is designed for either of the two.

I've tried the beta blockchain drivers (with DDU between attempts) and the newest drivers, with multiple uninstall/reinstalls. I'm on Windows 10 x64 if that matters.

I'm just really puzzled - I'd have thought that running two cards at the same spec would've yielded the same hash rate on either. It isn't the end of the world that they're different, but part of me really wants to know why there's such a big difference!