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

[viljy]

For some reason it seems to me that automatic settings for RX550 are not optimal.
How much does the average kh/s can give RX550? ~1.2 kh/s for cn-upx is this normal?

[psthoy]

Has anyone tried srbminer 1.8.3 on a hashvaultpro pool?  Im trying to merge mine haven+bloc, anyone can share how its done?  TIA.

[Iamtutut]

Which driver is meant by Adrenalin 2019 in the recommended drivers? Is that 18.12.2 only or also newer versions?

My 19.4.1 is disapproved by the miner...

Never mine with a driver newer than your miner version.

[doktor83]

1) When the miner freezes, the system error window appears.

But the SRB-Miner process is still in computer memory.

The SRB-Restarter utility thinks that the miner is working. But the miner is frozen.

Not every crash can be catched, restarter works for restarting the miner when a crash occurs because of miner bug code (access violation etc.)

RX562, miner 1.8.2 - 1.8.3 - not OK
{
"cryptonight_type" : "normalv4",
"intensity" : 0,
"double_threads" : true,
"retry_time" : 2,
"persistent_memory" : false,
"min_rig_speed" : 2100,  <<<<<<<<<==============   only works if set to 100. otherwise restart miner!!!!!!!
"min_rig_speed_duration" : 60,
"gpu_conf" :
[
{ "id" : 0, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 1, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 2, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 3, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 4, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8}
]
}

Someone else also reported the 'min_rig_speed' parameter is broken and triggers much earlier than it is set. I am testing it, but so far so good.
On a rig that does 38khs turtle i set it to 36khs and it's not triggering - restarting miner.

Im using with minimal config :

Code:

{
"cryptonight_type" : "turtle",
"intensity" : 0,
"double_threads" : true,
"min_rig_speed" : 36000
}

Going to check it out in code, maybe when gpu_conf is set this parameter gets omitted, or idk.

[doktor83]

Which driver is meant by Adrenalin 2019 in the recommended drivers? Is that 18.12.2 only or also newer versions?

My 19.4.1 is disapproved by the miner...

It is not disapproved, you just see a notice that you are not using one of the recommended drivers 
Basically you can use any driver, if it is working good for you. But the most tested driver is 18.6.1

[alexcrys1]

Which driver is meant by Adrenalin 2019 in the recommended drivers? Is that 18.12.2 only or also newer versions?

My 19.4.1 is disapproved by the miner...

It is not disapproved, you just see a notice that you are not using one of the recommended drivers 
Basically you can use any driver, if it is working good for you. But the most tested driver is 18.6.1

Hello, u will support uplexa new algo:UPXTWO ?

[Lio78]

Hello, u will support uplexa new algo:UPXTWO ?

Quote 

[symplink]

"+ --preparedata can now give a small boost with turtle, mox, heavy and gpu algo on Vega56. Did not test on other cards."
It is like going worse on vega64 - from 49700 to 48400 /12 vegas
Maybe I am doing something wrong?:
SRBMiner-CN.exe --config Config\config-mox.txt --pools poolsmox.txt --preparedata --resetvega --gpureorder --disablegpuwatchdog --logfile %LOGTIME%
or
SRBMiner-CN.exe --config Config\config-mox.txt --pools poolsmox.txt --resetvega --preparedata  --gpureorder --disablegpuwatchdog --logfile %LOGTIME%\
Same results
19.1.1

[doktor83]

Hello, u will support uplexa new algo:UPXTWO ?

Quote 

Sure i will, im waiting on the dev/devs

"+ --preparedata can now give a small boost with turtle, mox, heavy and gpu algo on Vega56. Did not test on other cards."
It is like going worse on vega64 - from 49700 to 48400 /12 vegas
Maybe I am doing something wrong?:
SRBMiner-CN.exe --config Config\config-mox.txt --pools poolsmox.txt --preparedata --resetvega --gpureorder --disablegpuwatchdog --logfile %LOGTIME%
or
SRBMiner-CN.exe --config Config\config-mox.txt --pools poolsmox.txt --resetvega --preparedata  --gpureorder --disablegpuwatchdog --logfile %LOGTIME%\
Same results
19.1.1

You are not doing anything wrong, looks like vega64 does not like it. I tested only on Vega56 with Hynix mem, cause thats what i got.

[commonorx]

Hello, u will support uplexa new algo:UPXTWO ?

Quote  

Sure i will, im waiting on the dev/devs

"+ --preparedata can now give a small boost with turtle, mox, heavy and gpu algo on Vega56. Did not test on other cards."
It is like going worse on vega64 - from 49700 to 48400 /12 vegas
Maybe I am doing something wrong?:
SRBMiner-CN.exe --config Config\config-mox.txt --pools poolsmox.txt --preparedata --resetvega --gpureorder --disablegpuwatchdog --logfile %LOGTIME%
or
SRBMiner-CN.exe --config Config\config-mox.txt --pools poolsmox.txt --resetvega --preparedata  --gpureorder --disablegpuwatchdog --logfile %LOGTIME%\
Same results
19.1.1

You are not doing anything wrong, looks like vega64 does not like it. I tested only on Vega56 with Hynix mem, cause thats what i got.

I really miss the stability that CASTXMR provided. Month after month and I never looked at the miner 100% uptime. Maybe you can reach out to him (CASTXMR DUDE) for some assistance? This miner crashes every week for me.

[duketasos]

Hello, u will support uplexa new algo:UPXTWO ?

Quote  

Sure i will, im waiting on the dev/devs

"+ --preparedata can now give a small boost with turtle, mox, heavy and gpu algo on Vega56. Did not test on other cards."
It is like going worse on vega64 - from 49700 to 48400 /12 vegas
Maybe I am doing something wrong?:
SRBMiner-CN.exe --config Config\config-mox.txt --pools poolsmox.txt --preparedata --resetvega --gpureorder --disablegpuwatchdog --logfile %LOGTIME%
or
SRBMiner-CN.exe --config Config\config-mox.txt --pools poolsmox.txt --resetvega --preparedata  --gpureorder --disablegpuwatchdog --logfile %LOGTIME%\
Same results
19.1.1

You are not doing anything wrong, looks like vega64 does not like it. I tested only on Vega56 with Hynix mem, cause thats what i got.

I really miss the stability that CASTXMR provided. Month after month and I never looked at the miner 100% uptime. Maybe you can reach out to him (CASTXMR DUDE) for some assistance? This miner crashes every week for me.

Weird i switched monthw ago to srb coz it was rock solid.cast wasnt good for me...(oem vegas 56) but was solid for my rx580s

[psthoy]

has anyone tried configuring haven+bloc in hashvaultpro?  could you share your configs please?

[doktor83]

has anyone tried configuring haven+bloc in hashvaultpro?  could you share your configs please?

You don't need configs, you need pools file :

Code:

{
"pools" :
[
	{"pool" : "pool.haven.hashvault.pro:80",
	 "wallet" : "HAVEN_WALLET:BLOC_WALLET",
	 "password" : "x"
	 }
]
}

WAtch the : sign between HAVEN and BLOC wallet.

[doktor83]

1) When the miner freezes, the system error window appears.

But the SRB-Miner process is still in computer memory.

The SRB-Restarter utility thinks that the miner is working. But the miner is frozen.

Not every crash can be catched, restarter works for restarting the miner when a crash occurs because of miner bug code (access violation etc.)

RX562, miner 1.8.2 - 1.8.3 - not OK
{
"cryptonight_type" : "normalv4",
"intensity" : 0,
"double_threads" : true,
"retry_time" : 2,
"persistent_memory" : false,
"min_rig_speed" : 2100,  <<<<<<<<<==============   only works if set to 100. otherwise restart miner!!!!!!!
"min_rig_speed_duration" : 60,
"gpu_conf" :
[
{ "id" : 0, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 1, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 2, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 3, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 4, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8}
]
}

Someone else also reported the 'min_rig_speed' parameter is broken and triggers much earlier than it is set. I am testing it, but so far so good.
On a rig that does 38khs turtle i set it to 36khs and it's not triggering - restarting miner.

Im using with minimal config :

Code:

{
"cryptonight_type" : "turtle",
"intensity" : 0,
"double_threads" : true,
"min_rig_speed" : 36000
}

Going to check it out in code, maybe when gpu_conf is set this parameter gets omitted, or idk.

Ok, so i think i know what is happening. Cn/R precompiles 10 kernels on miner startup. That takes some time * number of gpus in the system.
Your "min_rig_speed_duration" is 60, it is too low, because miner does not start hashing until it precompiled the kernels. So just set the "min_rig_speed_duration" to a higher number, try few minutes...

[doktor83]

if someone has managed to get Tube2 version 1.7.1 to Rx 580 8 GB without fail it will be thankful to share settings.with adrenaline 18.6.1 I tried a lot of combinations but for now without success. "heavy_mode": 3,"Fragments": 9999 was the best I got on 7 cards - 8310 h/s on heavy,  but one of seven cards did not go away."intensity" : 58.   Doc, how do you benchmark the cards when running the program? It is interesting that with the same settings they give different crashes. Have one reached half or some percentage of their power and there are stuck?

Version 1.7.1 ? You should switch to the latest version, i can't help you with older versions.

[Nero_]

1) When the miner freezes, the system error window appears.

But the SRB-Miner process is still in computer memory.

The SRB-Restarter utility thinks that the miner is working. But the miner is frozen.

Not every crash can be catched, restarter works for restarting the miner when a crash occurs because of miner bug code (access violation etc.)

RX562, miner 1.8.2 - 1.8.3 - not OK
{
"cryptonight_type" : "normalv4",
"intensity" : 0,
"double_threads" : true,
"retry_time" : 2,
"persistent_memory" : false,
"min_rig_speed" : 2100,  <<<<<<<<<==============   only works if set to 100. otherwise restart miner!!!!!!!
"min_rig_speed_duration" : 60,
"gpu_conf" :
[
{ "id" : 0, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 1, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 2, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 3, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8},
{ "id" : 4, "intensity" : 26, "worksize" : 32, "threads" : 2, "fragments" : 8}
]
}

Someone else also reported the 'min_rig_speed' parameter is broken and triggers much earlier than it is set. I am testing it, but so far so good.
On a rig that does 38khs turtle i set it to 36khs and it's not triggering - restarting miner.

Im using with minimal config :

Code:

{
"cryptonight_type" : "turtle",
"intensity" : 0,
"double_threads" : true,
"min_rig_speed" : 36000
}

Going to check it out in code, maybe when gpu_conf is set this parameter gets omitted, or idk.

Ok, so i think i know what is happening. Cn/R precompiles 10 kernels on miner startup. That takes some time * number of gpus in the system.
Your "min_rig_speed_duration" is 60, it is too low, because miner does not start hashing until it precompiled the kernels. So just set the "min_rig_speed_duration" to a higher number, try few minutes...

Getting the same here:
Miner does 6500
"min_rig_speed" : 5000
"gpu_conf" :
[
   { "id" : 0, "intensity" : 0, "worksize" : 0, "threads" : 2},
   { "id" : 1, "intensity" : 0, "worksize" : 0, "threads" : 2},
   { "id" : 2, "intensity" : 0, "worksize" : 0, "threads" : 2},
   { "id" : 3, "intensity" : 0, "worksize" : 0, "threads" : 2},
   
]
Miner restarts on the avg 5min mark constantly.

[doktor83]

Getting the same here:
Miner does 6500
"min_rig_speed" : 5000
"gpu_conf" :
[
   { "id" : 0, "intensity" : 0, "worksize" : 0, "threads" : 2},
   { "id" : 1, "intensity" : 0, "worksize" : 0, "threads" : 2},
   { "id" : 2, "intensity" : 0, "worksize" : 0, "threads" : 2},
   { "id" : 3, "intensity" : 0, "worksize" : 0, "threads" : 2},
   
]
Miner restarts on the avg 5min mark constantly.

Cn/r ? Weak cards? How many ? worksize 0 ?

[doktor83]

V1.8.4
- Performance increase for Vega56/64/Fe/Vii up to 15%!
- Minimum for 'main_pool_reconnect' is now 60 seconds instead of 180
- Added new cmd parameters: --disabletweaking, --cgputweakprofile
- Added new config parameter: tweak_profile
- Minor bug fixes


+ Yes, you read it right. Performance increase on ALL algos on Vega56/64/Fe/VII up to 15%!
A new parameter 'tweak_profile' was added, which if used can significantly boost your hashrate. There is section about this on the first page, but in short:

There are 5 (6) profiles, where :

0 - no change, uses your original settings
1 - light tweak
2
3
4
5 - max tweak

You should find separately for every gpu the profile it can handle, and set it in gpu_conf or cmdline.

VERY IMPORTANT :
1. Miner must run with administrator privileges [right click on SRBMiner-CN.exe->properties->compatibility-> check 'Run this program as an administrator' option-> click OK button
2. Don't apply any memory timings with ElioVP's tool before running miner
3. Don't start testing/tweaking with your highly optimised PPT and ODT settings - because it is probably optimised for a miner/algo combination.. Start with stock settings.
4. Be patient, it takes some time to find settings optimal for your gpu/rig

Tweaking is enabled by default, and can be used on supported devices.
If you want to disable it use the --disabletweaking parameter in start.bat

It can't get easier to use, just use + and - on your keyboard to move between the profiles.

Examples on how to use it can be found on the first page.

Special thanks goes to ElioVP for sharing the offset of the timings position.

[mk111]

V1.8.4
- Performance increase for Vega56/64/Fe/Vii up to 15%!
- Minimum for 'main_pool_reconnect' is now 60 seconds instead of 180
- Added new cmd parameters: --disabletweaking, --cgputweakprofile
- Added new config parameter: tweak_profile
- Minor bug fixes


+ Yes, you read it right. Performance increase on ALL algos on Vega56/64/Fe/VII up to 15%!
A new parameter 'tweak_profile' was added, which if used can significantly boost your hashrate. There is section about this on the first page, but in short:

There are 5 (6) profiles, where :

0 - no change, uses your original settings
1 - light tweak
2
3
4
5 - max tweak

You should find separately for every gpu the profile it can handle, and set it in gpu_conf or cmdline.

VERY IMPORTANT :
1. Miner must run with administrator privileges [right click on SRBMiner-CN.exe->properties->compatibility-> check 'Run this program as an administrator' option-> click OK button
2. Don't apply any memory timings with ElioVP's tool before running miner
3. Don't start testing/tweaking with your highly optimised PPT and ODT settings - because it is probably optimised for a miner/algo combination.. Start with stock settings.
4. Be patient, it takes some time to find settings optimal for your gpu/rig

Tweaking is enabled by default, and can be used on supported devices.
If you want to disable it use the --disabletweaking parameter in start.bat

It can't get easier to use, just use + and - on your keyboard to move between the profiles.

Examples on how to use it can be found on the first page.

Special thanks goes to ElioVP for sharing the offset of the timings position.

Legit!

[joseph32]

That tweaking, is it fixed per profile number (so it will be always the same depending on the number)? Or will it run some tweaks in the background to find a sweetspot?