Vega 64 - 54MH/s with less than 130w

[sozturk]

Initially when all cards were 1110/840/1028, I can compare my overjoy to mining a whole block solo when I saw all my 6 cards working at 56.17 MH/sec… But with that setting it could never run more than 5 minutes it got GPU Detected DEAD errors. Hence I lowered their settings.

Now my 6 cards working with a total speed similar to all my 7 cards used to work with phoenixminer, and I got a GPU6 which I need to understand and solve its problem. Because it has always been a headache. It worked with phoenixminer but it created minimum 60% invalid shares. I would think it is faulty, but when I connect it to HDMI i see the screen, and I am not capable to test it on more advanced level. Maybe I must put it in a bare desktop pc and try with some games.

[0Lucifer0]

Your GPU 6 is not flashed to 56 (stills show 64). AMDMEMTWEAK need to be run at start of rig using xinit.user.sh
You have to use team red miner not Phoenix miner and it need to be With config B

[sozturk]

Thanks mate, I tried all that but it didn’t work. That’s why I reverted back to 64 bios and disabled it.

Something else is wrong with that GPU which I couldn’t figure out yet.

[0Lucifer0]

You can still try lower like 1000 instead of 1028 and rise when it’s stable

[sozturk]

Here I tried it again. Now it is flashed to same Vega 56 bios and having same amdmemtweak settings. It just started and:

I tried many different things but it just doesn’t want to work with TRM

[vsimovic]

My Vega 64 settings are 1100/840/960

[sozturk]

Are you using the same amdmemtweak for Vega64?

[vsimovic]

yes. It took some time for me to find settings so my system is stable and dont get GPU dead errors like you do.

[sozturk]

I tried almost everything for this GPU… Lowering the core clock memory speeds, increasing them, increasing the voltage, setting memory voltage, removing all OC settings…

It only gives fault with Teamredminer, with phoenixminer it works.

It seems I’m gonna have to accept that somehow it might be faulty…

[renard]

How tweak your 5700 ?? Straps?

[drango]

I can’t exceed 38.15 with the vega 64 stock bios + OC. with readminer

1x vega 64
1 Rx 580

[0Lucifer0]

Yes that’s what happen if you don’t follow the guide >< you need amdmemtweak on startup and you have to flash to 56

[deepfriedbutter]

Thanks to all the contributors for the information on this thread. I have 2x Vega 56 running 51.6MH/s rock solid. Very happy with this.

I am, however, having issues with my timings being applied on boot. I have amdmemtweak entry properly because if I stop the miner and execute ./xinit.user.sh the timings load. I have GUI on boot enabled as well. Not sure what is going on. I’m running hive version 0.6-198@210210

Do I need to modify more things on the file?

Here’s my file:

#!/usr/bin/env bash

This script is run in terminal after X server start

uncomment the following line if you want to see miner log after start

#tail -f /run/hive/miner.1

amdmemtweak --i 2,3 --CL 20 --RAS 26 --RCDRD 12 --RCDWR 12 --RC 38 --RP 12 --RRDS 3 --RRDL 5 --RTP 6 --FAW 12 --CWL 8 --WTRS 4 --WTRL 9 --WR 13 --REF 24000 --RFC 2$
#amdmemtweak --gpu 2 --CL 16 --RAS 30 --RCDRD 12 --RCDWR 5 --RC 44 --RP 10 --RRDS 3 --RRDL 5 --RTP 5 --FAW 20 --CWL 7 --WTRS 4 --WTRL 9 --WR 16 --WRRD 1 --RDWR 19 $
#amdmemtweak --i 2 --CL 17 --RAS 26 --RP 10 --RC 36 --RFC 248 --RRDS 3 --RRDL 3 --RCDRD 12 --RCDWR 5 --REF 15000 --FAW 13 --RTP 4

[SybrBlue]

Are the dollar signs there because of the way you cut and paste? They appear in the terminal to denote that the line goes on to the right BUT they should not be in the actual text.

[drango]

It gives me the following error when flashing.
subsystem IDs do not match

[drango]

ok, from CMD if you let me and it seems to work

[deepfriedbutter]

The dollar signs are there because it’s the end of the terminal I was using. Ignore.

[itsafe]

I have similar issue.
xinit.user.sh is being ignored on two our rigs, have to run a command line manually, then it works
GUI enabled

[othermonitor]

I was having the same issue. Not sure if this safe to do or not, but I just inserted the amdmemtweak line to the miner script in /hive/miners/[your miner here]/h-run.sh above line that runs miner.

So I have, /hive/miners/teamredminer/h-run.sh

#!/usr/bin/env bash

export GPU_MAX_ALLOC_PERCENT=100
export GPU_SINGLE_ALLOC_PERCENT=100
export GPU_MAX_HEAP_SIZE=100
export GPU_USE_SYNC_OBJECTS=1

[[ `ps aux | grep "./teamredminer" | grep -v grep | wc -l` != 0 ]] &&
  echo -e "${RED}$MINER_NAME miner is already running${NOCOLOR}" &&
  exit 1

[[ -f ${MINER_LOG_BASENAME}_head.log ]] && rm "${MINER_LOG_BASENAME}_head.log"

cd $MINER_DIR/$MINER_VER
WATCHDOG=""
[[ -e watchdog.sh ]] && WATCHDOG="--watchdog_script"
amdmemtweak --RAS 26 --RCDRD 12 --RC 38 --RRDS 3 --RRDL 4 --REF 21000
./teamredminer ${WATCHDOG} $(< $MINER_NAME.conf) --api_listen=0.0.0.0:${MINER_API_PORT}   2>&1 | tee --append ${MINER_LOG_BASENAME}.log

Seems to be working correctly.

[Hans]

Hi thanks for the tweak i was able to get close to 50Mh, but the thing is my wattage is 170w-just too high,

Is it becouse I’m using the stable version of hive os?

My card is vega 56 with bios updated to the latest vega 56