Author

Topic: [ANN]: cpuminer-opt v3.8.8.1, open source optimized multi-algo CPU miner - page 186. (Read 444067 times)

full member
Activity: 231
Merit: 150
Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

Is the problem with cpuminer-opt or the rig?

Thanks for the info abnout cpumier-multi or . It seems to claim SSE2 optimizations so how does it
compare with cpuminer-opt? Maybe they did a better job.

You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because
I think some of my tweaks actually hurt performance.

Thanks for the extensive testing.

AMD is working with cpuminer-muti or cpuminer_x64_SSE2 in windows, in VM your versions doesn't work for the AMD rig.
Yours loads now and appears to be working but at a low load rate "20% or less" and I never see any shares completed with the amount of time I've let it run.
Over 2 hours nothing.

OK so there is a problem with cpuminer-opt in a Linux VM on your AMD rig but the other miners work ok
natively in Windows on the same machine? Is this correct?

How does the performance compare between cpuminer-multi and cpuminer_x64_SSE2. It has SSE2 in
the name but does it have the performance? It should be significantly faster than than multi.

Edit and another thing you could try the other miners in the VM?This eliminates another variable. Thanks.
When did it break for cpuminer-opt?
#1 yes

#2 Haven't had much of a change to take a close look at those numbers yet.
RL is being a pain atm.

#3 When I get the time I'll be more then glad to, I did try or tried to compile cpuminer-multi  with no luck in the VM so far.
No source for cpuminer_x64_SSE2 to compile that I'm aware of.

#4 hasn't never worked for the AMD rig, but by this time next week with any luck I'll have my new CPU's up and running so it will no longer matter.
I will try to get you as much info as I can if you'd like that before the change over, but looks like I maybe the only one running this type of setup.
So that may not help anyone, but I willing to see what I can dig up.
legendary
Activity: 1470
Merit: 1114
Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

Is the problem with cpuminer-opt or the rig?

Thanks for the info abnout cpumier-multi or . It seems to claim SSE2 optimizations so how does it
compare with cpuminer-opt? Maybe they did a better job.

You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because
I think some of my tweaks actually hurt performance.

Thanks for the extensive testing.

AMD is working with cpuminer-muti or cpuminer_x64_SSE2 in windows, in VM your versions doesn't work for the AMD rig.
Yours loads now and appears to be working but at a low load rate "20% or less" and I never see any shares completed with the amount of time I've let it run.
Over 2 hours nothing.

OK so there is a problem with cpuminer-opt in a Linux VM on your AMD rig but the other miners work ok
natively in Windows on the same machine? Is this correct?

How does the performance compare between cpuminer-multi and cpuminer_x64_SSE2. It has SSE2 in
the name but does it have the performance? It should be significantly faster than than multi.

Edit and another thing you could try the other miners in the VM?This eliminates another variable. Thanks.
When did it break for cpuminer-opt?
full member
Activity: 231
Merit: 150
Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

Is the problem with cpuminer-opt or the rig?

Thanks for the info abnout cpumier-multi or . It seems to claim SSE2 optimizations so how does it
compare with cpuminer-opt? Maybe they did a better job.

You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because
I think some of my tweaks actually hurt performance.

Thanks for the extensive testing.

AMD is working with cpuminer-muti or cpuminer_x64_SSE2 in windows, in VM your versions doesn't work for the AMD rig.
Yours loads now and appears to be working but at a low load rate "20% or less" and I never see any shares completed with the amount of time I've let it run.
Over 2 hours nothing.
legendary
Activity: 1470
Merit: 1114
Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

Is the problem with cpuminer-opt or the rig?

Thanks for the info abnout cpumier-multi. It seems to claim SSE2 optimizations so how does it
compare with cpuminer-opt? Maybe they did a better job.

You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because
I think some of my tweaks actually hurt performance.

Thanks for the extensive testing.
full member
Activity: 231
Merit: 150
Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.
legendary
Activity: 1470
Merit: 1114

Not sure if there is a un-compiled version or Linux version of cpuminer_x64_SSE2, but the .exe of this
will mine x11 & a few other coins in Windows7-x64. No AES support, none of the cpuminer_x64_avx flavors
worked on any of my rigs, unsupported hardware? Using the cpuminer_x64_sse2 on the AMD rig ATM mining x11.
AMD rig has Server R2012 R2 as a OS. 42 day trail left on that. lol

Edit: On another note I have a old AMD FX duel CPU motherboard & hardware setting collecting dust if I can fine the time
from RL I'll setup a test bed with that "IF" I can find a spare PSU to use with it & a spot somewhere to place it. lol

I don't follow your first paragraph. Are you talking about another fork? What is cpuminer_x64_sse2?
Its found in the \NiceHashMiner_v1.2.2.2\bin folder after unpacked

runline: cpuminer_x64_SSE2 -a x11 -o stratum+tcp://x11.usa.nicehash.com:3336 -u 18tvS3deKZK5q4eTtPRWYeEMWmGmuErHgz.H8QG6 -p d=0.01 -t 32



Edit: Apparently is seems to be cpuminer-muti 1.1 renamed.? My guess is Nichhash software conf line makes it use sse2 when ran with it,
but with no options for any of the GPU algos to be mined on the CPU. Since I don't see sse2 being used in my run or maybe sse2 is used but just not printed as in use.

 

OK, that's not mine, you'll have to talk to nicehash about that.
full member
Activity: 231
Merit: 150

Not sure if there is a un-compiled version or Linux version of cpuminer_x64_SSE2, but the .exe of this
will mine x11 & a few other coins in Windows7-x64. No AES support, none of the cpuminer_x64_avx flavors
worked on any of my rigs, unsupported hardware? Using the cpuminer_x64_sse2 on the AMD rig ATM mining x11.
AMD rig has Server R2012 R2 as a OS. 42 day trail left on that. lol

Edit: On another note I have a old AMD FX duel CPU motherboard & hardware setting collecting dust if I can fine the time
from RL I'll setup a test bed with that "IF" I can find a spare PSU to use with it & a spot somewhere to place it. lol

I don't follow your first paragraph. Are you talking about another fork? What is cpuminer_x64_sse2?
Its found in the \NiceHashMiner_v1.2.2.2\bin folder after unpacked

runline: cpuminer_x64_SSE2 -a x11 -o stratum+tcp://x11.usa.nicehash.com:3336 -u 18tvS3deKZK5q4eTtPRWYeEMWmGmuErHgz.H8QG6 -p d=0.01 -t 32



Edit: Apparently is seems to be cpuminer-muti 1.1 renamed.? My guess is Nichhash software conf line makes it use sse2 when ran with it,
but with no options for any of the GPU algos to be mined on the CPU. Since I don't see sse2 being used in my run or maybe sse2 is used but just not printed as in use.

 
legendary
Activity: 1470
Merit: 1114

Not sure if there is a un-compiled version or Linux version of cpuminer_x64_SSE2, but the .exe of this
will mine x11 & a few other coins in Windows7-x64. No AES support, none of the cpuminer_x64_avx flavors
worked on any of my rigs, unsupported hardware? Using the cpuminer_x64_sse2 on the AMD rig ATM mining x11.
AMD rig has Server R2012 R2 as a OS. 42 day trail left on that. lol

Edit: On another note I have a old AMD FX duel CPU motherboard & hardware setting collecting dust if I can fine the time
from RL I'll setup a test bed with that "IF" I can find a spare PSU to use with it & a spot somewhere to place it. lol

I don't follow your first paragraph. Are you talking about another fork? What is cpuminer_x64_sse2?
full member
Activity: 231
Merit: 150
It didn't take long to hit a brick wall with windows.

static inline void transform(cubehashParm *sp )

Expected '(' to follow 'inline'

WTF?

 I guess that settles it.

visual studio or mingw?

VS. It worked for me for ccminer so I was hoping if I got the project file right it would work.
I was very careful with it matching the file list side by side with the makefile. There were
windows versions of all the miners I harvested from but I didn't compile any of them
myself.

I'll try mingw again, this time in a VM

Epsylon3 version of cpuminer compiles well with visual studio

I'll try compiling darkcoin minerd and cp3u. If they don't both compile then I know I have a bigger
problem.

Not sure if there is a un-compiled version or Linux version of cpuminer_x64_SSE2, but the .exe of this
will mine x11 & a few other coins in Windows7-x64. No AES support, none of the cpuminer_x64_avx flavors
worked on any of my rigs, unsupported hardware? Using the cpuminer_x64_sse2 on the AMD rig ATM mining x11.
AMD rig has Server R2012 R2 as a OS. 42 day trail left on that. lol

Edit: On another note I have a old AMD FX duel CPU motherboard & hardware setting collecting dust if I can fine the time
from RL I'll setup a test bed with that "IF" I can find a spare PSU to use with it & a spot somewhere to place it. lol
legendary
Activity: 1470
Merit: 1114
It didn't take long to hit a brick wall with windows.

static inline void transform(cubehashParm *sp )

Expected '(' to follow 'inline'

WTF?

 I guess that settles it.

visual studio or mingw?

VS. It worked for me for ccminer so I was hoping if I got the project file right it would work.
I was very careful with it matching the file list side by side with the makefile. There were
windows versions of all the miners I harvested from but I didn't compile any of them
myself.

I'll try mingw again, this time in a VM

Epsylon3 version of cpuminer compiles well with visual studio

I'll try compiling darkcoin minerd and cp3u. If they don't both compile then I know I have a bigger
problem.
legendary
Activity: 1400
Merit: 1050
It didn't take long to hit a brick wall with windows.

static inline void transform(cubehashParm *sp )

Expected '(' to follow 'inline'

WTF?

 I guess that settles it.

visual studio or mingw?

VS. It worked for me for ccminer so I was hoping if I got the project file right it would work.
I was very careful with it matching the file list side by side with the makefile. There were
windows versions of all the miners I harvested from but I didn't compile any of them
myself.

I'll try mingw again, this time in a VM

Epsylon3 version of cpuminer compiles well with visual studio
legendary
Activity: 1470
Merit: 1114
It didn't take long to hit a brick wall with windows.

static inline void transform(cubehashParm *sp )

Expected '(' to follow 'inline'

WTF?

 I guess that settles it.

visual studio or mingw?

VS. It worked for me for ccminer so I was hoping if I got the project file right it would work.
I was very careful with it matching the file list side by side with the makefile. There were
windows versions of all the miners I harvested from but I didn't compile any of them
myself.

I'll try mingw again, this time in a VM
legendary
Activity: 2716
Merit: 1094
Black Belt Developer
It didn't take long to hit a brick wall with windows.

static inline void transform(cubehashParm *sp )

Expected '(' to follow 'inline'

WTF?

 I guess that settles it.

visual studio or mingw?
legendary
Activity: 1470
Merit: 1114
It didn't take long to hit a brick wall with windows.

static inline void transform(cubehashParm *sp )

Expected '(' to follow 'inline'

WTF?

 I guess that settles it.
full member
Activity: 231
Merit: 150

Got it working some what just hasn't shown any hashrate yet.


Was the CPU working or just idling? Did you try other pools or algos?

I've streamlined the check in v3.0.7. The check for SSE2 wasn't working and
with the plan to drop seperate generic x86_64 target the SSE check isn't
needed anymore.

Startimg in 3.0.7 it will display seperatety whether the CPU and build
support AES_NI and select the appropriate target. The startup display
will also be directly linked to the target selection, previously there were
two seperate checks.

Was a very low load on the CPU's no matter how many cores used. If a big amount was used the OS seemed to lock up.
But I think that may have more to do the the VM setup than miner software, but can't say for sure either way.
Yes I tried other algos, but not different pools yet. I'll give your v3.0.7 a go a later on today & post back the results.
full member
Activity: 231
Merit: 150
May have a better solution for these problems. Order 4x of these tonight AMD Opteron 6276 to replace the 4x 6124 HE.
http://www.cpu-world.com/CPUs/Bulldozer/AMD-Opteron%206276.html
 Grin

64 cores with aes is a good start ;-)
I wonder how they perform in cryptonight.
Should be here Wednesday or Thursday then I have to break it down and replace. Maybe some test before the weekend.
Doing my research now to see if there is anything extra I may need to do like Update or change bios.
I have the [H] overclocking bios installed atm.
legendary
Activity: 1470
Merit: 1114
legendary
Activity: 1470
Merit: 1114

Got it working some what just hasn't shown any hashrate yet.


Was the CPU working or just idling? Did you try other pools or algos?

I've streamlined the check in v3.0.7. The check for SSE2 wasn't working and
with the plan to drop seperate generic x86_64 target the SSE check isn't
needed anymore.

Startimg in 3.0.7 it will display seperatety whether the CPU and build
support AES_NI and select the appropriate target. The startup display
will also be directly linked to the target selection, previously there were
two seperate checks.
legendary
Activity: 2716
Merit: 1094
Black Belt Developer
May have a better solution for these problems. Order 4x of these tonight AMD Opteron 6276 to replace the 4x 6124 HE.
http://www.cpu-world.com/CPUs/Bulldozer/AMD-Opteron%206276.html
 Grin

64 cores with aes is a good start ;-)
I wonder how they perform in cryptonight.
full member
Activity: 231
Merit: 150
May have a better solution for these problems. Order 4x of these tonight AMD Opteron 6276 to replace the 4x 6124 HE.
http://www.cpu-world.com/CPUs/Bulldozer/AMD-Opteron%206276.html
 Grin
Jump to: