In 5+1 rig (5-RX 470 4gb+1 RX 480 8gb) i replaced Power Color Red devil (8gb) with Sapphire Nitro 8gb.
same effect, it is either crashing withinn 2-3 hr or dropping that 8gb RX 480 card.
Conclusion: Opti 1.6.1 does not tolerate mixing cards (ether 470 and 480 or 4gb and 8gb).
rig was not crashing on 5 Rx 470 4gb cards and Opti 1.6.1
maybe just a special case of adding a card, but then, why does it start going at first without crashing at very beginning.
I wonder what would happen if you started one optiminer instance to use your 5 RX470 cards and a separate instance to use your RX480 card.
3 Likes
I have made the following scripts because my miner crash after few hours/days.
**If something wrong with the miner, then kill and reboot, start:**
vim /home/yourusername/optiminer-zcash/watchdog-cmd.sh
echo "Uncomment any of the actions in 'watchdog-cmd.sh' to perform on stuck GPU!"
#========Start=======
#Kill if something wrong
PID=`ps -eaf | grep optiminer | grep -v grep | awk '{print $2}'`
if [[ "" != "$PID" ]]; then
echo "killing $PID"
kill -9 $PID
fi
## Reboot:
/sbin/shutdown -r now
#=======END=========
chmod +x /home/yourusername/optiminer-zcash/watchdog-cmd.sh
**This script will start miner in a tmux session if not running:**
Create script startminer.sh with the following content :
touch /home/yourusername/startminer.sh
vim /home/yourusername/startminer.sh
======START=======
#!/bin/bash
# Check if running
ps auxw | grep optiminer | grep -v grep > /dev/null
if [ $? != 0 ]
#if not, then start
then
/usr/bin/tmux new-session -d /home/cristian/optiminer-zcash/mine.sh
fi
#end
=======END========
This cron job will execute the startminer.sh on every 5 minute
chmod +x /home/yourusername/startminer.sh
Add the following job in crontab
*/5 * * * * /bin/sh /home/youusername/startminer.sh
In order to check your miner just exectue the following command :
See tmux sessions :
# tmux list-sessions
Attach to the session
# tmux a
Detach from session
ctrl+b c
Download scripts : https://github.com/cristiycv/mtools
I hope this will help some miners, i also accept donations 
1 Like
I always run 1 miner per GPU, even though all my GPUs are the same. I have found, over several miners and several coins, that 6-card rigs (mine at least) are clearly more stable that way. Essentially, you’re keeping the threading in the miner simple.
1 Like
can anyone here make bat file for restart gpu that does not work
watchdog i mean
for windows version???
Good idea, thanks. I will try this, however, after experiencing repeated shutdowns as a part of 6 cards config, even prior stable 5 cards configuration crashed last night.
We shall see, but i feel that 1.5.0 was the last truly stable config on my setup.
I will try doing 5+1 and also 1+1+1+1+1 (5 instances) as suggested by @dlehenky
thanks, i will try doing this.
Version 1.6.2 released!
Following up on the reported problems with 1.6.1 that seem related with PCI communication, this release brings a second mode that changes how the host interacts with the GPU. You can enabled it by adding the flag --pci-mode 1 to the command line.
Other than that, 1.6.2 just brings very small performance improvements.
1 Like
Hi Opti. Could you be a bit more specific about the differences between the 2 PCI modes?
Mode 0 is same as 1.6.1.
Mode 1 is similar to 1.5.0.
Thanks! v1.6.x has been like a rock for me, so default it is.
hello optiminer can post here please
the bat file for windows version
for watchdog enabled at cards
because one card need watchdog
windows version
How check GPU temp/fan% and change fan% on Ubuntu?
I just change Windows on Ubuntu and make 2150 sol/S, Crazy.
Hi Opti,
I have tried both options on the 1.6.2 miner on the same, “problematic” linux machines and the problem still remains. Several hangs in several hours. On every machine. Just to clear things a bit, all machines work great on windows and on linux 1.5 version of the miner. But no luck with 1.6.1 or 1.6.2. Can you do something to fix this? Am I the only one with this problems? Maybe because I am using older mobos (P35 chipset and dual and quad core processors), is this making some problems?
I have the same problem. Similarly several machines and all hang with a few hours maximum with 1.6.1 and 1.6.2. All rigts are running RX480s with stock settings (no overclocking or under volting). I’ve reverted back to 1.5.0 and running without any problems for 12 hours or more.
To check the temperature for example the gpu 0:
aticonfig --adapter=0 --od-gettemperature
I dont know what’s happened. All the time ive got the same problem. -i 7 all stock.
- when GPU 1 crash, ive got info ./watchdog-cmd.sh: Permision Denied.
Any one know what i make wrong?
EDIT
I add sudo chmod +x watchdog-cmd.sh
and run again all rig
i have the exact same situation and --pci-mode 0 or 1 does not help.
@Optiminer
I am not sure what the differences are between 1.5.0 and 1.6.1,2, but they are not good (at least for some rigs). The rumor was that the difference was only a smaller fee, but it is definitely NOT the case. I rather get 5-6% less hash with 1.5. and it being stable vs constant crashes on 1.6.1-1.6.2, which is very annoying. Could you fix or reverse the communication scheme, because this might be the culprit as the program starts nicely?
Suffice it to say that Nvidia EWBF software NEVER crashes.
maybe there is something here to seriously improve as far as stability is concerned.
Thanks
On Optiminer 1.6.2 I get a crash once every 24 hours. It’s almost like clock-work, every 24 hours. Strange… I’ll keep monitoring to see if it was just a coincidence or not.
Same issue, also on 1.6.1, this is why i have created the scripts :