Topic: SRBMiner Cryptonight AMD GPU Miner V1.9.3 - native algo switching - page 287. (Read 237247 times)

"cryptonight_type" : "normalv7",
"platform_index" : 1,
"intensity" : 0,
"double_threads" : true,
"target_temperature" : 75,
this config WORK

"cryptonight_type" : "normalv7",
"intensity" : 0,
"double_threads" : true,
"target_temperature" : 75,
This not WORK !

Try it yourself

if the miner hangs, and the main thread becomes unresponsive then pressing R would do nothing
Also you can set in config the "reboot_script" parameter which if set, will disable watchdog standard resetting mode and will run your custom script instead.
You can put in that script for example what Pennywis3 wrote.

There is no platform_index parameter in SRBMiner. So i don't know what are you doing and on which miner

Driver Amd Adrenalin 18.3.1
Intel integrated GPU Hd530 on Cpu
8xAmd Rx550
FIX IT
"platform_index" : 1, in config.txt change Platform !!
(in readmy.txt not comment this parameter )

Now new problem))
{ "id" : 0, "intensity" : 21, "worksize" : 8, "threads" : 2},
Intensity not change ! ( auto intensity =22
id ?

interesting how it found platform 0 as AMD platform, cause it checks by vendor name, and clearly on platform 0 vendor name is Intel(R) Corporation and not Advanced Micro Devices as it can be seen.
Detection actually works great so there is no option to enter platform, who knows what is going on there.
Which driver are you using?

Hi !
Litle problem (((

C:\Users\RIG\Desktop\XMR>start.bat
[2018-05-02 13:15:55] No AMD gpu devices found on platform 0.

C:\Users\RIG\Desktop\XMR>SRBMiner-CN.exe --listdevices
[2018-05-02 13:14:26] Platform #0
[2018-05-02 13:14:26]
[2018-05-02 13:14:26] Intel(R) OpenCL
[2018-05-02 13:14:26] Intel(R) Corporation
[2018-05-02 13:14:26] OpenCL 2.1
[2018-05-02 13:14:26] DEVICES: 2
[2018-05-02 13:14:26]
[2018-05-02 13:14:26] DeviceID

[2018-05-02 13:14:26] Type:GPU
[2018-05-02 13:14:26] Vendor: INTEL
[2018-05-02 13:14:26] CU: 12
[2018-05-02 13:14:26] Error CL_INVALID_VALUE when getting pcie bus id
[2018-05-02 13:14:26] RAM: 3155 MB
[2018-05-02 13:14:26] BUSID: -1
[2018-05-02 13:14:26]
[2018-05-02 13:14:26] DeviceID [1]
[2018-05-02 13:14:26] Type: CPU
[2018-05-02 13:14:26] Vendor: INTEL
[2018-05-02 13:14:26] CU: 2
[2018-05-02 13:14:26] Error CL_INVALID_VALUE when getting pcie bus id
[2018-05-02 13:14:26] RAM: 7887 MB
[2018-05-02 13:14:26] BUSID: -1
[2018-05-02 13:14:26]
[2018-05-02 13:14:26]
[2018-05-02 13:14:26] Platform #1
[2018-05-02 13:14:26]
[2018-05-02 13:14:26] AMD Accelerated Parallel Processing
[2018-05-02 13:14:26] Advanced Micro Devices, Inc.
[2018-05-02 13:14:26] OpenCL 2.1 AMD-APP (2527.
[2018-05-02 13:14:26] DEVICES: 9
[2018-05-02 13:14:26]
[2018-05-02 13:14:26] DeviceID
[2018-05-02 13:14:26] Type:GPU
[2018-05-02 13:14:26] Vendor: AMD
[2018-05-02 13:14:26] CU: 8
[2018-05-02 13:14:26] RAM: 2048 MB
[2018-05-02 13:14:26] BUSID: 4
[2018-05-02 13:14:26]


how to chenge Platform #0 Platform #1

Disabling ADL doesn't help, I've tried.
One way to solve this problem is to disable the watchdog with a shutdown script and add Taskkill /IM SRBMiner-CN.exe /F and OverdriveNTool.exe -r1 -r2 -r3 -r4 -r5 -r6 ( depend on how many cards you have ) to the script.

It has worked for me once, but don't know if it works every time.

It's on the higher side I know, I have to feed my GPUs a bit more V than you see here in general for them (bad asic quality maybe), and there's also two Scythe Ultra Kaze fans on that rig that suck power...

I am using AMD software version 18.3.4, you dont need to restart the vegas with this software version after each restart, it will auto detect. There is no special setting, I just had to reduce the OC on all cards, so running 900 on memory and 1300 on core. 950mV on core and 900mV on mem. 18kh/s, using SRBminer, tried cast too, same results. However, I have disabled the ADL reporting and I am using openhardwaremonitor to check temps and fan speed and I have also disabled the gpuwatchdog. That's it. Screens:

https://i.imgur.com/POSApjf.png

Doc,

Ive just tested the 1.5.0 with 6x Vega 64. Same good hash. I was still using 1.4.1 cuz it was working fine.

Ive tested the disable/enable feature.
After i press 2, took more than a min to disable the card. I have started to wait for the card come back and nothing. lol after 5 min ive decided to press 2 again and voila.
It may be good to add a message "card disable" 3 seconds after the event to be sure.

About JJ script, the script became more conservator and it added a separated disable and enable process with tempo to each card. ive always used the first version doing to all 6 at same time. It never hanged to me IF 5 seconds interval is respected. I mean, after disable all cards, if i hit a key to start the enable process in 3 seconds, the rig will BSOD. This is to all cards of course.

About the "restart_devices_on_startup" :  option,
Can i place this anywhere in the config file?

Next version will have a parameter that will run a .bat you specify after restarting the device, so you can put in something like start c:\OverdriveNTool\OverdriveNTool.exe -r1 -p1"GPU0"  ... etc

So, you are the first human being on planet earth to report a working 10 Vega rig. Can you share your full settings and hardware please?


Doc, im about to test the new version 1.5.0 in my Vega 6 Saphire 64 rig. My only concern is that enable/disable is only half of the job since when the driver hangs, the card lost its OverdriveNTool profile AND softpowertables.
This could be problematic cuz most Vega cards are working underclocked adn undervolted. And disable/enable only recover the softpowertable(registry mod) part making the card work at the default frequency that is 130hz  and 300mv above mining profile.
So, this card will tend to fail since will be regulated in SRB(Temp and RPM) to the previous mining configuration.

So...
The very same script, that is not a random script from the internet but the one made by Jericho Jones(JJ, please add credit), used to incorporate the function to your miner, has the ability to apply OverdriveNTool profiles. It must be made by the config.txt file since the user must edit the lines accordingly with his needs.
This could be a bypass solution to Vegas only. It`s easy since all Vega users already know how to do this and instructions can be found in vegaminingguides. And its only one line to be added to the procedure, takes 2 seconds, with the only requirement that OT and OT.ini must be in same folder...

wish me luck
Cya and thanks

Oh shoot, all this time I thought I was alone, I swear same issue. But it could be that I am having a 10gpu vega rig? I have turned off watchdog and ADL which looks like its helping but can't be certain yet, it has been just about 4 hours without issues.

I've had these problems from the start, hangs when it wants to restart, even with a custom shutdown script, the rig doesn't reboot it just hangs.
I've tried killing the process, resetting the cards before restart but nothing helps.
I have watchdogs that will power off/on the rig when this happens, but that's not good for the components health.

This is the only reason I've stopped using this miner.
I hope you can fix this Doktor 

Can you add a manual miner restart for those who don't want to use the watchdog? For example, press 'r' to manually restart miner. Very important when one gpu crashes in a rig and we dont wanna use the watchdog bec it makes the rig hang.

doctor. sometimes miner gives error:
"Invalid nonce; is miner not compatible with nicehash?"
Any suggestions?

doctor, when miner not found shares yet from start, after pressing "s" key miner writes:
"Boy, no shares found yet."
Of course it's plesure to hear that I'm a boy... I don't hear something like this long time allready...

Does it means, that SRB miner designed specially for man? Wiil be SRB miner for girls? )))))))

That is too much. my 3*vega56 flashed to 64 = 530-570W at the wall .
v7= 6000hr 555-560W

Check crossfire mode on one of cards. If it will be on, then choosing workload mode disabled.

Any chance full API support will be added any time soon, for complete command line control for multi algo switching.