Author

Topic: [ANN]: cpuminer-opt v3.8.8.1, open source optimized multi-algo CPU miner - page 148. (Read 444067 times)

legendary
Activity: 1510
Merit: 1003
Just for fun, like to build from source ))

Tried your latest 3.4.1 version. I have 2 laptops:
one (1) with core i5 3317U (Ivy Bridge) running win7 64. It has AVX only.
second (2) with core i5 5250U (Broadwell-U) running win8.1 64. It has AVX2.

I only tried Lyra2re on Nicehash.

First of all I built on laptop (1). It was successfull Compile and Run, I get something like 240khs in Lyra2re.

Code:
     Forked from TPruvot's cpuminer-multi with credits
     to Lucas Jones, elmad, palmd, djm34, pooler, ig0tik3d,
     Wolf0, Jeff Garzik and Optiminer.

CPU:        Intel(R) Core(TM) i5-3317U CPU @ 1.70GHz
CPU features: SSE2 AES AVX
SW built on Aug  9 2016 with GCC 4.8.3
SW features: SSE2 AES AVX
Algo features: SSE2 AES AVX AVX2
Start mining with SSE2 AES AVX

[2016-08-09 23:05:39] Starting Stratum on stratum+tcp://lyra2re.eu.nicehash.com:
3342
[2016-08-09 23:05:39] 4 miner threads started, using 'lyra2re' algorithm.
[2016-08-09 23:05:40] Stratum difficulty set to 0.8
[2016-08-09 23:05:40] lyra2re.eu.nicehash.com:3342 lyra2re block 83397
[2016-08-09 23:05:42] CPU #2: 65.54 kH, 57.24 kH/s
[2016-08-09 23:05:42] CPU #1: 65.54 kH, 56.74 kH/s
[2016-08-09 23:05:42] CPU #3: 65.54 kH, 56.74 kH/s
[2016-08-09 23:05:42] CPU #0: 65.54 kH, 56.50 kH/s

Then I copy this fresh compiled miner with needed dlls to laptop 2.
And it didn't even start. It is not showing any errors. Just close and all.
Then I tried to compile your miner on laptop 2.
It was successfull compile but again no run.
I tried to copy second compiled miner from laptop 2 to laptop 1 and tried to run it.
It shows some info
Code:
         **********  cpuminer-opt 3.4.1  ***********
     A CPU miner with multi algo support and optimized for CPUs
     with AES_NI and AVX extensions.
     BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
     Forked from TPruvot's cpuminer-multi with credits
     to Lucas Jones, elmad, palmd, djm34, pooler, ig0tik3d,
     Wolf0, Jeff Garzik and Optiminer.

CPU:        Intel(R) Core(TM) i5-3317U CPU @ 1.70GHz
CPU features: SSE2 AES AVX
SW built on Aug  9 2016 with GCC 4.8.3
SW features: SSE2 AES AVX AVX2
Algo features: SSE2 AES AVX AVX2
Start mining with SSE2 AES AVX

[2016-08-09 22:52:57] Starting Stratum on stratum+tcp://lyra2re.eu.nicehash.com
3342
[2016-08-09 22:52:57] 4 miner threads started, using 'lyra2re' algorithm.
and exit ...
No go ...
That is my experience. Fairly speaking I have no much free time to struggle with it ))
legendary
Activity: 1470
Merit: 1114
cpuminer-opt v3.4.1 is released. It fixes hmq1725 and adds more AVX2 optimizations to the
lyra2 core improving Lyra2RE 34% and Lyra2REv2 11%.

Source code:
https://drive.google.com/file/d/0B0lVSGQYLJIZak0xNVFBX3prMHc/view?usp=sharing
(new link, posted wrong file previously.)

It seems Cryptomining Blog have stopped producing binaries for cpuminer-opt. Windows users
may compile their own using mingw64. Instructions are in README.md.

Thanks to MrPlazu and ctgiant for confirming the hmq1725 bug.
hmq1725 should be a little faster also.
legendary
Activity: 1470
Merit: 1114
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.

You can use https://esp.maxminers.net.

I'm not sure of the last working version

Any update on this?

Maybe you could be a little more patient considering you declined my request for more info.
It will increase the chances I actually fix it next release. I'n not creating an account at maxminers
just for this. If I know what release broke it I will know exactly what broke it and how to fix it.

I'm working on some other optimizations, It'll be a couple of days before the next release.

I'm happy to help fix this. I know that Windows version 3-3-8 and Linux version 3-3-5 is working perfect. I haven't tested all Windows and Linux version.

Let me know if you want me to test something specific.

Thanks for the info. That's not what I expected. Are you sure 3.3.8 works? I made a change to hmq1275 in v3.3.8 and that is what
I suspected broke it. I have made no changes to this algo since. Which version is broken for you?

Edit: I made a change that affected many algos, including hmq1275, in 3.4.0. I tested several of the algos and none of them were
broken by it. I don't think this is what broke hm1275, I still suspect the 3.3.8 change.

The endian change broke hmq1725.  I just tested it on linux and replaced with the old code and it worked again.  I don't know what version you made that change on but I hope this helps isolate the issue for you.

the code I commented out:
Code:
        uint32_t endiandata[20] __attribute__((aligned(64)));
        uint32_t hash64[8] __attribute__((aligned(32)));

        flipend32_array( endiandata, pdata, 19 );
And the code I reverted:
Code:
	uint32_t _ALIGN(32) hash64[8];
uint32_t endiandata[32];

int kk=0;
for (; kk < 32; kk++)
{
be32enc(&endiandata[kk], ((uint32_t*)pdata)[kk]);
};

Thanks that's exactly what I suspected. Fix is on the way..
legendary
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
dont waste your time with this bad algo, their wallet is half broken at all level, and the price is reaching 1 LTC satoshi Wink not tradable so... soon only DOGE markets will be possible
sr. member
Activity: 298
Merit: 250
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.

You can use https://esp.maxminers.net.

I'm not sure of the last working version

Any update on this?

Maybe you could be a little more patient considering you declined my request for more info.
It will increase the chances I actually fix it next release. I'n not creating an account at maxminers
just for this. If I know what release broke it I will know exactly what broke it and how to fix it.

I'm working on some other optimizations, It'll be a couple of days before the next release.

I'm happy to help fix this. I know that Windows version 3-3-8 and Linux version 3-3-5 is working perfect. I haven't tested all Windows and Linux version.

Let me know if you want me to test something specific.

Thanks for the info. That's not what I expected. Are you sure 3.3.8 works? I made a change to hmq1275 in v3.3.8 and that is what
I suspected broke it. I have made no changes to this algo since. Which version is broken for you?

Edit: I made a change that affected many algos, including hmq1275, in 3.4.0. I tested several of the algos and none of them were
broken by it. I don't think this is what broke hm1275, I still suspect the 3.3.8 change.

The endian change broke hmq1725.  I just tested it on linux and replaced with the old code and it worked again.  I don't know what version you made that change on but I hope this helps isolate the issue for you.

the code I commented out:
Code:
        uint32_t endiandata[20] __attribute__((aligned(64)));
        uint32_t hash64[8] __attribute__((aligned(32)));

        flipend32_array( endiandata, pdata, 19 );
And the code I reverted:
Code:
	uint32_t _ALIGN(32) hash64[8];
uint32_t endiandata[32];

int kk=0;
for (; kk < 32; kk++)
{
be32enc(&endiandata[kk], ((uint32_t*)pdata)[kk]);
};
newbie
Activity: 56
Merit: 0
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.

You can use https://esp.maxminers.net.

I'm not sure of the last working version

Any update on this?

Maybe you could be a little more patient considering you declined my request for more info.
It will increase the chances I actually fix it next release. I'n not creating an account at maxminers
just for this. If I know what release broke it I will know exactly what broke it and how to fix it.

I'm working on some other optimizations, It'll be a couple of days before the next release.

I'm happy to help fix this. I know that Windows version 3-3-8 and Linux version 3-3-5 is working perfect. I haven't tested all Windows and Linux version.

Let me know if you want me to test something specific.

Thanks for the info. That's not what I expected. Are you sure 3.3.8 works? I made a change to hmq1275 in v3.3.8 and that is what
I suspected broke it. I have made no changes to this algo since. Which version is broken for you?

Edit: I made a change that affected many algos, including hmq1275, in 3.4.0. I tested several of the algos and none of them were
broken by it. I don't think this is what broke hm1275, I still suspect the 3.3.8 change.

Let me test 3.3.8 on Linux and see if it Works. Can you link to Google docs where I can download 3.3.8. Can only find 3.4.0 now
hero member
Activity: 700
Merit: 500
@joblo

just remembered i was using your miner with optiminers hodl pool with the proxy just fine for several days straight, so it seems indeed to be nh related, i will try to get in touch with them, as a temporary resolution i will most likely just restart the miner if my app detects it is "hanging"

thanks for your help though
legendary
Activity: 1470
Merit: 1114
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.

You can use https://esp.maxminers.net.

I'm not sure of the last working version

Any update on this?

Maybe you could be a little more patient considering you declined my request for more info.
It will increase the chances I actually fix it next release. I'n not creating an account at maxminers
just for this. If I know what release broke it I will know exactly what broke it and how to fix it.

I'm working on some other optimizations, It'll be a couple of days before the next release.

I'm happy to help fix this. I know that Windows version 3-3-8 and Linux version 3-3-5 is working perfect. I haven't tested all Windows and Linux version.

Let me know if you want me to test something specific.

Thanks for the info. That's not what I expected. Are you sure 3.3.8 works? I made a change to hmq1275 in v3.3.8 and that is what
I suspected broke it. I have made no changes to this algo since. Which version is broken for you?

Edit: I made a change that affected many algos, including hmq1275, in 3.4.0. I tested several of the algos and none of them were
broken by it. I don't think this is what broke hm1275, I still suspect the 3.3.8 change.
newbie
Activity: 56
Merit: 0
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.

You can use https://esp.maxminers.net.

I'm not sure of the last working version

Any update on this?

Maybe you could be a little more patient considering you declined my request for more info.
It will increase the chances I actually fix it next release. I'n not creating an account at maxminers
just for this. If I know what release broke it I will know exactly what broke it and how to fix it.

I'm working on some other optimizations, It'll be a couple of days before the next release.

I'm happy to help fix this. I know that Windows version 3-3-8 and Linux version 3-3-5 is working perfect. I haven't tested all Windows and Linux version.

Let me know if you want me to test something specific.
legendary
Activity: 1470
Merit: 1114
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.

You can use https://esp.maxminers.net.

I'm not sure of the last working version

Any update on this?

Maybe you could be a little more patient considering you declined my request for more info.
It will increase the chances I actually fix it next release. I'n not creating an account at maxminers
just for this. If I know what release broke it I will know exactly what broke it and how to fix it.

I'm working on some other optimizations, It'll be a couple of days before the next release.
newbie
Activity: 56
Merit: 0
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.

You can use https://esp.maxminers.net.

I'm not sure of the last working version

Any update on this?
hero member
Activity: 700
Merit: 500
for me it stays disconnected until restarted (that is killed and started again)

socks is just proxying tcp and udp connections, so its not protocol aware, will test when im back


br
legendary
Activity: 1470
Merit: 1114
hmmm i only had this issue if using socks, when connected directly the miner was running fine, socks is setup correctly and working fine with other services as well, will test if this also happens with other miners with nicehash to determine if its a miner or nh problem

will take some time as im leaving for vacations today though

br

Test with socks at other pools as well. If socks is protocol aware (not sure if it is) it may be specific to the stratum+tcp
protocol. But since I have seen stratum errors at nicehash with different miners and without socks I'm leaning more
toward nicehash. This is assuming the problem I'm seeing and yours have the same root cause.

In my case it reconnects promptly so it's not too disruptive.
hero member
Activity: 700
Merit: 500
hmmm i only had this issue if using socks, when connected directly the miner was running fine, socks is setup correctly and working fine with other services as well, will test if this also happens with other miners with nicehash to determine if its a miner or nh problem

will take some time as im leaving for vacations today though

br
legendary
Activity: 1470
Merit: 1114
hi joblo, im having a rather unusual problem

when running the miner (latest version in linux + 3.3.8 on windows tested) and using a socks proxy, after some time the miner loses the connection with

Quote
Stratum connection failed: Could not resolve host: lyra2re.eu.nicehash.com
...retry after 10 seconds

my proxy reports this:

Quote
sockd[11167]: info: block(3): tcp/accept ]: 10.1.0.47.36480 10.1.0.40.1080: error after reading 4 bytes in 0 seconds: eof from local client
sockd[11168]: info: block(3): tcp/accept ]: 10.1.0.47.36482 10.1.0.40.1080: error after reading 4 bytes in 0 seconds: eof from local client

such an error only happend before and directly when starting the miner if i did not specify my proxy was socks by "socks5://"

maybe you can look into it Smiley

br

I'm not sure how much I can help, I know nothing about socks. I also don't think it's a problem with the miner because that
has never been reported by anyone else.

I have seen a single failure on startup at times with other miners so I assume this is an issue with nicehash, or a bug common
to ccminer and cpuminer, either way protocol incompatibility of some kind that only occurs with nicehash..

If using socks causes new symptoms I would suspect a problem with socks and/or nicehash. I you don't see the disconnects at
other pools it may be the same nicehash incompatibility causing both errors.
hero member
Activity: 700
Merit: 500
hi joblo, im having a rather unusual problem

when running the miner (latest version in linux + 3.3.8 on windows tested) and using a socks proxy, after some time the miner loses the connection with

Quote
Stratum connection failed: Could not resolve host: lyra2re.eu.nicehash.com
...retry after 10 seconds

my proxy reports this:

Quote
sockd[11167]: info: block(3): tcp/accept ]: 10.1.0.47.36480 10.1.0.40.1080: error after reading 4 bytes in 0 seconds: eof from local client
sockd[11168]: info: block(3): tcp/accept ]: 10.1.0.47.36482 10.1.0.40.1080: error after reading 4 bytes in 0 seconds: eof from local client

such an error only happend before and directly when starting the miner if i did not specify my proxy was socks by "socks5://"

maybe you can look into it Smiley

br
newbie
Activity: 56
Merit: 0
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.

You can use https://esp.maxminers.net.

I'm not sure of the last working version
legendary
Activity: 1470
Merit: 1114
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.


I can't test this anymore at Suprnova. I have an idea what's wrong, it would help if you can tell me the last version
where it worked. I should easilly be able to restore it.
newbie
Activity: 56
Merit: 0
I'm tring to get this version to Work on Ubuntu 16 with algo HMQ1725.

./cpuminer -a hmq1725 -o stratum+tcp://esp.maxminers.net:5872 -u user.worker -p 123

All I get is this

CPU:            Intel(R) Xeon(R) CPU @ 2.30GHz
CPU features: SSE2 AES AVX AVX2
SW built on Aug  7 2016 with GCC 5.4.0
SW features: SSE2 AES AVX AVX2
Algo features: SSE2 AES AVX AVX2
Start mining with SSE2 AES AVX AVX2

[2016-08-07 17:33:33] Starting Stratum on stratum+tcp://esp.maxminers.net:5872
[2016-08-07 17:33:33] 8 miner threads started, using 'hmq1725' algorithm.
[2016-08-07 17:33:37] Stratum difficulty set to 1
[2016-08-07 17:33:37] esp.maxminers.net:5872 hmq1725 block 521805
[2016-08-07 17:33:38] CPU #1: 6630 H, 25.98 kH/s
[2016-08-07 17:33:38] accepted: 0/1 (0.0%), 6630 H, 25.98 kH/s nooooo
[2016-08-07 17:33:38] reject reason: low difficulty share of 0.000022975398271283902
[2016-08-07 17:33:38] factor reduced to : 0.67
[2016-08-07 17:33:38] CPU #2: 17.57 kH, 25.69 kH/s
[2016-08-07 17:33:38] accepted: 0/2 (0.0%), 24.20 kH, 51.67 kH/s nooooo
[2016-08-07 17:33:38] reject reason: low difficulty share of 0.00007275697698863166
[2016-08-07 17:33:38] factor reduced to : 0.44
[2016-08-07 17:33:39] CPU #4: 32.69 kH, 26.84 kH/s
[2016-08-07 17:33:39] accepted: 0/3 (0.0%), 56.89 kH, 78.51 kH/s nooooo
[2016-08-07 17:33:39] reject reason: low difficulty share of 0.00001634549809734534
[2016-08-07 17:33:39] factor reduced to : 0.30
[2016-08-07 17:33:39] CPU #3: 35.29 kH, 23.53 kH/s
[2016-08-07 17:33:39] CPU #6: 39.41 kH, 26.12 kH/s
[2016-08-07 17:33:39] accepted: 0/4 (0.0%), 131.60 kH, 128.16 kH/s nooooo
[2016-08-07 17:33:39] reject reason: low difficulty share of 0.000017716213142858426
[2016-08-07 17:33:39] factor reduced to : 0.20
[2016-08-07 17:33:39] accepted: 0/5 (0.0%), 131.60 kH, 128.16 kH/s nooooo
[2016-08-07 17:33:39] reject reason: low difficulty share of 0.00004261769515069645
[2016-08-07 17:33:39] factor reduced to : 0.13
[2016-08-07 17:33:39] CPU #1: 34.05 kH, 24.98 kH/s
[2016-08-07 17:33:39] accepted: 0/6 (0.0%), 159.02 kH, 127.16 kH/s nooooo
[2016-08-07 17:33:39] reject reason: low difficulty share of 0.00005388752541995255
[2016-08-07 17:33:39] factor reduced to : 0.09
[2016-08-07 17:33:39] esp.maxminers.net:5872 hmq1725 block 521806
[2016-08-07 17:33:39] CPU #2: 24.05 kH, 23.01 kH/s
[2016-08-07 17:33:39] CPU #6: 3348 H, 15.04 kH/s
[2016-08-07 17:33:39] CPU #1: 1804 H, 15.04 kH/s
[2016-08-07 17:33:39] CPU #4: 10.57 kH, 20.45 kH/s
[2016-08-07 17:33:39] CPU #7: 41.26 kH, 23.86 kH/s
[2016-08-07 17:33:39] CPU #0: 40.93 kH, 23.52 kH/s
[2016-08-07 17:33:39] CPU #3: 3664 H, 15.51 kH/s
[2016-08-07 17:33:39] CPU #5: 43.05 kH, 24.83 kH/s

And after a min or to

[2016-08-07 17:34:39] esp.maxminers.net:5872 hmq1725 block 521810
[2016-08-07 17:34:39] CPU #2: 153.68 kH, 15.52 kH/s
[2016-08-07 17:34:39] CPU #6: 123.07 kH, 15.50 kH/s
[2016-08-07 17:34:39] CPU #4: 151.63 kH, 15.31 kH/s
[2016-08-07 17:34:39] CPU #5: 153.09 kH, 15.46 kH/s
[2016-08-07 17:34:39] CPU #7: 151.47 kH, 15.30 kH/s
[2016-08-07 17:34:39] CPU #0: 152.34 kH, 15.39 kH/s
[2016-08-07 17:34:39] CPU #1: 149.37 kH, 15.09 kH/s
[2016-08-07 17:34:39] CPU #3: 149.26 kH, 15.07 kH/s
[2016-08-07 17:34:48] esp.maxminers.net:5872 hmq1725 block 521811
[2016-08-07 17:34:48] CPU #5: 139.25 kH, 15.46 kH/s
[2016-08-07 17:34:48] CPU #7: 138.15 kH, 15.34 kH/s
[2016-08-07 17:34:48] CPU #0: 138.86 kH, 15.42 kH/s
[2016-08-07 17:34:48] CPU #2: 140.02 kH, 15.54 kH/s
[2016-08-07 17:34:48] CPU #1: 136.06 kH, 15.10 kH/s
[2016-08-07 17:34:48] CPU #6: 139.75 kH, 15.51 kH/s
[2016-08-07 17:34:48] CPU #4: 138.34 kH, 15.36 kH/s
[2016-08-07 17:34:48] CPU #3: 136.03 kH, 15.10 kH/s
[2016-08-07 17:34:59] esp.maxminers.net:5872 hmq1725 block 521812
[2016-08-07 17:34:59] CPU #7: 168.57 kH, 15.31 kH/s
[2016-08-07 17:34:59] CPU #4: 168.87 kH, 15.34 kH/s
[2016-08-07 17:34:59] CPU #6: 170.61 kH, 15.50 kH/s
[2016-08-07 17:34:59] CPU #5: 170.04 kH, 15.44 kH/s
[2016-08-07 17:34:59] CPU #2: 170.90 kH, 15.52 kH/s
[2016-08-07 17:34:59] CPU #0: 169.72 kH, 15.42 kH/s
[2016-08-07 17:34:59] CPU #1: 166.12 kH, 15.09 kH/s
[2016-08-07 17:34:59] CPU #3: 166.18 kH, 15.09 kH/s

And that just go on and on and on

I have another machine with a older version where it's working. What I'm doing wrong here??
legendary
Activity: 1470
Merit: 1114
Back to our regularly scheduled program. In the last episode it was discovered there are two copies of the
scryptjane code, one for the scryptjane algo itself and one for argon2 algo. The former compiles, the latter doesn't using
gcc 5.4.0 using the -flto option.

The main difference betwen these two file sets is in the presence of AVX2 code in the argon2 copy. The technique
for selecting a target at compile time is the same in both copies. There are some small code differences between the duplicated files
but it is unclear if they are functional or if they would affect compilation, specifically LTO optimization.

The plan forward is to create a build environment using gcc 5.4.0 to do more testing and comparing of the two scryptjane
versions with the goals of finding the cause of the compile error, merging the two copies of scryptjane and using the AVX2
code to optimize the scryptjane algo.

In the meantime users of gcc 5.4.0 can either compile without -flto or apply the workaround to disable argon2 and the GRS
SSE2 code as previusly described.
Jump to: