[LOCKED] cpuminer-opt v3.12.3, open source optimized multi-algo CPU miner

[guytechie]

Quiet for a couple days?  Wow... things must be going very well (Maytag).

So far so good over here.  3.8.2.1 has been super.

[1714158483]

1714158483

[Slkp]

Hexx mining(lyra2zoin) 7800x
3.8.0 gives 1200h
3.8.2 only 900h

[joblo]

Hexx mining(lyra2zoin) 7800x
3.8.0 gives 1200h
3.8.2 only 900h

Fixed in next release.

[zabidin2]

Is it possible install this on ubuntu box?

[guytechie]

Is it possible install this on ubuntu box?

I would think so.  You'll have to compile it from source, no?

I did it once, but it was on a VM.  I was hoping to utilize SHA before it was available in the Windows binary.

Of course, SHA acceleration is not exposed to the VM, so it was more of a learning experience on how to compile.

[joblo]

cpuminer-opt-3.8.3

https://github.com/JayDDee/cpuminer-opt/releases/tag/v3.8.3

More restoration of lost lyra2 hash.
8 way AVX2 and 4way AVX optimization for blakecoin, vanilla & blake2s.
8 way AVX2 for lbry.
Scaled hashrate for API output.
A couple of GBT fixes.

[joblo]

Is it possible install this on ubuntu box?

I would think so. 

Not if it has an ARM CPU. x86_64 only!

[felixbrucker]

Scaled hashrate for API output.

Thanks for this (y)

[s0ftcorn]

anyone else getting 100% rejected shares?
pools and algos:
yescrypt.mine.zpool.ca:6233
-a cryptonight -o stratum+tcp://mine.graft.cryptocrush.cc:3333

using cpuminer-avx2.exe on a core i5 7600K

3.8.2 worked flawless

[joblo]

anyone else getting 100% rejected shares?
pools and algos:
yescrypt.mine.zpool.ca:6233
-a cryptonight -o stratum+tcp://mine.graft.cryptocrush.cc:3333

using cpuminer-avx2.exe on a core i5 7600K

3.8.2 worked flawless

Yes, I just discovered this with m7m, I'm investigating.

Edit: It looks like a problem submitting shares by algos that don't use 4 way.

Edit: Fix coming soon.

[joblo]

cpuminer-opt-3.8.3.1

https://github.com/JayDDee/cpuminer-opt/releases/tag/v3.8.3.1

Fixed regression in v3.8.3 that broke several algos.

New in 3.8.3:

More restoration of lost lyra2 hash.
8 way AVX2 and 4way AVX optimization for blakecoin, vanilla & blake2s.
8 way AVX2 for lbry.
Scaled hashrate for API output.
A couple of GBT fixes.

[felixbrucker]

Scaled hashrate for API output.

Thanks for this (y)

oh, nvm, though you changed it to H/s instead of KH/s, but you made it dynamic

[joblo]

Scaled hashrate for API output.

Thanks for this (y)

oh, nvm, though you changed it to H/s instead of KH/s, but you made it dynamic

It scales up  at 10,000, just like the console display. It makes parsing the output a little trickier,
you'll need to test the scale prefix.

I had to do this because some algos (since 3.8.3) can now reach over 100000 kH/s on a mainstream CPU.
It also gives more precision for very slow algos.

[felixbrucker]

as this is api, why not just output H/s (a large number for fast algos), gives most fine granular control over display and formatting etc

for display it makes sense to scale it, but not for api

[Nokedli]

Code:

./cpuminer -a yescryptr16 -o http://192.168.1.150:49751 -u XY -p XY1 --no-stratum

         **********  cpuminer-opt 3.8.2.1  *********** 
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700X Eight-Core Processor         .
SW built on Feb 18 2018 with GCC 7.3.0.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 AVX SHA.
Start mining with AVX SHA.

[2018-02-23 22:17:46] 16 miner threads started, using 'yescryptr16' algorithm.
[2018-02-23 22:17:47] Current block is 83506
[2018-02-23 22:17:47] No payout address provided, switching to getwork
[2018-02-23 22:19:30] CPU #0: 4096 H, 40.29 H/s
[2018-02-23 22:19:32] yescryptr16 block 83507, diff 0.11, net 3051.57 kH/s
[2018-02-23 22:19:32] CPU #10: 2419 H, 28.93 H/s
[2018-02-23 22:19:32] CPU #14: 1988 H, 26.30 H/s
[2018-02-23 22:19:32] CPU #13: 2221 H, 28.62 H/s
[2018-02-23 22:19:32] CPU #11: 2362 H, 28.94 H/s

Code:

./cpuminer -a yescryptr16 -o http://192.168.1.150:49751 -u XY -p XY1 --no-stratum

         **********  cpuminer-opt 3.8.3.1  *********** 
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX2 and SHA extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT

CPU: AMD Ryzen 7 1700X Eight-Core Processor         .
SW built on Feb 23 2018 with GCC 7.3.0.
CPU features: SSE2 AES AVX AVX2 SHA.
SW features: SSE2 AES AVX AVX2 SHA.
Algo features: SSE2 AVX SHA.
Start mining with AVX SHA.

[2018-02-23 22:17:18] 16 miner threads started, using 'yescryptr16' algorithm.
[2018-02-23 22:17:19] Current block is 83506
[2018-02-23 22:17:19] No payout address provided, switching to getwork
Segmentation fault

Solo mining looks bad, only for me?
Dual E5-2640 similar error, Manjaro 17 on both

[joblo]

Solo mining looks bad, only for me?
Dual E5-2640 similar error, Manjaro 17 on both

Possibly not. If you can compile please replace function std_longpoll_rpc_call in file
cpu-miner.c with this and let me know what happens.

Code:

json_t *std_longpoll_rpc_call( CURL *curl, int *err, char* lp_url )
{
   json_t *val;
   char *req = NULL;
   if (have_gbt)
   {
       req = (char*) malloc( strlen(gbt_lp_req) + strlen(lp_id) + 1 );
       sprintf( req, gbt_lp_req, lp_id );
   }
//TODO this code makes no sense, this first call should be removed.
// also remove conditional expression in second call, no getwork.
//   val = json_rpc_call( curl, rpc_url, rpc_userpass, getwork_req, err,
//                        JSON_RPC_LONGPOLL );
   val = json_rpc_call( curl, lp_url, rpc_userpass, req ? req : getwork_req,
                        err, JSON_RPC_LONGPOLL);
//   val = json_rpc_call( curl, lp_url, rpc_userpass, req,
//                        err, JSON_RPC_LONGPOLL );
   free(req);
   return val;
}

[wildzer0]

The API change breaks Awesome Miner support. Can a compatibility mode be added or is an Awesome Miner update needed?

[Nokedli]

still the same error

[joblo]

The API changes were a request, but it seems the old way was preferred. I'll revert the change
if the majority want.

Edit: there are more changes to the API coming, adding solved block count.

[joblo]

still the same error

OK, try going back to block version 3:

Code:

#define BLOCK_VERSION_CURRENT 3

Edit: if that doesn't work go back to std_longpoll_rpc_call from v3.8.2. Those were the only
changes I made that affect solo mining and I made no changes to yescrypt.