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

[joblo]

Me again, something was clearly the older cpu 1156 and 1366 do not provide anything more than hs with older miner such as the fx processors have not improved anything if someone has a solution thank you in advance

Pleas provide the command line and program output.

cpuminer-btver1 -a lyra2z -o stratum+tcp://xzc.suprnova.cc:5595  -u darun.darun -p 1 --hide-diff

You were already told AMD perfoms poorly. And don't hijack another user's conversation.

[1714971997]

1714971997

[1714971997]

1714971997

[1714971997]

1714971997

[1714971997]

1714971997

[felixbrucker]

joblo, regarding compiling, is there anything up against replacing the windows compiles with "-march=<some march>" by compiles with just the extensions added (-msse2/-maes/-mavx/-mavx2) like nicehash does it for their compiles? would boil the resulting bins down to essentially 3 or 4 and the right one can easily be identified

cheers

[joblo]

joblo, regarding compiling, is there anything up against replacing the windows compiles with "-march=<some march>" by compiles with just the extensions added (-msse2/-maes/-mavx/-mavx2) like nicehash does it for their compiles? would boil the resulting bins down to essentially 3 or 4 and the right one can easily be identified

cheers

My intent is to do something similar but continue to compile on architecture boundaries instead of cherry picking features.
If I ignore the manufacturer differences there are 4 levels I want to support: SSE2, +AES, +AVX, +AVX2. Unfortunately
Westmere doesn't have its own compile arch defined but is SSE4.1+AES. I am targetting redundant builds one at a time.

For AMD I'm relying on your advice of what special AMD builds would be useful, meaning there isn't a compatible Intel
build with the same critical features. I would also need to be able to build it which could be an issue.

For now I'm targetting btver1. Any reason to keep it?

[felixbrucker]

joblo, regarding compiling, is there anything up against replacing the windows compiles with "-march=<some march>" by compiles with just the extensions added (-msse2/-maes/-mavx/-mavx2) like nicehash does it for their compiles? would boil the resulting bins down to essentially 3 or 4 and the right one can easily be identified

cheers

My intent is to do something similar but continue to compile on architecture boundaries instead of cherry picking features.
If I ignore the manufacturer differences there are 4 levels I want to support: SSE2, +AES, +AVX, +AVX2. Unfortunately
Westmere doesn't have its own compile arch defined but is SSE4.1+AES. I am targetting redundant builds one at a time.

For AMD I'm relying on your advice of what special AMD builds would be useful, meaning there isn't a compatible Intel
build with the same critical features. I would also need to be able to build it which could be an issue.

For now I'm targetting btver1. Any reason to keep it?

for cryptonight the difference between a native amd compile and a generic "-maes -mavx" compile is minimal, like sub 10 H/s

for lyra2re the difference was (afaik) larger, however all other compiles packaged resulted in even worse performance

for zcoin the difference is not observable

as a conclusion best amd compile is native, and after that a generic with -m flags

[joblo]

joblo, regarding compiling, is there anything up against replacing the windows compiles with "-march=<some march>" by compiles with just the extensions added (-msse2/-maes/-mavx/-mavx2) like nicehash does it for their compiles? would boil the resulting bins down to essentially 3 or 4 and the right one can easily be identified

cheers

My intent is to do something similar but continue to compile on architecture boundaries instead of cherry picking features.
If I ignore the manufacturer differences there are 4 levels I want to support: SSE2, +AES, +AVX, +AVX2. Unfortunately
Westmere doesn't have its own compile arch defined but is SSE4.1+AES. I am targetting redundant builds one at a time.

For AMD I'm relying on your advice of what special AMD builds would be useful, meaning there isn't a compatible Intel
build with the same critical features. I would also need to be able to build it which could be an issue.

For now I'm targetting btver1. Any reason to keep it?

for cryptonight the difference between a native amd compile and a generic "-maes -mavx" compile is minimal, like sub 10 H/s

for lyra2re the difference was (afaik) larger, however all other compiles packaged resulted in even worse performance

for zcoin the difference is not observable

as a conclusion best amd compile is native, and after that a generic with -m flags

I can't build native for any AMD arch, that's why I want to know which don't have an equivalent Intel build.

[felixbrucker]

joblo, regarding compiling, is there anything up against replacing the windows compiles with "-march=<some march>" by compiles with just the extensions added (-msse2/-maes/-mavx/-mavx2) like nicehash does it for their compiles? would boil the resulting bins down to essentially 3 or 4 and the right one can easily be identified

cheers

My intent is to do something similar but continue to compile on architecture boundaries instead of cherry picking features.
If I ignore the manufacturer differences there are 4 levels I want to support: SSE2, +AES, +AVX, +AVX2. Unfortunately
Westmere doesn't have its own compile arch defined but is SSE4.1+AES. I am targetting redundant builds one at a time.

For AMD I'm relying on your advice of what special AMD builds would be useful, meaning there isn't a compatible Intel
build with the same critical features. I would also need to be able to build it which could be an issue.

For now I'm targetting btver1. Any reason to keep it?

for cryptonight the difference between a native amd compile and a generic "-maes -mavx" compile is minimal, like sub 10 H/s

for lyra2re the difference was (afaik) larger, however all other compiles packaged resulted in even worse performance

for zcoin the difference is not observable

as a conclusion best amd compile is native, and after that a generic with -m flags

I can't build native for any AMD arch, that's why I want to know which don't have an equivalent Intel build.

yes i know there is no native compile, best compile for my fx and a6/a10 cpus was a generic compile (after native), every one of the intel compiles was slower than the generic one

[joblo]

joblo, regarding compiling, is there anything up against replacing the windows compiles with "-march=<some march>" by compiles with just the extensions added (-msse2/-maes/-mavx/-mavx2) like nicehash does it for their compiles? would boil the resulting bins down to essentially 3 or 4 and the right one can easily be identified

cheers

My intent is to do something similar but continue to compile on architecture boundaries instead of cherry picking features.
If I ignore the manufacturer differences there are 4 levels I want to support: SSE2, +AES, +AVX, +AVX2. Unfortunately
Westmere doesn't have its own compile arch defined but is SSE4.1+AES. I am targetting redundant builds one at a time.

For AMD I'm relying on your advice of what special AMD builds would be useful, meaning there isn't a compatible Intel
build with the same critical features. I would also need to be able to build it which could be an issue.

For now I'm targetting btver1. Any reason to keep it?

for cryptonight the difference between a native amd compile and a generic "-maes -mavx" compile is minimal, like sub 10 H/s

for lyra2re the difference was (afaik) larger, however all other compiles packaged resulted in even worse performance

for zcoin the difference is not observable

as a conclusion best amd compile is native, and after that a generic with -m flags

I can't build native for any AMD arch, that's why I want to know which don't have an equivalent Intel build.

yes i know there is no native compile, best compile for my fx and a6/a10 cpus was a generic compile (after native), every one of the intel compiles was slower than the generic one

What options? I will try to build it with those optoins on my Intel and see if it works well on your AMD.

[felixbrucker]

What options? I will try to build it with those optoins on my Intel and see if it works well on your AMD.

you can compile with: "-maes -mavx", this results an a binary with sse2, aes and avx support (none of my amd cpus has avx2)

ill gladly test it

on a side note: benchmark of lyra2z was broken in ocminers miner and yours as well, are you planning on fixing it or just wait till mpt (or whatever the new zcoin algo is named) gets released?

[felixbrucker]

also i just remembered the mysterious bug/error where cpuminer-opt would crash in my lxc containers with a buffer overflow, its not happening anymore, seems it was indeed some lxc or kernel related issue

edit: it might also be related to me not catching the stdout/stderr when spawning cpuminer-opt as a child process, anyways its working now

[joblo]

What options? I will try to build it with those optoins on my Intel and see if it works well on your AMD.

you can compile with: "-maes -mavx", this results an a binary with sse2, aes and avx support (none of my amd cpus has avx2)

ill gladly test it

on a side note: benchmark of lyra2z was broken in ocminers miner and yours as well, are you planning on fixing it or just wait till mpt (or whatever the new zcoin algo is named) gets released?

I'll build one and pm you the link. It should be the same as corei7-avx or core-avx-i.

[felixbrucker]

here are the results (surprised me that the packaged builds did better on lyra2*, last time i tested them they where worse):

fx-8320e:

lyra2re:

aes-avx: 396kh/s
corei7-avx: 412kh/s
core-avx-i: 411kh/s

lyra2rev2:

aes-avx: 539kh/s
corei7-avx: 568kh/s
core-avx-i: 568kh/s

cryptonight (8 threads):

aes-avx: 229h/s
corei7-avx: 228h/s
core-avx-i: 229h/s

cryptonight (7 threads):

aes-avx: 231h/s
corei7-avx: 230h/s
core-avx-i: 229h/s


a10-6800k:

lyra2re:

aes-avx: 311kh/s
corei7-avx: 314kh/s
core-avx-i: 317kh/s

lyra2rev2:

aes-avx: 344kh/s
corei7-avx: 364kh/s
core-avx-i: 364kh/s

cryptonight (4 threads):

aes-avx: 48h/s
corei7-avx: 47h/s
core-avx-i: 47h/s

cryptonight (3 threads):

aes-avx: 56h/s
corei7-avx: 58h/s
core-avx-i: 58h/s

[joblo]

here are the results (surprised me that the packaged builds did better on lyra2*, last time i tested them they where worse):

fx-8320e:

Thanks, those are the results I expected. Next release i will drop the btver1 build and recommend AMD users choose the Intel build
with the same features.

[Marvell1]

here are the results (surprised me that the packaged builds did better on lyra2*, last time i tested them they where worse):

fx-8320e:

Thanks, those are the results I expected. Next release i will drop the btver1 build and recommend AMD users choose the Intel build
with the same features.

trying to run this miner on my desk top but I crashes right away

I have and dual core Pentium g4400 cpu (Sandy bridge)

cpuminer-core-avx2 -a zcoin -o stratum+tcp://xzc.pool.mn:2428 -u user -p x

any help ?

[joblo]

trying to run this miner on my desk top but I crashes right away

I have and dual core Pentium g4400 cpu (Sandy bridge)

cpuminer-core-avx2 -a zcoin -o stratum+tcp://xzc.pool.mn:2428 -u user -p x

any help ?

Read the readme, Sandybridge doesn't support avx2.

[Nik4691]

here are the results (surprised me that the packaged builds did better on lyra2*, last time i tested them they where worse):

fx-8320e:

Thanks, those are the results I expected. Next release i will drop the btver1 build and recommend AMD users choose the Intel build
with the same features.

trying to run this miner on my desk top but I crashes right away

I have and dual core Pentium g4400 cpu (Sandy bridge)

cpuminer-core-avx2 -a zcoin -o stratum+tcp://xzc.pool.mn:2428 -u user -p x

any help ?

No AVX2 or even AVX for this processor.
Use the cpuminer-corei7.exe kenel.

[Madmach]

Hi, After several hours mining displays "Unauthorized worker".
After rebooting cpu-miner its all OK. I have AMD FX,  cpuminer-btver1 -a lyra2z -o stratum+tcp://xzc.suprnova.cc:5595 -u user.worker -p pass

[felixbrucker]

what about renaming the binaries to the corresponding arch?

4.8.x didnt include this (thats why you where not able to compile westmere, or in general other arch's by their name)

gcc 4.8.5, 4.9.4, 5.4.0 and 6.2.0 indicate the following:

core2 == core2
corei7 == nehalem
corei7-avx == sandybridge
core-avx-i == ivybridge
core-avx2 == haswell

missing in 4.8.x:

westmere (4.9.4)
broadwell (4.9.4)
skylake (6.2.0)

i suppose there is no easy way to upgrade gcc in a mingw environment on windows, im using gnustep which is the current version and only ships with gcc 4.8.x sadly

i also attempted cross compiling but ran into issues with linking the libs in the last step, has anybody done that before?

[joblo]

what about renaming the binaries to the corresponding arch?

4.8.x didnt include this (thats why you where not able to compile westmere, or in general other arch's by their name)

gcc 4.8.5, 4.9.4, 5.4.0 and 6.2.0 indicate the following:

core2 == core2
corei7 == nehalem
corei7-avx == sandybridge
core-avx-i == ivybridge
core-avx2 == haswell

missing in 4.8.x:

westmere (4.9.4)
broadwell (4.9.4)
skylake (6.2.0)

i suppose there is no easy way to upgrade gcc in a mingw environment on windows, im using gnustep which is the current version and only ships with gcc 4.8.x sadly

i also attempted cross compiling but ran into issues with linking the libs in the last step, has anybody done that before?

I just defaulted to the gcc arch but I'm flexible. A case could also be made for going strictly by the best feature like is displayed
by the miner.

Using the Intel brand doesn't help with AMD users who are unfamiliar with them. The newer architectures have the best
feature in their name so AMD users can key on that (assuming the issue of compatibility and performance of Intel builds
on AMD CPUs is resolved).

I'm also considering dropping core-avx-i as their is no specifically targetted code for this arch. If there are any performance differences
over corei7-avx it's all from the compiler. I would like to hear if there are users with Ivybridge CPUs that get lower perfomance using
corei7-avx vs core-avx-i.

[mikhan]

tried cpuminer-xzc, it shows weird (?) 220-230 kh/s hashrate but a pool shows about 60-70 h/s, 48 threads

are you referring to ocminers cpuminer-xzc or cpuminer-opt by joblo?

please share the parameters used to start the miner for further debug, might just be a wrong algo specified

this one is from https://github.com/ocminer/cpuminer-xzc/releases
command line: cpuminer -a lyra2rev2 -o stratum+tcp://xzc.suprnova.cc:5595 -u ***.*** -p *** -t 48

[mikhan]

so this is default (no -t) run of joblo's miner on dual xeon e5-2690v3
48 threads

(pls don't mind *-3.4.5 folder name, it's a clean 3.4.
-t 12