Pages:
Author

Topic: [ANN] lolMiner-mnx, Mining MinexCoin(MNX) New version: v0.4 (July 22) - page 7. (Read 30713 times)

member
Activity: 433
Merit: 48
I am trying v 0.34, but the screen got drowned when I started up.
Is this normal? The executable file is "run_miner.cmd".

Screen should turn off and right on again. That is because the cmd turns of the GPU driver and tries to turn it on again in the watch dog. This helped a bit better in case were 0 sol/s occur to recover without rebooting. If you do not want this, comment out line 257 ("resetGpus") in watchdog-lolminer-v0.34.ps1.

That may also solve the problem for people that use software over clock that is reset by this procedure.



Also, is it possible to adjust the load with v 0.4?

0.4 is almost a complete rewrite of everything because the current design is not bad but not flexible enough for what I want to do in future. Well ... I can give a try to implement that.




Well I think you were lucky you never had that with 0.31. The new watchdog script detects this graphical error messages, kills everything and should be able to get it back to live.

tirou todas minhas duvidas
obrigado

What doubts? XD
Welcome here Smiley
newbie
Activity: 9
Merit: 0
newbie
Activity: 20
Merit: 0
Hi. I am new to mining. After a whole night of mining with this version, i got this error

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

I am using 2 different AMD gpu, R9 280x and HD 5770 at workbatch of 8. Before I got my R9 280x, I've been mining with lolminer 0.31 and set the workbatch at auto with HD 5770 24/7 without a hiccup.
newbie
Activity: 44
Merit: 0
Hello, Lolliedieb.

I am trying v 0.34, but the screen got drowned when I started up.
Is this normal? The executable file is "run_miner.cmd".

Also, is it possible to adjust the load with v 0.4?

What do you mean 'drowned'?
sr. member
Activity: 896
Merit: 289
Twitter: @BTY_11073
Hello, Lolliedieb.

I am trying v 0.34, but the screen got drowned when I started up.
Is this normal? The executable file is "run_miner.cmd".

Also, is it possible to adjust the load with v 0.4?
member
Activity: 433
Merit: 48
YES! I confirm that 0.3.4 works with AMD Fiji (Fury).
You can add to your benchmark on first page:

AMD R9 Nano: 11.5k (stock clock, same result at 2G or HIGH).
AMD R9 Fury X: 12.3K
AMD Radeon Pro Duo: 23.8k

Great to hear that and thank you for the numbers Smiley Also good to know that the 2G / High deliver approx the same. That was purpose of the changes to bring this a bit closer together, so people with virtual memory problems do not lose any more so much hash speed when using lower work batch sizes.

How can I calculate profitability for this mining?

Every hour there are on average 20 Blocks found with 2.5 MNX each, so 50 MNX per hour created. Of this 40 go to miners and the other 10 to minex bank. So your income is approximately

(Your Hash / Network Hash) * 960 MNX / day

From that still you must subtract pool and miner fee.

For example with rigs doing 190k one has 0.1% of network hash rate and thus gets 0.96 - pool and miner fee per day.
Be careful with mining calculators, many existing forget the MNX going to the minex bank and thus overestimating the income.   


newbie
Activity: 142
Merit: 0
Lolliedieb,
0.34 running faster (4%) in my R9 200 cards!
Thank You!
jr. member
Activity: 159
Merit: 5
Hi,

YES! I confirm that 0.3.4 works with AMD Fiji (Fury).
You can add to your benchmark on first page:

AMD R9 Nano: 11.5k (stock clock, same result at 2G or HIGH).
AMD R9 Fury X: 12.3K
AMD Radeon Pro Duo: 23.8k

Good job!

How can I calculate profitability for this mining?



member
Activity: 433
Merit: 48
Working through your pdf and I see the following, thanks a TON for including this feature!! I wish all miners did this.  Cheesy

Your welcome Cheesy
For 0.4 I plan to allow that users can switch sol/s for ksol/s ... lets see Wink


Lolliedieb,
Could we expect any performance improvements in future releases?

Yes, the 0.34 already mines a bit faster compared to 0.33. On my Linux test rig about 2% faster on my Windows AMD Kaveri machine about 5%. Not that big numbers, but faster is faster Wink
For 0.4 I will likely add the support for binary kernels. That may allow the download of Kernels that are optimized for your GPU. Also maybe I provide Kernels then with inline assembler code for more speed. Lets see ^^

Reading your release notes it seems that this version still does not support Fiji (AMD R9 Fury / Nano)? Is this still planned for 0.4.0?

The release notes are not complete - else it would take longer to write them as to write the new code ^^
I just checked the OpenCL Kernel code in AMD offline compiler:

Code:
Target GPU detected:
Fiji (Graphics IP v8)
AMD FirePro (TM) S9300 x2
AMD MxGPU
AMD Radeon (TM) Graphics Processor
AMD Radeon (TM) R9 Fury Series
AMD Radeon Graphics Processor
Radeon (TM) Pro Duo
Radeon Pro SSG Development Kit

Building for Fiji... succeeded.

So the Kernel builds and also the code runs on my test systems with other AMD GCN V3 platforms (I have no Fiji card here, but well - chip architecture is the same) ... so IMAO it should run with 0.34 Smiley
jr. member
Activity: 159
Merit: 5
Hi,
Thanks for 0.34 release!

Reading your release notes it seems that this version still does not support Fiji (AMD R9 Fury / Nano)? Is this still planned for 0.4.0?

Many thanks!
newbie
Activity: 10
Merit: 0
Lolliedieb,
Could we expect any performance improvements in future releases?
newbie
Activity: 44
Merit: 0
Working through your pdf and I see the following, thanks a TON for including this feature!! I wish all miners did this.  Cheesy

--stats-digits
This parameter adjusts the number of decimal digits that are printed in the statistics output. The minimum is
0, which corresponds to no digits beyond decimal delimiter printed. The default is one.
newbie
Activity: 44
Merit: 0
Finally its there: A new version of lolMiner ^^   

More concretely: 0.34 is there


Similar as in the past I did not plan to do this version, but the changes I planned for 0.4 were too fundamental and so well – I wanted to have this intermediate version with its fixes done before.

Here are the download links:
[Lin 64] Click me (0.34, Linux) MD5 of package: 4b6a86481177fe4d09e4dcc71fe20828
[Win 64] Click me (0.34, Windows) MD5 of package: 27075cf216144bc4f47fc139b1fdb7e7


So, what is new:

Windows only: A new watchdog.
This version of lolMiner comes with a watchdog script written in power shell. Basically it detects when the miner is stuck / has 0 sol/s and so on and can restart it if required. You can start it by clicking run_miner.cmd while the old run script is also on board.

To configure the new watchdog you have to edit the watchdog-lolminer-v0.34.ps1 file. Especially you should give the name of your config file in the beginning.

The watchdog is able to reboot the rig after a certain number of miner restarts. To enable this feature you have to set $global:allowReboot=$true in the header of the watchdog on line 18. For the reboot feature to be effective you should place the watchdog in your auto-start.

Also you can allow it do do a screen shot on every crash by setting  $global:allowScreenshot=$true on line 19. The parameter in line 18 controls the number of restarts before the watchdog closes itself or reboots the rig.

In order to handle 0 sol/s cases the watchdog will require admin rights on your computer when it starts. This allows the watchdog to reset your GPU drivers before a new restart attempt. Because of this we made the watchdog open source so you can verify it does no harm to your system.

Fixed a bug in stratum code:
Before the miner got stuck when the first pool in list was not avaiable. This is fixed now. Also the miner no longer closes when the first pool reports an authorization of miner problem, but it now tests all failover pools in a row before exiting due to missing authorization

Changed behavior of –exit-on-zero
Well … did not find a better headline for this. The parameter –exit-on-zero does not exist any more / has no effect. Instead when the miner detects that a GPU did not come back from a mining task properly it will print an error message and close. The same is true when a device does not finish a task for more then 30 seconds.

New internal stratum behavior
The stratum code now monitors for shares not tested (if they are below target) or not submitted. When there was no single share tested for a minute or no share submitted for 10 minutes, the miner will stop working and close itself, so the restart scripts or the Windows watchdog can restart it.

New parameter: --continue-on-error
This is the counterpart to the above two described changes. If this parameter is set the miner will continue working even when an error occurred.
New parameter: --stats-digits
This will set the number of digits printed in statistics messages. E.g. if you set it to 0 the decimal delimiter and everything behind will be cut of. The default is 1.

Slight changes in work management
The change is something I wanted to test for 0.4. Basically it now tries to do the transfer from and to the cards asynchronously. Especially on AMD cards this helps reducing the difference in speed between –work-batch HIGH and AUTO and lower values. Higher is still mining faster but the difference to lower values is now smaller

New parameter: --disable-double-buffer
This is the counterpart to the work management change. For some drivers (the open source Clover OpenCL drivers on Linux) this is required because not all drivers support the change. 

New parameter: --disable-suprnova-for-fee
This parameter is meant to remove the suprnova pool from the list of fall-backs for the miners dev fee. Reason is that this pool is blocked in some countries and that the miner may get stuck on an ISP block of this pool.

lolMiner-mnx now got a license.txt file
Well that was required xD

lolMiner-mnx now displays proper AMD device names
Well its not yet perfect, but I think “Radeon RX 470/480; RX 570/580” is better then “Ellesmere” - but I may be wrong ^^


I hope you have fun with this release. I will now prepare the next steps. As said the planned changes to 0.4 are rather massive, therefore it may take some time.

---------------------
For the stats:
Overall hash rate of all lolMiners out there: 185 Msol/s
Market share: 84% Cheesy

This is AWESOME news!! I will be migrating my farm to this version today, starting with my most troublesome rigs. Thanks so much!   Cheesy
member
Activity: 433
Merit: 48
Finally its there: A new version of lolMiner ^^   

More concretely: 0.34 is there


Similar as in the past I did not plan to do this version, but the changes I planned for 0.4 were too fundamental and so well – I wanted to have this intermediate version with its fixes done before.

Here are the download links:
[Lin 64] Click me (0.34, Linux) MD5 of package: 4b6a86481177fe4d09e4dcc71fe20828
[Win 64] Click me (0.34, Windows) MD5 of package: 27075cf216144bc4f47fc139b1fdb7e7


So, what is new:

Windows only: A new watchdog.
This version of lolMiner comes with a watchdog script written in power shell. Basically it detects when the miner is stuck / has 0 sol/s and so on and can restart it if required. You can start it by clicking run_miner.cmd while the old run script is also on board.

To configure the new watchdog you have to edit the watchdog-lolminer-v0.34.ps1 file. Especially you should give the name of your config file in the beginning.

The watchdog is able to reboot the rig after a certain number of miner restarts. To enable this feature you have to set $global:allowReboot=$true in the header of the watchdog on line 18. For the reboot feature to be effective you should place the watchdog in your auto-start.

Also you can allow it do do a screen shot on every crash by setting  $global:allowScreenshot=$true on line 19. The parameter in line 18 controls the number of restarts before the watchdog closes itself or reboots the rig.

In order to handle 0 sol/s cases the watchdog will require admin rights on your computer when it starts. This allows the watchdog to reset your GPU drivers before a new restart attempt. Because of this we made the watchdog open source so you can verify it does no harm to your system.

Fixed a bug in stratum code:
Before the miner got stuck when the first pool in list was not avaiable. This is fixed now. Also the miner no longer closes when the first pool reports an authorization of miner problem, but it now tests all failover pools in a row before exiting due to missing authorization

Changed behavior of –exit-on-zero
Well … did not find a better headline for this. The parameter –exit-on-zero does not exist any more / has no effect. Instead when the miner detects that a GPU did not come back from a mining task properly it will print an error message and close. The same is true when a device does not finish a task for more then 30 seconds.

New internal stratum behavior
The stratum code now monitors for shares not tested (if they are below target) or not submitted. When there was no single share tested for a minute or no share submitted for 10 minutes, the miner will stop working and close itself, so the restart scripts or the Windows watchdog can restart it.

New parameter: --continue-on-error
This is the counterpart to the above two described changes. If this parameter is set the miner will continue working even when an error occurred.
New parameter: --stats-digits
This will set the number of digits printed in statistics messages. E.g. if you set it to 0 the decimal delimiter and everything behind will be cut of. The default is 1.

Slight changes in work management
The change is something I wanted to test for 0.4. Basically it now tries to do the transfer from and to the cards asynchronously. Especially on AMD cards this helps reducing the difference in speed between –work-batch HIGH and AUTO and lower values. Higher is still mining faster but the difference to lower values is now smaller

New parameter: --disable-double-buffer
This is the counterpart to the work management change. For some drivers (the open source Clover OpenCL drivers on Linux) this is required because not all drivers support the change. 

New parameter: --disable-suprnova-for-fee
This parameter is meant to remove the suprnova pool from the list of fall-backs for the miners dev fee. Reason is that this pool is blocked in some countries and that the miner may get stuck on an ISP block of this pool.

lolMiner-mnx now got a license.txt file
Well that was required xD

lolMiner-mnx now displays proper AMD device names
Well its not yet perfect, but I think “Radeon RX 470/480; RX 570/580” is better then “Ellesmere” - but I may be wrong ^^


I hope you have fun with this release. I will now prepare the next steps. As said the planned changes to 0.4 are rather massive, therefore it may take some time.

---------------------
For the stats:
Overall hash rate of all lolMiners out there: 185 Msol/s
Market share: 84% Cheesy
newbie
Activity: 9
Merit: 0
Hi people. And what you get by means of lolminer now?
newbie
Activity: 22
Merit: 0
hey anyone knows is lolMiner works on SMOS i cant run it  Angry on windows and hiveos works great!
sr. member
Activity: 896
Merit: 289
Twitter: @BTY_11073
Any watchdog type of script that might help with this issue?

0.34 will come next week (hopefully) ... and with a new (real) watchdog script for Windows power shell that should do the job Smiley
I am looking forward to the new version.
I hope it will be more stable behavior on Windows version.

Good job, Lolliedieb.
Pages:
Jump to: