[ANN] cpuminer-opt v24.2, Optimized multi-algo CPU miner for x86_64 and AArch64

[JayDDee]

Hi,
I'm sorry if this not the correct location for this, but i would like to suggest or to know if the algo verushash 2.0 can be added to cpuminer. That would be great, since seems to be a good coin to be mined with cpu's.

Thanks,

The existing miner is written in c++ and would be too difficult to import.

[minerja]

Hi,
I'm sorry if this not the correct location for this, but i would like to suggest or to know if the algo verushash 2.0 can be added to cpuminer. That would be great, since seems to be a good coin to be mined with cpu's.

Thanks,

The existing miner is written in c++ and would be too difficult to import.

Shame, cos it's a great coin to mine, but the team haven't bothered optimising the miners for nearly 12 months.
I'm sure yours would be much better

[JayDDee]

Hi,
I'm sorry if this not the correct location for this, but i would like to suggest or to know if the algo verushash 2.0 can be added to cpuminer. That would be great, since seems to be a good coin to be mined with cpu's.

Thanks,

The existing miner is written in c++ and would be too difficult to import.

Shame, cos it's a great coin to mine, but the team haven't bothered optimising the miners for nearly 12 months.
I'm sure yours would be much better

It looks like it's pure AES and is already using AES_NI instructions. VAES is possible but that would be limited to
2 ways on Ryzen 5xxx. 4 way VAES is only available on a few Intel laptop CPUs with VAES & AVX512.

It's hard to estimate the gain due to the larger data load, extra data manipulation and lower clock rate. Theoretically
2 ways would be in the range of 1.5x and 4 ways would around 2.5x.

If it's only a matter of importing the Haraka hash function it could be a trivial addition.

I'll give it a quick test, but if it doesn't work as is I'll give up because I wouldn't know where to start debugging.

Edit: It won't work, the stratum protocol is different.

[BoozyTalking]

Hi, is it possible to somehow compile it on android?

[JayDDee]

Hi, is it possible to somehow compile it on android?

The first post lists the requirements. Clearly the answer is no.

Apple and Android operating systems are not supported.

[legcramp]

Can someone point me in the right direction. I have multiple machines that I plan to use for mining Raptoreum, the question is do I have to install and sync a separate Raptoreum core / wallet for each machine?

Or is it simply sync and create the wallet on the main machine and just run the miner off the other machines and redirecting the rewards to the wallet address?

[JayDDee]

Can someone point me in the right direction. I have multiple machines that I plan to use for mining Raptoreum, the question is do I have to install and sync a separate Raptoreum core / wallet for each machine?

Or is it simply sync and create the wallet on the main machine and just run the miner off the other machines and redirecting the rewards to the wallet address?

This isn't the right place to ask, but you only need one wallet.

[legcramp]

Can someone point me in the right direction. I have multiple machines that I plan to use for mining Raptoreum, the question is do I have to install and sync a separate Raptoreum core / wallet for each machine?

Or is it simply sync and create the wallet on the main machine and just run the miner off the other machines and redirecting the rewards to the wallet address?

This isn't the right place to ask, but you only need one wallet.

Ahhh thank you, didn't realize this isn't the official thread for the coin.

[JayDDee]

Here's a plug for a useful link on some of the latest malware phishing techniques using fake github accounts,
and some tips. I have also seen where a legitimate github user was hacked (lax security settings) and his wiki
hijacked to link to malware. A phishing link to it was then posted. The legitimate githiub account made the phishing
link look more legitimate.

Allways verify links are legitimate, hover over the link to show the actual destination. Look for subtle spellling
changes for familiar sites, "I" changed to "1" etc.

Direct links to files are dangerous and should be avoided, always go to the download directory to confirm you
are in the right place before you download a file.

Stay safe.

https://bitcointalk.org/index.php?topic=5182222.0

[JayDDee]

cpuminer-opt-3.18.0

See the first post in this thread for download links.

Release notes:

Complete rewrite of Scrypt code, optimized for large N factor (scryptn2):
  - AVX512 & SHA support for SHA256, AVX512 has priority,
  - up to 50% increase in hashrate,
  - memory requirements reduced 30-60% depending on CPU architecture,
  - memory usage displayed at startup,
  - scrypt, default N=1024 (LTC), will likely perform slower.

Improved stale share detection and handling for Scrypt with large N factor:
  - abort and discard partially computed hash when new work is detected,
  - quicker response to new job, less time wasted mining stale job.

Improved stale share handling for all algorithms:
  - report possible stale share when new work received with a previously
    submitted share still pending,
  - when new work is detected report the submission of an already completed,
    otherwise valid, but likely stale, share,
  - fixed incorrect block height in stale share log.

Small performance improvements to sha, bmw, cube & hamsi for AVX512 & AVX2.

When stratum disconnects miner threads go to idle until reconnected.

Colour changes to some logs.

Some low level function name changes for clarity and consistency.

The reference hashrate in the summary log and the benchmark total hashrate
are now the mean hashrate for the session.

Aditional details, tips and suggested testing:

A lot has changed so  lot of testing would be helpful.

The first tip is for Ryzen users and scryptn2. Ryzen can mine scryptn2 using SHA or AVX2. The current default for
is to use SHA. Users may want to try an AVX2 build to see if that gets better performance.
In future release I may change the default to use AVX2. For Rocket lake CPUs the default is to use AVX512.

Hash rate reporting has significant changes. The reference hash rate in the summary may be slghtly lower but
it will be more accurate and directly comparable with the effective hash rate for the session. The longer the
session the more they converge. The previous hash rate calculation is still reported in the TTF of the New Block log
for comparison. The old way didn't account for overhead such as the stratum thread or other applications or
system processes running. The old way is also used for thread hash rates displayed with the --hash-meter option.

The differences also show why miner wars fighting over who has a slightly faster hash rate is a waste of time.
The only thing that matters is what the pool sees.

Changes were made to stale share detection and handling, in particular for scryptn2. It's notorious for stale shares
for the simple reason it takes nearly a full second to calculate hash. I've found the processing bottlenecks and
placed checkpints at strategic places to check for new jobs that will cause the current hash to be aborted and work
will begin with the new job. This should reduce the number of stale shares.

In addition some new logs are introduced related to stale shares that apply to all aggos. One is output when a new
job is received with a submitted share still pending a response. In this situation the share may be accepted or
rejected, there's no way to know it it arrived at the server in time. There is nothing that the miner can do about this
except report it.

The other new log is displayed when new work is deteced but a fresh valid share is in hand. All the hard work has
already been done so it might as well be submitted. These have a higher likelyhood of being stale unless the pool
offers a grace period where it accepts late shares. I have seen shares accepted after this log so I'm hesitant to
discard them and possibly waste a perfectly good share for the sake of keeping stale numbers low.

Scryptn2 now uses much less memory. In addition to organic improvements to the code it may be possible for
CPUs that were previously limited in the number of threads they could run to be able to run more threads
and possibly increase performance further.

Please test and report your results.
 

[southmining.eu]

I would like to ask to you how to pass the founder reward parameter to cpuminer (charity address). thanks

[JayDDee]

I would like to ask to you how to pass the founder reward parameter to cpuminer (charity address). thanks

Thank you for your interest. I'm not sure I understand your question but if you want to make a donation
the simplest way is a direct BTC transaction to the donation address displayed at miner startup. You may
also direct your mining rewards to the donation address if you prefer. Only BTC donations are supported
at this time.
 
cpuminer-opt has no fee of any kind therefore has no built in mechanism for collecting one. Donations are
purely voluntary and made at the discretion of the users.

[southmining.eu]

My question was mostly focused on solving this problem...

Code:

[2021-10-01 21:00:00] CPU affinity [!!!!!!!!!!!!!!!!]
[2021-10-01 21:00:00] API listnening to 127.0.0.1:49999
[2021-10-01 21:00:00] 10 of 16 miner threads started using 'scrypt' algorithm
[2021-10-01 21:00:01] CPU temp: curr 16 C max 0, Freq: 0.800/4.697 GHz
[2021-10-01 21:00:01] New Block 1740237, Net Diff 0.00024414, Ntime b05a5761
                      Miner TTF @ 0.00 h/s NA, Net TTF @ 7507.86 kh/s 0m00s
[2021-10-01 21:00:12] 1 Submitted Diff 0.00036104, Block 1740237, Ntime bb5a5761
[2021-10-01 21:00:12] 1 A0 S0 Rejected 1 B0, 11.746 sec (2ms)
                      Diff 0.00036104, Block 1740237
[2021-10-01 21:00:12] Reject reason: no-founder-reward
                      Hash:   00000ad1ce379ee7d100000000000000ffffffffffffffff
                      Target: 00000fffff00000000000000000000000000000000000000
[2021-10-01 21:00:14] 2 Submitted Diff 0.00031864, Block 1740237, Ntime bc5a5761
[2021-10-01 21:00:14] 2 A0 S0 Rejected 2 B0, 2.119 sec (2ms)
                      Diff 0.00031864, Block 1740237
[2021-10-01 21:00:14] Reject reason: no-founder-reward
                      Hash:   00000c4256735adad000000000000000ffffffffffffffff
                      Target: 00000fffff00000000000000000000000000000000000000

[JayDDee]

My question was mostly focused on solving this problem...

I've never seen that error, I don't know what it means. Please post more info, what coin, command line
arguments, start up messages? Also please add -D -P options.

Edit: I assume this was with v3.18.0, does it also occur on v3.17.1. I would assume so since I didn't change anything
in the protocol.

[JayDDee]

Intel Alder Lake, I'm not impressed.

It looks like Alder Lake desktop CPUs won't have AVX512 or VAES which is a step backward for Intel and
puts them behind AMD. which supports 256 bit VAES on zen3.

Alder Lake will use a big-little architecture with a combination of big powerfull cores with smaller more efficient cores.
AVX512 is being disabled on the big cores to provide a consistent instruction set with the small cores which don't support
it. If AMD adds AVX512 on Zen4 they will go 2 steps ahead of Intel, just in architecture.

Alder Lake will finally ditch the 14nm process but Zen4 will likely involve another die shrink keeping AMD ahead of Intel
in that department as well. If Zen4 also includes PCIe5 and DDR5 it will be clearly ahead or equal with Intel in every department.
That hasn't happened since AMD was first with 64 bit desktop CPUs, Debian packages still use the amd64 name. Zen4 could also
have some surprises like another boost in core count or bigger cache.

I really don't see a fit for big-little architecture. On the desktop the small cores are pretty useless and on a laptop I  prefer more
smaller more efficient cores than the big cores that are either sitting idle or draining the battery.

I don't think I'll be buying any Intel CPUs for a while.

[minerja]

My question was mostly focused on solving this problem...

Code:

[2021-10-01 21:00:00] CPU affinity [!!!!!!!!!!!!!!!!]
[2021-10-01 21:00:00] API listnening to 127.0.0.1:49999
[2021-10-01 21:00:00] 10 of 16 miner threads started using 'scrypt' algorithm
[2021-10-01 21:00:01] CPU temp: curr 16 C max 0, Freq: 0.800/4.697 GHz
[2021-10-01 21:00:01] New Block 1740237, Net Diff 0.00024414, Ntime b05a5761
                      Miner TTF @ 0.00 h/s NA, Net TTF @ 7507.86 kh/s 0m00s
[2021-10-01 21:00:12] 1 Submitted Diff 0.00036104, Block 1740237, Ntime bb5a5761
[2021-10-01 21:00:12] 1 A0 S0 Rejected 1 B0, 11.746 sec (2ms)
                      Diff 0.00036104, Block 1740237
[2021-10-01 21:00:12] Reject reason: no-founder-reward
                      Hash:   00000ad1ce379ee7d100000000000000ffffffffffffffff
                      Target: 00000fffff00000000000000000000000000000000000000
[2021-10-01 21:00:14] 2 Submitted Diff 0.00031864, Block 1740237, Ntime bc5a5761
[2021-10-01 21:00:14] 2 A0 S0 Rejected 2 B0, 2.119 sec (2ms)
                      Diff 0.00031864, Block 1740237
[2021-10-01 21:00:14] Reject reason: no-founder-reward
                      Hash:   00000c4256735adad000000000000000ffffffffffffffff
                      Target: 00000fffff00000000000000000000000000000000000000

Hi,
I've only seen this once or twice before.
From memory it is a coin where the block gets split between the "miner" and the "founder", ie if a block is say 25 coins, the "miner" gets 20 coins, and the "founder" gets 5 coins.
When i have seen this you MUST use the wallet inbuilt miner (usually console, and "setgenerate true x") all external miners only allow to mine to 1 address.
Knowing the coin would really help us. Send me a PM if you don't want to publicly name the coin.

Jaydee, there is code out their that allows the miner to "split" the rewards, if i come across it again i will send you a link.
I think the "miner" probably had an internal "pool" written, making the reward split much easier

[JayDDee]

Hi,
I've only seen this once or twice before.
From memory it is a coin where the block gets split between the "miner" and the "founder", ie if a block is say 25 coins, the "miner" gets 20 coins, and the "founder" gets 5 coins.
When i have seen this you MUST use the wallet inbuilt miner (usually console, and "setgenerate true x") all external miners only allow to mine to 1 address.
Knowing the coin would really help us. Send me a PM if you don't want to publicly name the coin.

Jaydee, there is code out their that allows the miner to "split" the rewards, if i come across it again i will send you a link.
I think the "miner" probably had an internal "pool" written, making the reward split much easier

Thanks for the info. I couldn't find anything relevant with Google, just coins claiming they have no founder reward.
Yet I don't how the miner would be involved with the founder reward, It has no knowledge of the block reward or how
it's shared and such knowledge would not be of any use anyway.

It seems an intentional way to limit mining. If that's the case I'm not interested. I'm also interested in any coin with a
founder reward. The founders can afford to pay developpers, they won't get aything free from me.

[otacilioricardo]

Hello.

Is it possible to run Cpuminer on all nodes of a cluster? I intend to set up a cluster to work, and I want to take advantage of mining.

What or would you like to do is:

1. I have main node;
2. Node 1 (no1);
3. Node 2 (no2);
4. Node 3 (no3).

I set it to run on the main node, and use nodes 1, 2 and 3 to mine.

Thank you for your attention.

Olá.

É possível rodar o Cpuminer em todos os nós de um cluster? Pretendo montar um cluster para trabalhar, e quero aproveitar minerar.

O que ou gostaria de fazer é:

1. Tenho nó principal;
2. Nó 1 (no1);
3. Nó 2 (no2);
4. Nó 3 (no3).

Coloco para rodar no nó principal, e uso os nós 1, 2 e 3 para minerar.

Obrigado pela sua atenção.

[JayDDee]

Hello.

Is it possible to run Cpuminer on all nodes of a cluster? I intend to set up a cluster to work, and I want to take advantage of mining.

First you should ask your boss if it's permitted, then the system administrator how to do it.

[ThinkBeyondRich]

Hello.

Is it possible to run Cpuminer on all nodes of a cluster? I intend to set up a cluster to work, and I want to take advantage of mining.

Hello.

Is it possible to run Cpuminer on all nodes of a cluster? I intend to set up a cluster to work, and I want to take advantage of mining.

First you should ask your boss if it's permitted, then the system administrator how to do it.

I don't think he said "at" work. He said "to" work.