Author

Topic: SRBMiner Cryptonight AMD GPU Miner V1.9.3 - native algo switching - page 260. (Read 237263 times)

hero member
Activity: 2548
Merit: 626
@doktor83.  Is there a virtual memory requirement or free ssd/hdd space requirement when compiling the kernels and copying over the device?

Virtual mem is same as for every other Cryptonight miner(the more, better Cheesy) , kernel compile happens on GPU, after that it's written to the hdd, miner directory, and it takes up approx 2.1-2.3mb / binary.
Or what was your question ? Cheesy

Cause this is weird behavior I observed today.  Same I/W/T settings 59/8/1 on 4gb RX 570 which was working fine.  Restarting the miner it gave the error related to below error message,


sometimes drivers or windows behaves strange, i bet if you delete the .srb files and reboot it will all be good.
full member
Activity: 729
Merit: 114
@doktor83.  Is there a virtual memory requirement or free ssd/hdd space requirement when compiling the kernels and copying over the device?

Virtual mem is same as for every other Cryptonight miner(the more, better Cheesy) , kernel compile happens on GPU, after that it's written to the hdd, miner directory, and it takes up approx 2.1-2.3mb / binary.
Or what was your question ? Cheesy

Cause this is weird behavior I observed today.  Same I/W/T settings 59/8/1 on 4gb RX 570 which was working fine.  Restarting the miner it gave the error related to below error message,


So yup is there a specific memory requirement during the compilation process?  Trying to debug why this happened.
Note:  Rebooting system, the miner will work with same config.
hero member
Activity: 2548
Merit: 626
@doktor83.  Is there a virtual memory requirement or free ssd/hdd space requirement when compiling the kernels and copying over the device?

Virtual mem is same as for every other Cryptonight miner(the more, better Cheesy) , kernel compile happens on GPU, after that it's written to the hdd, miner directory, and it takes up approx 2.1-2.3mb / binary.
Or what was your question ? Cheesy

How many memory you recomend for 4 rx 580, being 3 8gb version dual threath.

i think 32gb vmem should be enough
newbie
Activity: 24
Merit: 0
@doktor83.  Is there a virtual memory requirement or free ssd/hdd space requirement when compiling the kernels and copying over the device?

Virtual mem is same as for every other Cryptonight miner(the more, better Cheesy) , kernel compile happens on GPU, after that it's written to the hdd, miner directory, and it takes up approx 2.1-2.3mb / binary.
Or what was your question ? Cheesy

How many memory you recomend for 4 rx 580, being 3 8gb version dual threath.
hero member
Activity: 2548
Merit: 626
@doktor83.  Is there a virtual memory requirement or free ssd/hdd space requirement when compiling the kernels and copying over the device?

Virtual mem is same as for every other Cryptonight miner(the more, better Cheesy) , kernel compile happens on GPU, after that it's written to the hdd, miner directory, and it takes up approx 2.1-2.3mb / binary.
Or what was your question ? Cheesy
hero member
Activity: 2548
Merit: 626
- Fixed kernels 3 & 4 for heavy algo

+ Kernels 3 & 4 couldn't mine heavy algo, as no result was ever found. Fixed.

Sorry can you clarify that bit? What exactly are kernels 3 & 4? You mean GPUs 3 and 4 would not have found any share prior to this update?

1st page and readme:

Quote
Some additional parameters you can use in gpu_conf:
"kernel" : 0-4 , IF 0, MINER WILL SELECT MOST SUITABLE KERNEL, OTHERS ARE : 1-FOR GCN CARDS, 2-FOR PRE-GCN CARDS, 3-FOR PRE-GCN EXPERIMENTAL 1, 4-FOR PRE-GCN EXPERIMENTAL 2

hero member
Activity: 2548
Merit: 626
V1.5.5
- No need for all that stuff in start.bat anymore, because watchdog can now restart the whole miner process without it
- Fixed kernels 3 & 4 for heavy algo
- Fixed OCL binary caching for Vega cards
- Added last job received info in stats and API
- Added 'job_timeout' parameter to pools config, it reconnects to pool if no job received for 'job_timeout' time (in seconds), default is 5 min
- Pool reload now checks if current pool is first in reloaded pools list, if not, it switches to first pool


+ Finally, start.bat has been cleaned, as of now miner (watchdog) can restart itself. (woohoo)
Now you can run multi instances of miner if you need to, without renaming hacks etc.. Every miner instance can restart itself.

+ Kernels 3 & 4 couldn't mine heavy algo, as no result was ever found. Fixed.

+ New parameter that goes in pools config is 'job_timeout'. If not set default is 300 sec (5 min). This can be helpful if for some reason the pool stops sending us new jobs, after the defined time the miner automatically reconnects to the pool.

+ Pressing 'r' reloads pools (defined through CMD line + pools list), and now if it finds the currently running pool on the first place in the new list, it just sets the pointer to it, and pressing 'p' to change pool will jump to the second pool from the list.
If the currently running pool is not same as the one on the first place on this new list, miner automatically connects to that new first pool.


Hey man, just downloaded this release for the first time trying the miner.
I'm using 6 RX580s overclocked correctly (no noob bios flashing shit)
I been using Claymore's CryptoNote AMD GPU Miner v11.3 with a total hash on Stellite coin of 5120
Now with your miner I'm getting 4841 only, much less hashrate. Intensity is 0

Double Thread Enabled Gives me 5100, this is closer to Claymore

Is there anything that can be done to try and tweak this for higher hash for me to switch to your miner.
Claymore one is currently dev free now.


Yes, 1st page and readme has a lot of info about manual gpu tweaking.
hero member
Activity: 2548
Merit: 626
V1.5.5
- No need for all that stuff in start.bat anymore, because watchdog can now restart the whole miner process without it
Hi and thanks for the updates.

Is it possible to configure SRBMiner to simply terminate in case it run into any crashes/issues - so it doesn't restart itself?

When using SRMMiner from applications like Awesome Miner, it's preferred to let Awesome Miner detect and restart any crashed/stopped mining processes.

Thanks!

Quote
"reboot_script" : FILENAME, TURN OFF BUILT IN WATCHDOG AND INSTEAD RUN A USER DEFINED .BAT FILE ON GPU FAILURE (included windows restart .bat)

Example:

tskill SRBMiner-CN
newbie
Activity: 24
Merit: 0
Cryptonighy-Heavy.  1200+ H/s on RX 570 8gb (Micron).  Also, serving as primary display card so couldn't push them too far.  Makes my Vega 56 look bad hashing at 1400 H/s  Cry

https://i.imgur.com/D6pGkAF.jpg

have left them hashing at 1130 H/s 1250/2200 clocks @ 906mV.

Are you using PBE timmings? me at 1150/2200 clocks i got 930h/S
newbie
Activity: 24
Merit: 0
From my experience Micron kinda shit for cryptonight, my hynix at  1150/2100 can do 1050hs but i got 1 micron at 1150/2200 com PB timmings and doing around 940h/s. and sometimes  it starts at 1020h/s but go down to 940h/s, its a shame atleast at heavy the values are not stable.
No 1 wanna share some good timmings for Micron for cryptonight? PBE timmings not giving good values
full member
Activity: 729
Merit: 114
@doktor83.  Is there a virtual memory requirement or free ssd/hdd space requirement when compiling the kernels and copying over the device?
newbie
Activity: 11
Merit: 0
.....B3715

This is my mod from the forum bits media. Still it is more correct to give references to the source, using someone else's work for free.
I Apologize, I will consider on the future. (corrected)
hero member
Activity: 1274
Merit: 556
- Fixed kernels 3 & 4 for heavy algo

+ Kernels 3 & 4 couldn't mine heavy algo, as no result was ever found. Fixed.

Sorry can you clarify that bit? What exactly are kernels 3 & 4? You mean GPUs 3 and 4 would not have found any share prior to this update?
newbie
Activity: 143
Merit: 0
V1.5.5
- No need for all that stuff in start.bat anymore, because watchdog can now restart the whole miner process without it
- Fixed kernels 3 & 4 for heavy algo
- Fixed OCL binary caching for Vega cards
- Added last job received info in stats and API
- Added 'job_timeout' parameter to pools config, it reconnects to pool if no job received for 'job_timeout' time (in seconds), default is 5 min
- Pool reload now checks if current pool is first in reloaded pools list, if not, it switches to first pool


+ Finally, start.bat has been cleaned, as of now miner (watchdog) can restart itself. (woohoo)
Now you can run multi instances of miner if you need to, without renaming hacks etc.. Every miner instance can restart itself.

+ Kernels 3 & 4 couldn't mine heavy algo, as no result was ever found. Fixed.

+ New parameter that goes in pools config is 'job_timeout'. If not set default is 300 sec (5 min). This can be helpful if for some reason the pool stops sending us new jobs, after the defined time the miner automatically reconnects to the pool.

+ Pressing 'r' reloads pools (defined through CMD line + pools list), and now if it finds the currently running pool on the first place in the new list, it just sets the pointer to it, and pressing 'p' to change pool will jump to the second pool from the list.
If the currently running pool is not same as the one on the first place on this new list, miner automatically connects to that new first pool.


Hey man, just downloaded this release for the first time trying the miner.
I'm using 6 RX580s overclocked correctly (no noob bios flashing shit)
I been using Claymore's CryptoNote AMD GPU Miner v11.3 with a total hash on Stellite coin of 5120
Now with your miner I'm getting 4841 only, much less hashrate. Intensity is 0

Double Thread Enabled Gives me 5100, this is closer to Claymore

Is there anything that can be done to try and tweak this for higher hash for me to switch to your miner.
Claymore one is currently dev free now.


legendary
Activity: 3346
Merit: 1094
V1.5.5
- No need for all that stuff in start.bat anymore, because watchdog can now restart the whole miner process without it
Hi and thanks for the updates.

Is it possible to configure SRBMiner to simply terminate in case it run into any crashes/issues - so it doesn't restart itself?

When using SRMMiner from applications like Awesome Miner, it's preferred to let Awesome Miner detect and restart any crashed/stopped mining processes.

Thanks!
hero member
Activity: 2548
Merit: 626
V1.5.5
- No need for all that stuff in start.bat anymore, because watchdog can now restart the whole miner process without it
- Fixed kernels 3 & 4 for heavy algo
- Fixed OCL binary caching for Vega cards
- Added last job received info in stats and API
- Added 'job_timeout' parameter to pools config, it reconnects to pool if no job received for 'job_timeout' time (in seconds), default is 5 min
- Pool reload now checks if current pool is first in reloaded pools list, if not, it switches to first pool


+ Finally, start.bat has been cleaned, as of now miner (watchdog) can restart itself. (woohoo)
Now you can run multi instances of miner if you need to, without renaming hacks etc.. Every miner instance can restart itself.

+ Kernels 3 & 4 couldn't mine heavy algo, as no result was ever found. Fixed.

+ New parameter that goes in pools config is 'job_timeout'. If not set default is 300 sec (5 min). This can be helpful if for some reason the pool stops sending us new jobs, after the defined time the miner automatically reconnects to the pool.

+ Pressing 'r' reloads pools (defined through CMD line + pools list), and now if it finds the currently running pool on the first place in the new list, it just sets the pointer to it, and pressing 'p' to change pool will jump to the second pool from the list.
If the currently running pool is not same as the one on the first place on this new list, miner automatically connects to that new first pool.
newbie
Activity: 79
Merit: 0
Guys, how is the latest driver 18.4.1 faring compared to 18.3.4 using SRBminer?

I didn't realise the major Windows update would flatten the display drivers (f***ing Microsoft). I now have the choice between putting my previous driver on or use the latest.

Any experiences to share?

EDIT: oh well, what the hell, went with the new driver and... no big change in hashrate.
Good driver 18.4.1 with SRBMiner 1.5.2
RX 570 4GB Elpida V7Normal intensity 58, w8, T2  1022H/S
http://prntscr.com/jkqhp6

Could you please share your MOD?
For Elpida cclock=1252MHZ mclock=2040MHZ
777000000000000022AA1C00AC615B3CA0550F142C8C1506006004007C041420CA8980A9020004C 01712262B612B3715

This is my mod from the forum bits media. Still it is more correct to give references to the source, using someone else's work for free.
newbie
Activity: 21
Merit: 0
Hi @doktor83, I would really like to see more logging information in the log file. A starting point would be all the text that is yellow in the miner command window - Temp, Fan RPM, Clock Speeds, etc. There is also some info missing in the logfile when I hit 's'.
hero member
Activity: 2548
Merit: 626
Mini suggestion: If reloaded pool's first pool is different from the currently connected pool, change to the first pool instead of the second.

Now, reload does not disconnect from the currently connected pool, and that's awesome, please do not change that behaviour. I use pools.txt to do coin switching, so I may be naturally mining on the 2nd or 3rd pool etc etc., they are not backup pools.

Ok, i already made it work like you first suggested:
If on pool reload first pool on the list is the same as the currently running pool, no pool change occurs , but when pressing P the next pool is the second from the list.
But if after pools reload the first pool on the list isn't the currently running pool, immediate pool switch happens to the first pool from the new list.

newbie
Activity: 11
Merit: 0
Guys, how is the latest driver 18.4.1 faring compared to 18.3.4 using SRBminer?

I didn't realise the major Windows update would flatten the display drivers (f***ing Microsoft). I now have the choice between putting my previous driver on or use the latest.

Any experiences to share?

EDIT: oh well, what the hell, went with the new driver and... no big change in hashrate.
Good driver 18.4.1 with SRBMiner 1.5.2
RX 570 4GB Elpida V7Normal intensity 58, w8, T2  1022H/S
http://prntscr.com/jkqhp6

Could you please share your MOD?
For Elpida cclock=1252MHZ mclock=2040MHZ
First timeng for Elpida from Mind163 ....B3715 https://forum.bits.media/index.php?/topic/38510-eth-xmr%D1%82%D0%B0%D0%B9%D0%BC%D0%B8%D0%BD%D0%B3%D0%B8rx470480570580-%D0%BE%D0%BF%D1%82%D0%B8%D0%BC%D0%B8%D0%B7%D0%B8%D1%80%D0%BE%D0%B2%D0%B0%D0%BD%D0%BD%D1%8B%D0%B5-%D1%82%D0%B0%D0%B9%D0%BC%D0%B8%D0%BD%D0%B3%D0%B8-%D0%B8-%D0%BA%D0%BE%D0%BD%D1%84%D0%B8%D0%B3%D0%B8-%D0%BF%D0%BE%D0%B4-%D0%BC%D0%B0%D0%B9%D0%BD%D0%B5%D1%80%D1%8B/

How much core voltage and Memory voltage?
thank you.
gpu - 906mV mem- 930mV
Jump to: