[ANN] cpuminer-opt v24.5, Optimized multi-algo CPU miner for x86_64 and AArch64

[Erdloch]

Hello together!

First of all, thanks to JayDDee / joblo for your fork of cpuminer-multi, I think it's pretty cool (when I finally can make it run ;-))
I would like to use this piece of software to mine for Nicehash.com
I have just compiled the latest version of cpuminer-opt from Github on my Ubuntu server and ran the following start-command:

./cpuminer -a scrypt -o stratum+tcp://scrypt.eu-west.nicehash.com:3333 -u 3GoaYEXEYLJ8Kri6apdt6yZsvSvHdqDksc.testminer1 -p x

For me, it seems like the program is running properly, and I can also see the machine appeard in the menu of Nicehash, but always with "Accepted speed: 0.0000 MH/s".
When I run htop, I can see my CPUs are at 100%.

After some time, I get this report, which also confirms that "0" has been submitted / accepted.

Does anyone has an idea of what I'm doing wrong? I have also tried other algorithms, but with the same result...
Any help appreciated!



Report:

[2021-02-28 07:56:31] scrypt: scrypt.eu-west.nicehash.com:3333
                      Periodic Report     5m49s        18m45s
                      Share rate        0.00/min     0.00/min
                      Hash rate          0.00h/s      0.00h/s   (57.72kh/s)
                      Submitted             0            0
                      Accepted              0            0
                      Hi/Lo Share Diff  0 /  9e+99

Log after the run command:

        **********  cpuminer-opt 3.15.6  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AVX512, SHA and VAES extensions by JayDDee.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

[2021-02-28 07:37:46] Scrypt paramaters: N= 1024, R= 1.
CPU: Intel(R) Xeon(R) CPU E5-2640 v4 @ 2.40GHz
SW built on Feb 28 2021 with GCC 9.3.0
CPU features:  AVX2    AES
SW features:   AVX2    AES
Algo features: AVX2

Starting miner with AVX2...

[2021-02-28 07:37:46] Stratum connect scrypt.eu-west.nicehash.com:3333
[2021-02-28 07:37:46] Miner thread priority 0 (nice 19)
[2021-02-28 07:37:46] 4 of 4 miner threads started using 'scrypt' algorithm
[2021-02-28 07:37:46] Stratum extranonce1= 000020551b, extranonce2 size= 3
[2021-02-28 07:37:46] Extranonce subscription enabled
[2021-02-28 07:37:46] Stratum connection established
[2021-02-28 07:37:46] New Stratum Diff 135033, Block 12505396, Job 00000000f7c3c61f
                      Diff: Net 1.2374e+05, Stratum 1.3503e+05, Target 2.0604
[2021-02-28 07:37:47] CPU temp: curr 0 C (max 0), Freq: 0.000/0.000 GHz
[2021-02-28 07:37:49] New Block 12505397, Job 00000000f7c3c620
                      Diff: Net 1.1761e+05, Stratum 1.3503e+05, Target 2.0604
                      TTF @ 50.90 kh/s: Block 8103y73d, Share 2d00h
                      Net hash rate (est) 252.56 Th/s
[2021-02-28 07:37:54] New Block 12505398, Job 00000000f7c3c621
                      Diff: Net 1.1251e+05, Stratum 1.3503e+05, Target 2.0604
                      TTF @ 56.37 kh/s: Block 8103y73d, Share 1d19h
                      Net hash rate (est) 120.81 Th/s
[2021-02-28 07:38:04] New Block 12505399, Job 00000000f7c3c622
                      Diff: Net 1.0791e+05, Stratum 1.3503e+05, Target 2.0604
                      TTF @ 57.29 kh/s: Block 8103y73d, Share 1d18h
                      Net hash rate (est) 77.24 Th/s

Thank you!

[Nighthog]

Hello together!
---------------
Thank you!

Do not mine scrypt with a cpu... not even a GPU is worthwhile.

Basically it's so slow it never finds a share it can submit.
Use another algorithm for the other coins, specifically the CPU coins. At least you can get something for your effort there.

[Erdloch]

Okay, that makes sense.
Can you suggest any algorithm where CPU-mining still makes sense and that also works with cpuminer-opt?

[Nighthog]

Okay, that makes sense.
Can you suggest any algorithm where CPU-mining still makes sense and that also works with cpuminer-opt?

yescryptR32 [DMS]
yespowerR16 [YTN]
yespower [several]

There are many more choices, these are just some I've used with this cpu-miner for now.

[JayDDee]

[2021-02-28 07:38:04] New Block 12505399, Job 00000000f7c3c622
                      Diff: Net 1.0791e+05, Stratum 1.3503e+05, Target 2.0604
                      TTF @ 57.29 kh/s: Block 8103y73d, Share 1d18h
                      Net hash rate (est) 77.24 Th/s

Take a look at the TTF, you can expect about 1 share every 2 days mining scrypt with that CPU.
Share TTF should be less than 1 minute for useful mining. Anything more than 5 minutes
will likely cause stratum timeouts.

[Nighthog]

Any plans to add RX2 algorithm for this? [LUX]

It's quite recent and was looking if it could be incorporated into this multi-miner.

[JayDDee]

cpuminer-opt development is winding down. That might have been obvious by the slower pace of new
releases and less content in each one

I've achieved everything I wanted to from a technical standpoint. All of the suitable algorithms have been
optimized to support AVX512, SHA and VAES. This brings support to CPUs that haven't even been released yet.

Adding new algorithms was never part of my plan but many were easy to integrate into the cpuminer design.
Tha's no longer the case. Most new algorithms are written differently and don't fit well with cpuminer.
Many use c++ and cmake while cpuminer uses c and automake.

The number of different algorithms has increased with more algorithms being used by only one coin making them
more work to support.

Coins are frequently changing algorithms making even more work to support. The days of an all in one cpu miner
are coming to an end. Custom miners are required for many coins. I wouldn't be able to keep up even if I wanted to.

I have 1 more planned release with a few minor code tweaks and some cleanup, but nothing realy new.
After that it's bug fixes only, which hopefully will be few.

I have no plans of going anywhere and I may get bored, so who knows.

[chrysophylax]

cpuminer-opt development is winding down. That might have been obvious by the slower pace of new
releases and less content in each one

I've achieved everything I wanted to from a technical standpoint. All of the suitable algorithms have been
optimized to support AVX512, SHA and VAES. This brings support to CPUs that haven't even been released yet.

Adding new algorithms was never part of my plan but many were easy to integrate into the cpuminer design.
Tha's no longer the case. Most new algorithms are written differently and don't fit well with cpuminer.
Many use c++ and cmake while cpuminer uses c and automake.

The number of different algorithms has increased with more algorithms being used by only one coin making them
more work to support.

Coins are frequently changing algorithms making even more work to support. The days of an all in one cpu miner
are coming to an end. Custom miners are required for many coins. I wouldn't be able to keep up even if I wanted to.

I have 1 more planned release with a few minor code tweaks and some cleanup, but nothing realy new.
After that it's bug fixes only, which hopefully will be few.

I have no plans of going anywhere and I may get bored, so who knows.

Let's chat ...

You never know what that could lead to

#crysx

[JayDDee]

cpuminer-opt-3.16.0

Added verthash algorithm. It's slow but it works.

https://github.com/JayDDee/cpuminer-opt/releases

The following is superseded by v3.16.1.

Mining verthash uses a data file of over 1 GB. It is not included in the download
but may be available for download elsewehere, YMMV.

For cpuminer-opt the verthash data file must be named verthash.dat and located in the same
directory as the cpuminer executable. A Linux symlink works.

The verthash data file must be obtained seperately. If you already use VerthashMiner you can
simply copy or link the existing data file to the cpuminer directory, using the required name.

Otherwise it may be created using https://github.com/CryptoGraphics/VerthashMiner/releases
following the instructions. A GPU is not necessary to create the file.

The same data file can be used by both cpuminer and VerthashMiner simultaneously.

Launching cpuminer to mine verthash is the same as any other algorithm, no extra options are required.

[JayDDee]

cpuminer-opt-3.16.1

https://github.com/JayDDee/cpuminer-opt/releases

New options for verthash:
  --data-file to specify the name, and optionally the path, of the verthash
              data file, default is "verthash.dat" in the current directory.
  --verify to perform the data file integrity check at startup, default is
           not to verify data file integrity.

Support for creation of default verthash data file if:
   1) --data-file option is not used,
   2) no default data file is found in the current directory, and,
   3) --verify option is used.

More detailed logs related to verthash data file.

Small verthash performance improvement.

Fixed detection of corrupt stats caused by networking issues.

Quickstart for Verthash:

First time mining verthash or don't have a Verthash data file:

--algo verthash --verify --url ...

Verthash data file already exists:

--algo verthash --data-file /path/to/verthash.dat --url ...

See verthash-help.txt for more details.

[trogdorjw73]

Is there any chance AVX512 with Rocket Lake would be worthwhile for mining Ethereum / Ethash? I saw this post and am trying to corroborate their 'findings': https://www.hardwaretimes.com/intel-core-i9-11900k-reportedly-on-par-w-the-nvidia-rtx-3060-in-ether-mining-courtesy-of-avx512/

[JayDDee]

Is there any chance AVX512 with Rocket Lake would be worthwhile for mining Ethereum / Ethash? I saw this post and am trying to corroborate their 'findings': https://www.hardwaretimes.com/intel-core-i9-11900k-reportedly-on-par-w-the-nvidia-rtx-3060-in-ether-mining-courtesy-of-avx512/

It's interesting but it was 5 GHz, not very practical. My i9-9940x slows from 4.3 GHz (all cores) to 3.5 when running AVX512.

I've been trying to vectorize verthash and AVX512 is slower than AVX2. I presume the significantly lower clock rate
slows down data file access, negating any gains from code optimization.
I also presume eth would have the same issues with the DAG.

I'm also not aware of any usable CPU miners for Eth.

[alucard20724]

Is there any chance AVX512 with Rocket Lake would be worthwhile for mining Ethereum / Ethash? I saw this post and am trying to corroborate their 'findings': https://www.hardwaretimes.com/intel-core-i9-11900k-reportedly-on-par-w-the-nvidia-rtx-3060-in-ether-mining-courtesy-of-avx512/

It's interesting but it was 5 GHz, not very practical. My i9-9940x slows from 4.3 GHz (all cores) to 3.5 when running AVX512.

I've been trying to vectorize verthash and AVX512 is slower than AVX2. I presume the significantly lower clock rate
slows down data file access, negating any gains from code optimization.
I also presume eth would have the same issues with the DAG.

I'm also not aware of any usable CPU miners for Eth.

the 11900K runs hot with AVX512 under constant load.. i was getting 100C steady with a 360mm AIO system (corsair h150i).
I had to stop running avx512 for the time being.

I've got a cooler master sub-zero system (thermoelectric cooling system) from newegg showing up in a few days... will put that in the system and try avx512 again.
https://www.newegg.com/cooler-master-liquid-cooling-system-masterliquid-ml360-sub-zero/p/N82E16835103310?Item=N82E16835103310

[JayDDee]

the 11900K runs hot with AVX512 under constant load.. i was getting 100C steady with a 360mm AIO system (corsair h150i).
I had to stop running avx512 for the time being.

I've got a cooler master sub-zero system (thermoelectric cooling system) from newegg showing up in a few days... will put that in the system and try avx512 again.
https://www.newegg.com/cooler-master-liquid-cooling-system-masterliquid-ml360-sub-zero/p/N82E16835103310?Item=N82E16835103310

Rocketlake is going to be hot, it uses 14nm for an architecture designed for 10nm. It added AVX512 & SHA but
had to leave out VAES and the top end 10 core model.

Performance will depend on the algo and the AVX512 clock penalty, and possibly thermal throttling.

[JayDDee]

cpuminer-opt-3.16.2

Verthash: midstate prehash optimization for all architectures.
Verthash: AVX2 optimization.
GBT: added support for Bech32 addresses, untested.
Linux: added CPU frequency to benchmark log.
Fixed integer overflow in time calculations.


For solo mining with getblocktemplate Bech32 addressing is not tested and segwit is uncertain.
One, the other, or both may not work. Testing is welcome. Questions can be discussed here or
by opening an issue with git. I'll do my best but I can't make any promises.

Please add -D command line option if trying Bech32 or segwit.

[paperoga]

What --param-key to mine Resistance (RES)?

I tried different variants but shares are always rejected. I don't want to use rplant miner

------------
2021-04-14 23:44:38] 10 Submitted Diff 1.0876e-06, Block 895384, Job 76bd
[2021-04-14 23:44:38] 10 A0 S0 Rejected 10 B0, 85.333 sec (350ms)
                      Diff 1.0876e-06, Block 895384, Job 76bd
[2021-04-14 23:44:38] Reject reason: incorrect size of nonce 8RES
                      Hash:   000e07be624feb118000000000000000ffffffffffffffff
                      Target: 00140000000000000000000000000000ffffffffffffffff
------------

Coin is in this pool: https://pool.rplant.xyz/

[JayDDee]

What --param-key to mine Resistance (RES)?

I tried different variants but shares are always rejected. I don't want to use rplant miner

------------
2021-04-14 23:44:38] 10 Submitted Diff 1.0876e-06, Block 895384, Job 76bd
[2021-04-14 23:44:38] 10 A0 S0 Rejected 10 B0, 85.333 sec (350ms)
                      Diff 1.0876e-06, Block 895384, Job 76bd
[2021-04-14 23:44:38] Reject reason: incorrect size of nonce 8RES
                      Hash:   000e07be624feb118000000000000000ffffffffffffffff
                      Target: 00140000000000000000000000000000ffffffffffffffff
------------

Coin is in this pool: https://pool.rplant.xyz/

Short answer is I don't know, I don't try to find every coin's parameters.

Usually the coin developpers should publish the POW specifications including parameters.
A pool that supports the coin will often have mining instructions for the coins it supports
which should include parameters if they are required.

However, I'm concerned about the reject reason in this case. I would expect "invalid share"
if the parameters were incorrect. But the reason suggests a nonce size issue, not a parameter
issue. The reason also includes the coin's symbol suggesting something specific to RES.

I've never seen this error before, there may be something unique about the implementation
of the stratum protocol used by this coin that cpuminer-opt doesn't support. If so I don't know what it is.

If the coin has a thread in the ANN section I suggest asking there. You could also ask rplant,
he should know. His pool, and presumably his miner, support it.

You could also add -D to the command line and note the log messages at startup related to stratum,
That may provide a clue.

[paperoga]

-D added

        **********  cpuminer-opt 3.16.2  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AVX512, SHA and VAES extensions by JayDDee.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

[2021-04-15 03:32:19] High priority mining threads may cause system instability
[2021-04-15 03:32:19] Yespower parameters: N= 2048, R= 32
[2021-04-15 03:32:19] Key= "yespowerres"

CPU: AMD G-T56N Processor
SW built on Apr 14 2021 with GCC 9.3.0
CPU features:  SSE2 
SW features:   SSE2 
Algo features: SSE2   SHA

Starting miner with SSE2...

[2021-04-15 03:32:19] Creating stratum thread
[2021-04-15 03:32:19] Stratum connect stratum-eu.rplant.xyz:17040
[2021-04-15 03:32:19] Miner thread priority 4 (nice -10)
[2021-04-15 03:32:19] Binding thread 0 to mask 0000000000000003
[2021-04-15 03:32:19] 2 of 2 miner threads started using 'yespower' algorithm
[2021-04-15 03:32:19] Binding thread 1 to mask 0000000000000003
[2021-04-15 03:32:20] Stratum session id: 322239241633170
[2021-04-15 03:32:20] Stratum extranonce1= 4ffff344, extranonce2 size= 4
[2021-04-15 03:32:21] Method extranonce.subscribe is not supported
[2021-04-15 03:32:21] Threads restarted for new work.
[2021-04-15 03:32:21] Stratum connection established
[2021-04-15 03:32:21] CPU temp: curr 74 C max 0, Freq: 0.000/0.000 GHz
[2021-04-15 03:32:21] Threads restarted for new work.
[2021-04-15 03:32:21] New Stratum Diff 0.1, Block 895616, Job 7837
                      Diff: Net 8.0568e-05, Stratum 0.1, Target 1.5259e-06
[2021-04-15 03:32:34] CPU temp: curr 81 C max 74, Freq: 0.000/0.000 GHz
[2021-04-15 03:32:34] Threads restarted for new work.
[2021-04-15 03:32:34] New Work: Block 895616, Net diff 8.0568e-05, Job 7838
[2021-04-15 03:33:18] 1 Submitted Diff 4.7359e-06, Block 895616, Job 7838
[2021-04-15 03:33:18] Thread 0, Nonce 60030000, Xnonce2 00000000
[2021-04-15 03:33:18] Data[0:19]: 04000000 0c6831ef 3bf8d0d5 40fb9d53 0cc3a9e3 ccc8dbbc 42cd98b0 2cf99196 d370e4be 3220f5a9
[2021-04-15 03:33:18]           : 5045e544 6543e0c0 50ba895a 3ca541c9 294c736f 8d509b0f d120af2d d1b37760 bf7b301e 60030000
[2021-04-15 03:33:18] Hash[7:0]: 000338d1 e3f4fd9c f5242b71 d6127636 f35513ac 47901520 00ef2105 0f2f396f
[2021-04-15 03:33:18] Targ[7:0]: 000a0000 00000000 00000000 00000000 ffffffff ffffffff ffffffff ffffffff
[2021-04-15 03:33:18] 1 A0 S0 Rejected 1 B0, 58.369 sec (352ms)
                      Diff 4.7359e-06, Block 895616, Job 7838
[2021-04-15 03:33:18] Reject reason: incorrect size of nonce 8RES
                      Hash:   000338d1e3f4fd9d0000000000000000ffffffffffffffff
                      Target: 000a0000000000000000000000000000ffffffffffffffff
[2021-04-15 03:33:18] CPU temp: curr 82 C max 81, Freq: 0.000/0.000 GHz
[2021-04-15 03:33:29] Threads restarted for new work.
[2021-04-15 03:33:29] New Work: Block 895616, Net diff 8.0568e-05, Job 7839

[JayDDee]

I don't see anything obvious in the output. The log about extranonce not being supported is bogus
and will be fixed in the next release.

The only thing that seems certain is there are 2 problems. First is the reject reason. It either has nothing to
do with the algo parameters or the reason itself is in error. The second issue is the algo parameters.
I presume you took a guess at the key/pers.

I can't do anything about the first problem. If the reject reason is legit then cpuminer-opt is incompatible
in some way. If the reject reason is bogus finding and using the correct parameters may solve it.

[paperoga]

any chance you can add parameters for Raptoreum coin?
Unfortunately adding new algos is beyond my capabilities.

It's using Ghostrider algo, infos I found:
https://docs.raptoreum.com/#/howtomine

and this miner:
https://github.com/npq7721/cpuminer-gr