Topic: [ANN] cpuminer-opt v3.14.2, open source optimized multi-algo CPU miner - page 16. (Read 9676 times)

When I was doing solo mining of LCC (sha256d algo) I found this Bug report as marked in the screenshot.



Miner of both versions (v3.13.1.1 and 3.13.2) is showing this bug report.

Here is the log file: https://drive.google.com/open?id=1OGFmZKF9ejfFKQtNHwgmkjl5tof-c8ka

I left the miner running more than 20 minutes during solo mining, but it didn't show 5 mins summary.
So, now I can say the miner of both versions (v3.13.1.1 and 3.13.2) is not showing summary during solo mining with Minotaur and sha256d algo.
It only shows summary during pool mining.



LTNCG wallet synchronization is complete, but I postponed it testing, cause I got busy with some other stuffs.
I will test it when Minotaur testing will complete.




When I do solo mining first it shows "new block report" in new format, then it shows in old format and it continues showing in old format.





I think you're right. I'm going to test Ring solo mining again and will post the result after few hours.



You're welcome dude.  
I'm also learning with these testing.  

I think you may find you get both formats if you let it run long enough. I have an idea why but
will wait for your data.


Edit: I did some analysis and I found there may be a conflict with longpoll. With stratum, getwork,
or GBT without longpoll the miner threads request new work by signalling the workio thread.
When longpoll is enabled the longpoll thread requests new work, however, the miner threads via workio
also continue to request new work.

This is why the new block reporting is inconsistent. When workio requests new work the new format is
used, longpoll still uses the old format.

You previously reported an HTTP error that went away when you disabled longpoll. The error may have been
due to the conflict of both workio & longpoll requesting work.

As a result I want to change the focus of testing. Contrary to my last recommendation you should continue
to use -P as well as -D so we can see any HTTP errors.

I would like 2 tests to run long enough for several new blocks to see if they are reported consistenly the same way
and if there are any HTTP errors. Then post the full log files.

1. test with default (longpoll enabled)

2. test with --no-longpoll (disabled)

This is a special test targetted at the log reporting. Regression testing with other algos is still needed.

Thanks for your outstanding help testing.

Found first block of Ring in about 5 hours using cpuminer-opt-3.13.2-segwit-test with 2 threads, while I was doing other stuffs too on my PC.
My CPU is Intel® Core™ i5-2400

No, I didn't do anything different. I will test again and will post the result.

I did solo mining of sha256d algo (LCC) with both version 3.13.1.1 and 3.13.2 and found a Bug.
I will test again with 3.13.1.1 and will post the result.

Now, I'm synchronizing LitecoinCash Gold (LTNCG) wallet.
When syncing will be completed then will do solo mining of yespowerLTNCG algo and will post the result.

There's an inconsistency in your testing. In the above the new block is reported the old way,
but in the first debug log file: https://drive.google.com/file/d/1NWx7BHPhz9IlvIyV0j8eidXq6Yj81VvR/view
it's reported the new way:


Did you do something different? Can you watch this closely in your testing?
-P is no longer necessary but still use -D for testing.

That's great work, thanks.

I'll look into the log issues and wait for regression testing results. If nothing got broken the next build
will be a general release.

I keep it running more than 20 mins, it didn't show 5 mins summary.
But, it shows 5 mins summary when I do pool mining.

I get around 50 kh/s with 1 thread using LitecoinCash cpuminer-multi, but your miner is generating 80-85 kh/s.
It is about 60% more hash rate than LitecoinCash cpuminer-multi.
You can easily compare the hash rate with these screenshots.

cpuminer-opt-3.13.2-segwit-test



LitecoinCash cpuminer-multi



And this is the summary of 3 hours and 20 mins using cpuminer-opt v3.13.1.1. Yesterday, I did pool mining with 1 thread to see the performance of your miner.



My CPU is i5 2400
Intel® Core™ i5-2400 Specifications

That was unexpected. I guess if you prepare for the worst it doesn't happen.

This is the first time anyone has confirmed to me that GBT works in cpuminer-opt.
I just kept my hands off it and waited for users to complain.

I see there's still some work to be done, GBT still uses the old "new block" log.
I also didn't see a 5 minute summary log, did you run long enough? Without the new
"new block" log that's the only way you'll see a hash rate. It would be nice to know
how the performance compares. The hash rate tends to vary so it might be difficult to compare.
Please include your CPU model.

I've been thinking about Discord but will wait, development has slowed considerably. The code has
been optimized for the latest x86_64 CPUs and many new algos are incompatible with cpuminer.
I don't see a lot of activity in the immediate future. If that changes I may reconsider.

JayDDee, it will be great if you join LitecoinCash (LCC) or/& Ring discord, then we can easily chat and discuss errors/bugs.

LCC discord:  https://discord.gg/3AnFpvy
Ring discord:  https://discord.gg/Qd2wGkP

If you don't want to join discord then please ignore this post.

Congratulations dude, it's working. 

I tested Minotaur algo with flag `-t 1` `-t 2` `-t 3` and without flag `--threads=n`.
It worked without any issue.

Here is the debug log file.
https://drive.google.com/open?id=1NWx7BHPhz9IlvIyV0j8eidXq6Yj81VvR




I will check other algo one by one and will post the results here.

cpuminer-opt-3.13.2-segwit-test is released.

Adds support for Segregated Witness (segwit) to GetBlockTemplate.

https://github.com/JayDDee/cpuminer-opt/releases/tag/v3.13.2-segwit-test

This is a test release not intended for production use. Users mining solo are invited to test this release and
report their results. Other users may want to wait for the next general release before upgrading.

Test notes: Minotaur algo (Ring coin) should be tested against the litecoincash fork of cpuminer-multi. Other algos should
be tested against cpuminer-opt-3.13.1.1 for any regressions (things that worked in v3.13.1.1 but not in v3.13.2).
Testing the new segwit support for coins other than Ring should probably wait until it's known to work with Ring.

I don't think those are related to longpoll, it looks more like a networking issue. They didn't seem to cause any serious
problems, it reconnected and continued on and submitted valid blocks. If the're only displayed with -P it's likely because they
can be ignored.

I got the following errors

This is the same error that OBTfx received. You can see it in his shared files.
https://drive.google.com/file/d/1hagUcLp4QxYphdB6YIFHC-9Sf9_D3dxY

What errors?

Excellent! In the future you can cut the repeated block emmisions to reduce the size. Always show everything up to the first block,
any errors and the result of any blocks accepted or rejected.

I have made the code changes and it compiles. I want to review it again just in case I missed something. I'll be posting a
test release on github when it's ready for testing.

Today, when I did solo mining first time with this miner adding flag -D -P then I got those error too.
Then added `--no-longpoll` in my batch file and restarted my miner.
Then I didn't get any error.
https://drive.google.com/open?id=1xAhgO8r8UFTPyi33LiK5Cmb27hIwGzbI

Do not allow sending files (in the end it is good), I leave the file in the cloud here; I hope it serves what you want to look at. 

https://drive.google.com/open?id=1hagUcLp4QxYphdB6YIFHC-9Sf9_D3dxY

Both things confirmed, I have the output here I will try to send it to you by direct message.
The tests were against the testnet, so I am not struggling with the current difficulty of the chain.

I will not mind to get instructions from you.  


I have windows binaries of this miner.
https://drive.google.com/open?id=1RiK5nzi-M6NlemiUsb0BC9yCoQ-GBUtJ
I did a lot of solo mining with this miner in testnet of Ring and found hundreds of blocks without any issue.
I also did solo mining with this in mainnet without any issue, but didn't able to find any block.
Cause my PC is old and this miner generates low hash rate compare to others.
And with this hash rate it will take 2-3 days to find a block if I leave the miner running 24 hours a day.

Edit: I also did pool mining of Ring with this miner without any issue.


Okay.

Ok, we have this controlled, yes I confirm in advance that we have mined it in the past with the LitecoinCash cpuminer-multi alone;
Anyway, I execute an iteration with the flags that you indicate me;
I save the result to you in a plain text file and send it to you.

Thanks for the explanation of the Getwork and GBT, it makes everything very clear