Bitcoin Forum

Alternate cryptocurrencies => Mining (Altcoins) => Topic started by: thegreenturtle on January 14, 2018, 06:07:43 PM



Title: Please help - hashrate of one card dropped from 25 Mh/s to 16
Post by: thegreenturtle on January 14, 2018, 06:07:43 PM
I could use some help with a weird hashrate drop that I just experienced.

I'm trying to optimize my cards, and one of my RX 570s just went form 25 to 16 MH/s after doing this:
-Used Polaris one click bios editor - boosted hashrate a little
-Adjusted memory speed in ASUS GPU Tweak - boosted hashrate a little
-Undervolted memory and GPU voltage by -25 mV - dropped to 16 Mh/s
-When I adjust the voltages back up, I'm still getting the 16 Mh/s, I can't get the hashrate back up.

Any ideas? Thanks in advance for the input.


Title: Re: Please help - hashrate of one card dropped from 25 Mh/s to 16
Post by: doktor83 on January 14, 2018, 06:18:50 PM
lower your memory clocks


Title: Re: Please help - hashrate of one card dropped from 25 Mh/s to 16
Post by: thegreenturtle on January 14, 2018, 06:24:39 PM
I reset memory clocks to stock, I even flashed the stock bios back, and the hashrate is still low.


Title: Re: Please help - hashrate of one card dropped from 25 Mh/s to 16
Post by: Miningstore on January 14, 2018, 06:54:31 PM
Hey thegreenturtle,
Have you attempted adjusting the voltage from your bios? I prefer to have everything done at the BIOS level, in case there's ever any problems with the software configuration. It's possible that the voltage change to revert your settings back from Polaris isn't persisting. Modifying the BIOS and giving it a reboot could potentially help. Additionally, setting everything from the BIOS will keep you from worrying about software configuration down the road, and makes it easier to move GPUs around. If this doesn't work let me know and I can explore a few other ideas!

For instructions on how to make these changes, see step 6 of this blog post: http://www.cryptobadger.com/2017/04/build-ethereum-mining-rig-optimization/


Title: Re: Please help - hashrate of one card dropped from 25 Mh/s to 16
Post by: thegreenturtle on January 15, 2018, 04:00:52 AM
Hey thegreenturtle,
Have you attempted adjusting the voltage from your bios? I prefer to have everything done at the BIOS level, in case there's ever any problems with the software configuration. It's possible that the voltage change to revert your settings back from Polaris isn't persisting. Modifying the BIOS and giving it a reboot could potentially help. Additionally, setting everything from the BIOS will keep you from worrying about software configuration down the road, and makes it easier to move GPUs around. If this doesn't work let me know and I can explore a few other ideas!

For instructions on how to make these changes, see step 6 of this blog post: http://www.cryptobadger.com/2017/04/build-ethereum-mining-rig-optimization/

Thanks! In the future I will just adjust everything in the bios, I didn't really realize that you could write the voltage directly to the bios, that certainly will make things more streamlined. But for now the problem is solved with software:

I resolved the issue, with a lot of help from someone on the gpumining subreddit. This issue was in fact an issue with software configuration - I had EVGA Precision and Asus Gpu Tweak installed and they were conflicting, even though I wasn't using Precision, and dramatically reducing the core and memory clock speeds. We uninstalled both programs and switched to OverDriveNTool to optimize the fan speed, voltage, and memory & core clock (after strapping the timings in Polaris for each card). As of a few minutes ago, my 6 RX rig is back online and is running at 175 Mh/s, which I'm happy with.

Wanted to document what the result was in case someone else with a similar problem stumbles upon this post, and I'm happy to answer any questions.



Title: Re: Please help - hashrate of one card dropped from 25 Mh/s to 16
Post by: Miningstore on January 15, 2018, 04:54:04 AM
Glad to hear you resolved the problem! I figured it was probably a software configuration issue, and thanks for documenting it.