We have released ethOS 1.1.0 with RX 400 Series Support
You can get ethOS at: http://gpushack.com/collections/gpushackChangelog: http://ethosdistro.com/changelog/
How to Update: http://ethosdistro.com/kb/#updating
Sample configs: http://ethosdistro.com/pool.txt
How to mine with RX 400 series on ethOS
NOTE: For this update, do not update rigs remotely. Be on-site/local to your rigs.NOTE: You cannot mix and match RX series with any other GPUs in your rig. They are run on two different drivers and cannot coexist at the same time. In the future, when the new driver is able to support older GPUs, we may fix this issue.
0. Update to ethOS 1.1.0 , How to Update: http://ethosdistro.com/kb/#updating
1. Boot the rig with all GPUs attached.
2. Define the RX 400 driver (amdgpu) in the ethOS config as shown in the examples on http://ethosdistro.com/pool.txt and http://ethosdistro.com/changelog/ :
* Added driver [worker] amdgpu (worker is 6-digit rig name) option to config file (required for RX 400 series rig).
3. Wait for reboot prompt, then reboot. Machine will reboot and install the drivers, then it will reboot automatically.
4. Log in and type ‘allow’ to start mining.
NOTE: If you use remote.conf, globally defined configs (globalcore, globalmem, globalpowertune) are not interchangeable between fglrx and amdgpu. It is recommended to use two different remote.conf sources for each class of rigs. If you use local.conf, don't worry about this.
RX 400 Series Overclocking guide
RX Series / AMDGPU-pro overclocking is based on percentages and levels instead of direct values like what are used with previous AMD cards on the older FGLRX driver stack.# SAMPLE RX 400 SERIES CONFIGS FOR INDIVIDUAL RIGS
#
# If you have global configs defined for fglrx rigs, you must use individual rig configs for rigs that have RX 400 series GPUS.
#
# KNOWN GOOD WORKING (for a 4-gpu rig):
#
# driver c94e13 amdgpu
# flg c94e13 --cl-global-work 16384 --farm-recheck 200
# pwr c94e13 4 4 4 4
# cor c94e13 0 0 0 0
#
# EXAMPLES (for a 4-gpu rig):
#
# driver c94e13 amdgpu
# flg c94e13 --cl-global-work 16384 --farm-recheck 200
# cor c94e13 1 1 1 1
# mem c94e13 3 3 3 3
# pwr c94e13 4 4 4 4
#
# If you have global configs defined for fglrx rigs, you must use individual rig configs for rigs that have RX 400 series GPUS.
#
# KNOWN GOOD WORKING (for a 4-gpu rig):
#
# driver c94e13 amdgpu
# flg c94e13 --cl-global-work 16384 --farm-recheck 200
# pwr c94e13 4 4 4 4
# cor c94e13 0 0 0 0
#
# EXAMPLES (for a 4-gpu rig):
#
# driver c94e13 amdgpu
# flg c94e13 --cl-global-work 16384 --farm-recheck 200
# cor c94e13 1 1 1 1
# mem c94e13 3 3 3 3
# pwr c94e13 4 4 4 4
Core - cor range of 0-20
Example: cor 1 changes core clock from 1266 to 1278
Mem - mem range of 0-20
Example: mem 1 changes memory memory from 2000 to 2020
Powertune - pwr range of 0-7
For RX 400 series, powertune is DPM state of your card (Dynamic Power Management). Here is an example of DPM states and corresponding clocks for a reference RX480:
ethos@ethosrig <ip> /home/ethos # cat /sys/class/drm/card0/device/pp_dpm_sclk
0: 300Mhz
1: 608Mhz
2: 910Mhz
3: 1077Mhz *
4: 1145Mhz
5: 1191Mhz
6: 1236Mhz
7: 1278MhzPwr 3 tells the card to use state 3 (active dpm state is noted by the asterisk *)
DPM State is directly linked to power used by the gpu.
Note: Due to a bug in the current version of AMD’s AMDGPU-pro driver, it is likely that pwr and cor are EXCLUSIVE.
If you set cor, you should not set pwr, if you set pwr, you should not set cor, set both at your own risk. ->More like cor does not work at lower pwr