SRBMiner Cryptonight AMD GPU Miner V1.9.3 - native algo switching

[melloyellow]

doktor, my hawaii cards are being detected as pre-GCN and kernel 2 gets loaded by default.  If you don't set a global intensity then it tries to use 3 threads.  

edit: If I raise the intensity to 42 then I good get speeds with kernel 2 and 2 threads even though I put kernel 1 in my config....

kernel2 is now for 'older' cards, so thats why it selects that one automatically.
Also i got info from someone with 390x cards that for v7 3 threads and intensity 30 works best,  i couldn't test that i dont have those cards.

Do you have 290x ?

you should be able to change to kernel 1, that setting is set per card like :

"gpu_conf" :
[
   {
     "id" : 0,
     "intensity" : 124,
     "worksize" : 8,
     "threads" : 2,
     "kernel" : 1
   }
]

Yes I have 290x, a few 290, and one 390x.  All in 6 card rigs.  When 3 threads are used, not all cards will start.  Here are a couple examples

I:42 W:8 T:2 K:2 (even though K:1 was put in config globally) 290x - 848 h/s & 290 - 776 h/s on average
I:38 W:8 T:2 K:1 (no kernel value in config) 390x - 842 h/s & 390 - 825, 787, 786, 787, 833

So kernel 2 was assigned to 290 series and kernel 1 was assigned to 390 series.  Kind of strange but maybe it's based on memory.  I will try to override K:1 for the 390's in gpu_conf section.
If you could tell me what that person was getting for his 390x with 3 threads and 30 intensity, that would be very helpful.

[1675423771]

1675423771

[1675423771]

1675423771

[1675423771]

1675423771

[1675423771]

1675423771

[1675423771]

1675423771

[1675423771]

1675423771

[doktor83]

Yes I have 290x, a few 290, and one 390x.  All in 6 card rigs.  When 3 threads are used, not all cards will start.  Here are a couple examples

I:42 W:8 T:2 K:2 (even though K:1 was put in config globally) 290x - 848 h/s & 290 - 776 h/s on average
I:38 W:8 T:2 K:1 (no kernel value in config) 390x - 842 h/s & 390 - 825, 787, 786, 787, 833

So kernel 2 was assigned to 290 series and kernel 1 was assigned to 390 series.  Kind of strange but maybe it's based on memory.  I will try to override K:1 for the 390's in gpu_conf section.
If you could tell me what that person was getting for his 390x with 3 threads and 30 intensity, that would be very helpful.

There is no such a thing as global kernel setting in the config, so that's why it isn't working
Kernel parameter can be set only in gpu_conf.
You should try kernel 2 on the 390x also, the 3 thread setup, because kernel 1 is now really more for 4xx 5xx and vegas.

[melloyellow]

Yes I have 290x, a few 290, and one 390x.  All in 6 card rigs.  When 3 threads are used, not all cards will start.  Here are a couple examples

I:42 W:8 T:2 K:2 (even though K:1 was put in config globally) 290x - 848 h/s & 290 - 776 h/s on average
I:38 W:8 T:2 K:1 (no kernel value in config) 390x - 842 h/s & 390 - 825, 787, 786, 787, 833

So kernel 2 was assigned to 290 series and kernel 1 was assigned to 390 series.  Kind of strange but maybe it's based on memory.  I will try to override K:1 for the 390's in gpu_conf section.
If you could tell me what that person was getting for his 390x with 3 threads and 30 intensity, that would be very helpful.

There is no such a thing as global kernel setting in the config, so that's why it isn't working
Kernel parameter can be set only in gpu_conf.
You should try kernel 2 on the 390x also, the 3 thread setup, because kernel 1 is now really more for 4xx 5xx and vegas.

Ah well, that explains some of it.  I'll test kernel 2 with 3 threads and 30 intensity.  Do you remember what that persons 390x was hashing at.

[fapar]

Anyone else having issues with Vegas on CN-Fast? I have multiple vega rigs that give lower numbers than expected. It seems only 3-4 cards will give the appropriate hashrate while the rest take a ~300 drop for some reason. This happens to both 56 rigs and 64s. All my rigs have minimum 200GB of virtual memory on the systems. It's really wierd. Doesn't happen with CN-LiteV7, StelliteV4 or regular CNV7 only with CN-Fast.

All rigs are using Adrenalin 18.6.1. I have tried all the miners from V1.6.1- 1.6.7.


This is a Vega 56 rig:


This is a Vega 64/56 rig:

There is a problem on RX480 - hashrate drop from 850 to 480 hashes. It is then restored and a few good share again. Until I figured out the reason.

[doktor83]

Anyone else having issues with Vegas on CN-Fast? I have multiple vega rigs that give lower numbers than expected. It seems only 3-4 cards will give the appropriate hashrate while the rest take a ~300 drop for some reason. This happens to both 56 rigs and 64s. All my rigs have minimum 200GB of virtual memory on the systems. It's really wierd. Doesn't happen with CN-LiteV7, StelliteV4 or regular CNV7 only with CN-Fast.

All rigs are using Adrenalin 18.6.1. I have tried all the miners from V1.6.1- 1.6.7.


This is a Vega 56 rig:


This is a Vega 64/56 rig:

There is a problem on RX480 - hashrate drop from 850 to 480 hashes. It is then restored and a few good share again. Until I figured out the reason.

So what was the reason, if you figured it out ?

[bobymoore]

Finally i found solution, how to mine and play some games simultaneously! I found two ways, one from device manager, where i can disable the gpu where Will play the games. The second way comes from srb miner Who is a great tool, if i press 0-9 while mining i can disable-enable gpus! So i start miner, disable the card with my display port or hdmi or whatever and i play games! I had crashes in first time, but i had to set driver ppt again....

[ViWe]

Hi;

I am new in SBminer. I am testing miner with Arto algorithm but I am not sure about the hashrate. What is the optimum hashrate with RX580 GPU's?

Kind regards

I don't know whether I am hitting optimal hash rates, but this is what I get on 6xRX580 8gb:
 CN-arto    6.4 kH/s
 CN-litev7    13.5
 CN-alloy    3.2
 CN-heavy 6.8
 CN-fast    12.5
 CN-v7    6.5

Hi;

Thank you for the info. Can you share your config and driver version please? I can't get higer hash rate than 2,2kH/s with Adrenalin 18.6.1 and Adrenalin 18.3.4. but I can get 4kH/s with 18.8.1 but only 2 minute than it's reboot.

Kind regards

Driver 18.6.1
GPU 1340 MHz/1050 mV
Mem 2065 MHz/880 mV
Config for Arto:
"cryptonight_type" : "artocash",
"intensity" : 81,
"double_threads" : true,

Hi;

Once again thank you. I don't know why, these settings are not work with me. System crash after 5 minutes of miner startup. I guess something wrong with my OS or setup.

Kind regards

What could be different is that you may have Hynix memory (mine are Samsung). I also applied the ubermix 3.1 mem straps in BIOS. You did of course set your cards in compute mode?

[starcat]

CRASH
There are configurations that exist which cannot change to the drivers suggested, because it will destroy performance on other mining applications or even worse make them inoperable.
This program CRASHES my Vega 64 system pretty quickly and there is NO WAY I can change the drivers, so essentially it is another miner that will not run for me.
For whatever reason Avermore is the most well behaved of this style of miner and the Monero miners are the worst and this is on every system I have to include * Vega 64, GTX 1071Ti, and RX-570.
X16R, Neoscrypt, Ethash, all workable.
Monero while desirable is not worth the effort and also very slow to mine when running.

[doktor83]

Vega56 users, please try v1.6.7 and a heavy algo (haven, bittube, heavy, italo) with :

intensity : 61
worksize : 9
threads : 2

Do you get better results than earlier ?

[sundownz06]

I would like to see WebChain support added if you have time doktor =)

[wgestickley]

Vega56 users, please try v1.6.7 and a heavy algo (haven, bittube, heavy, italo) with :

intensity : 61
worksize : 9
threads : 2

Do you get better results than earlier ?

How many performance increase have you found?
Maybe try it later when I am back home

Which driver do you use?

[doktor83]

Vega56 users, please try v1.6.7 and a heavy algo (haven, bittube, heavy, italo) with :

intensity : 61
worksize : 9
threads : 2

Do you get better results than earlier ?

How many performance increase have you found?
Maybe try it later when I am back home

Which driver do you use?

Instead of 1500hs i get around 1550 hs without the 'start the miner with intensity 60 then ... ' trick.
As this is VERY nice, i want to be sure im not the only one getting it

[Pennywis3]

doktor - during low hashrate or miner hang events, miner restarts will trigger (not RIG reboot) - under what circumstances will the "reboot-windows.BAT" file triggered ? Can we select via CLI to do RIG REBOOT instead of miner restart in config file? Thanks

Set in config.txt

"reboot_script" : filename to a batch file in miner directory, if set it turns off built in miner reset procedure on gpu failure, and instead runs this script

Strange, but after 1.6.5 version this doesn't seem to work anymore, and the miner just wants to restart and hangs.
Any ideas why?

[citronick]

doktor - during low hashrate or miner hang events, miner restarts will trigger (not RIG reboot) - under what circumstances will the "reboot-windows.BAT" file triggered ? Can we select via CLI to do RIG REBOOT instead of miner restart in config file? Thanks

Set in config.txt

"reboot_script" : filename to a batch file in miner directory, if set it turns off built in miner reset procedure on gpu failure, and instead runs this script

Strange, but after 1.6.5 version this doesn't seem to work anymore, and the miner just wants to restart and hangs.
Any ideas why?

Doktor,

With v1.6.6 on 2 rigs, 2 x Vega56 and 4 x Vega64, and reboot script updated in the config file.

The Vega64 rig is stable mining XMR and has not triggered the reboot file since Aug 28.

The Vega56 rig however, reboots every half hour like clockwork.

I am tempted to revert to v1.6.5 but if you can figure out whats wrong - I will avoid the downtime.

[doktor83]

doktor - during low hashrate or miner hang events, miner restarts will trigger (not RIG reboot) - under what circumstances will the "reboot-windows.BAT" file triggered ? Can we select via CLI to do RIG REBOOT instead of miner restart in config file? Thanks

Set in config.txt

"reboot_script" : filename to a batch file in miner directory, if set it turns off built in miner reset procedure on gpu failure, and instead runs this script

Strange, but after 1.6.5 version this doesn't seem to work anymore, and the miner just wants to restart and hangs.
Any ideas why?

Doktor,

With v1.6.6 on 2 rigs, 2 x Vega56 and 4 x Vega64, and reboot script updated in the config file.

The Vega64 rig is stable mining XMR and has not triggered the reboot file since Aug 28.

The Vega56 rig however, reboots every half hour like clockwork.

I am tempted to revert to v1.6.5 but if you can figure out whats wrong - I will avoid the downtime.

sorry but i can't help without looking at log files.

Add --logfile log.txt in your start.bat (at the end of SRBMiner-CN.exe line) to enable logging then when you got the log (after 30 minutes where you say it restarts), upload the logs somewhere and give me the link please.

[Grumo]

i get windows error telling me that it blocked the miner from accessing the hardware

[amorphis22]

Hi Doktor,

Accoding to you recomendation I changed drivers to Adr.18.6.1(prev. was blockchain)
and faced with hash rate double decreased.
Iw was ~1900h/s on old drivers, but now it ~830h/s for rx570-4 on CN fast.
Compute mode enabled(SRBMiner-CN.exe --setcomputemode).

Please, help. Ireaaly dont want to return to blockchain drv.

[dragonmike]

Isn't there a Monero fork in the works supposed to go live very soon?
Cryptonight v2?
I thought XMR was supposed to fork on 1st September... Maybe I got that wrong.

Have you got a new miner version ready in any case?

EDIT: I found it. September 16th. There's some time left. Bleh. 16th of Sep was correct... but 2017!

[doktor83]

Isn't there a Monero fork in the works supposed to go live very soon?
Cryptonight v2?
I thought XMR was supposed to fork on 1st September... Maybe I got that wrong.

Have you got a new miner version ready in any case?

EDIT: I found it. September 16th. There's some time left.

can you share the link?

[doktor83]

Hi Doktor,

Accoding to you recomendation I changed drivers to Adr.18.6.1(prev. was blockchain)
and faced with hash rate double decreased.
Iw was ~1900h/s on old drivers, but now it ~830h/s for rx570-4 on CN fast.
Compute mode enabled(SRBMiner-CN.exe --setcomputemode).

Please, help. Ireaaly dont want to return to blockchain drv.

Hi i can think of 2 things:

1. You did not run setcompute mode as admin
2. You are using 1 thread instead of 2