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

[JayDDee]

Info about GCC support for znver4...

https://www.phoronix.com/news/AMD-Zen-4-Znver4-GCC-Enable

So what this amounts to at this point is getting -march=native working for Zen 4, honoring -march=znver4, and then over the Znver3 target just flipping on AVX512F, AVX512DQ, AVX512IFMA, AVX512CD, AVX512BW, AVX512VL, AVX512BF16, AVX512VBMI, AVX512VBMI2, GFNI, AVX512VNNI, AVX512BITALG, and AVX512VPOPCNTDQ.

The current advice for compiling is to use znver3 and add the AVX512 extensions manually. For the binary Windows builds avx512-sha-vaes will provide near optimum performance.

[alucard20724]

Info about GCC support for znver4...

https://www.phoronix.com/news/AMD-Zen-4-Znver4-GCC-Enable

So what this amounts to at this point is getting -march=native working for Zen 4, honoring -march=znver4, and then over the Znver3 target just flipping on AVX512F, AVX512DQ, AVX512IFMA, AVX512CD, AVX512BW, AVX512VL, AVX512BF16, AVX512VBMI, AVX512VBMI2, GFNI, AVX512VNNI, AVX512BITALG, and AVX512VPOPCNTDQ.

The current advice for compiling is to use znver3 and add the AVX512 extensions manually. For the binary Windows builds avx512-sha-vaes will provide near optimum performance.

I like that the new a7XXX series cpu from AMD have igpu.
the one thing that i don't like about the new a7XXX series cpu is that you can't disable the igpu.     not even in current bios.

[JayDDee]

I like that the new a7XXX series cpu from AMD have igpu.
the one thing that i don't like about the new a7XXX series cpu is that you can't disable the igpu.     not even in current bios.

Changing the primary display adapter setting in the BIOS to external GPU might do it.

[alucard20724]

I like that the new a7XXX series cpu from AMD have igpu.
the one thing that i don't like about the new a7XXX series cpu is that you can't disable the igpu.     not even in current bios.

Changing the primary display adapter setting in the BIOS to external GPU might do it.

Asrock 670E  Steel legend that i have my 7700X doesn't have any options for display.
And neither does the Asrock 670E Taichi... maybe it's just an asrock thing.   I normally go with Asus boards.

[deedeeranged]

I like that the new a7XXX series cpu from AMD have igpu.
the one thing that i don't like about the new a7XXX series cpu is that you can't disable the igpu.     not even in current bios.

Changing the primary display adapter setting in the BIOS to external GPU might do it.

Asrock 670E  Steel legend that i have my 7700X doesn't have any options for display.
And neither does the Asrock 670E Taichi... maybe it's just an asrock thing.   I normally go with Asus boards.

It's not an ASRock thing. All suffer from the introduction of AM5 socket and chips. Keep an eye out for BIOS updates the next couple of months as they will come as it is a brand new platform.

[alucard20724]

I like that the new a7XXX series cpu from AMD have igpu.
the one thing that i don't like about the new a7XXX series cpu is that you can't disable the igpu.     not even in current bios.

Changing the primary display adapter setting in the BIOS to external GPU might do it.

Asrock 670E  Steel legend that i have my 7700X doesn't have any options for display.
And neither does the Asrock 670E Taichi... maybe it's just an asrock thing.   I normally go with Asus boards.

It's not an ASRock thing. All suffer from the introduction of AM5 socket and chips. Keep an eye out for BIOS updates the next couple of months as they will come as it is a brand new platform.

Asrock just released bios update (1.11) that now gives you the ability to boot from internal(igpu) or external display... but still no option to disable igpu

[JayDDee]

Asrock 670E  Steel legend that i have my 7700X doesn't have any options for display.
And neither does the Asrock 670E Taichi... maybe it's just an asrock thing.   I normally go with Asus boards.

It's not an ASRock thing. All suffer from the introduction of AM5 socket and chips. Keep an eye out for BIOS updates the next couple of months as they will come as it is a brand new platform.

Asrock just released bios update (1.11) that now gives you the ability to boot from internal(igpu) or external display... but still no option to disable igpu

Maybe power management got smarter and can detect if/when the iGPU is not being used. It shouldn't be difficult for zen4 CPUs because the
iGPU is only to "light up a display" according to AMD CEO. Just speculating.

[alucard20724]

Asrock 670E  Steel legend that i have my 7700X doesn't have any options for display.
And neither does the Asrock 670E Taichi... maybe it's just an asrock thing.   I normally go with Asus boards.

It's not an ASRock thing. All suffer from the introduction of AM5 socket and chips. Keep an eye out for BIOS updates the next couple of months as they will come as it is a brand new platform.

Asrock just released bios update (1.11) that now gives you the ability to boot from internal(igpu) or external display... but still no option to disable igpu

Maybe power management got smarter and can detect if/when the iGPU is not being used. It shouldn't be difficult for zen4 CPUs because the
iGPU is only to "light up a display" according to AMD CEO. Just speculating.

no matter what i do, i can't get rid of it(igpu).  it always shows up in on afterburner and in devices.. disabled, and uninstalled... still pops up in ab and device manager.   it's like that itch that won't go away after a night of drinking and you wake up in a ditch with your pants around your ankles and you wonder how you got there.. 

[deedeeranged]

Asrock 670E  Steel legend that i have my 7700X doesn't have any options for display.
And neither does the Asrock 670E Taichi... maybe it's just an asrock thing.   I normally go with Asus boards.

It's not an ASRock thing. All suffer from the introduction of AM5 socket and chips. Keep an eye out for BIOS updates the next couple of months as they will come as it is a brand new platform.

Asrock just released bios update (1.11) that now gives you the ability to boot from internal(igpu) or external display... but still no option to disable igpu

Maybe power management got smarter and can detect if/when the iGPU is not being used. It shouldn't be difficult for zen4 CPUs because the
iGPU is only to "light up a display" according to AMD CEO. Just speculating.

no matter what i do, i can't get rid of it(igpu).  it always shows up in on afterburner and in devices.. disabled, and uninstalled... still pops up in ab and device manager.   it's like that itch that won't go away after a night of drinking and you wake up in a ditch with your pants around your ankles and you wonder how you got there.. 

Intel CPU with iGPU usually show the same behaviour. It uses the iGPU with an attached display and when it senses a dGPU with a display it will switch to that but you will probably still see the iGPU in device manager. This is from memory as I have not used and Intel for a long time.

[alucard20724]

Asrock 670E  Steel legend that i have my 7700X doesn't have any options for display.
And neither does the Asrock 670E Taichi... maybe it's just an asrock thing.   I normally go with Asus boards.

It's not an ASRock thing. All suffer from the introduction of AM5 socket and chips. Keep an eye out for BIOS updates the next couple of months as they will come as it is a brand new platform.

Asrock just released bios update (1.11) that now gives you the ability to boot from internal(igpu) or external display... but still no option to disable igpu

Maybe power management got smarter and can detect if/when the iGPU is not being used. It shouldn't be difficult for zen4 CPUs because the
iGPU is only to "light up a display" according to AMD CEO. Just speculating.

no matter what i do, i can't get rid of it(igpu).  it always shows up in on afterburner and in devices.. disabled, and uninstalled... still pops up in ab and device manager.   it's like that itch that won't go away after a night of drinking and you wake up in a ditch with your pants around your ankles and you wonder how you got there.. 

Intel CPU with iGPU usually show the same behaviour. It uses the iGPU with an attached display and when it senses a dGPU with a display it will switch to that but you will probably still see the iGPU in device manager. This is from memory as I have not used and Intel for a long time.

I haven't had the same problem with intel... intel i can disable the igpu in the bios.  even with the 5x00 series amd igpu i can disable the igpu, just not with the 7xxx igpu.     one thing i've noticed with the amd 7xxx series igpu, is that if you have a discrete AMD gpu (6xxx/5xxx gpu) is that the igpu attaches itself to the discrete gpu and runs in a hybrid mode.   doesn't happen with nvidia disrete gpu, though the amd igpu will still show up on af and control panel.   I'm sure eventually AMD and the board manufacturers will allow me to disable it.

[JayDDee]

The future of cpuminer-opt.

A number of things are converging that require some decisions to be made.

After 7 years I think I've squeezed about as much juice from cpuminer as I can. I have a couple more small optimisations so there will be,
at least, one more release.

Another factor is I'm planning to upgrade my main system, where I build the Windows binaries package. That means I will need to rebuild
the mingw environment for Ubuntu-22.04 from scratch or build a Ubuntu-20.04 VM and port the existing environment to it.
Without a plan for more development it's hard to get motivated to do the extra work. There is also now a more friendly way for Windows users
to compile their own, and with better tuning for their CPU.

cpuminer-opt's usefulness is diminishing fast, and falling further behind GPUs. CPUs can't provide large generational performance increases
that GPUs do. The crypto bear market isn't helping.

The mining environement has changed, a multi-algo miner is no longer feasible. Most new coins use their own custom algo now.
Most new miners are also written in C++ and use cmake so they are very difficult to integrate with the cpuminer architecture which is
written all in C and uses autoconf. There is also the bloating issue with trying to support many dissimilar algorithms.

I'm not sure what I'll do or when but it seems the end is in sight.

[chrysophylax]

The future of cpuminer-opt.

A number of things are converging that require some decisions to be made.

After 7 years I think I've squeezed about as much juice from cpuminer as I can. I have a couple more small optimisations so there will be,
at least, one more release.

Another factor is I'm planning to upgrade my main system, where I build the Windows binaries package. That means I will need to rebuild
the mingw environment for Ubuntu-22.04 from scratch or build a Ubuntu-20.04 VM and port the existing environment to it.
Without a plan for more development it's hard to get motivated to do the extra work. There is also now a more friendly way for Windows users
to compile their own, and with better tuning for their CPU.

cpuminer-opt's usefulness is diminishing fast, and falling further behind GPUs. CPUs can't provide large generational performance increases
that GPUs do. The crypto bear market isn't helping.

The mining environement has changed, a multi-algo miner is no longer feasible. Most new coins use their own custom algo now.
Most new miners are also written in C++ and use cmake so they are very difficult to integrate with the cpuminer architecture which is
written all in C and uses autoconf. There is also the bloating issue with trying to support many dissimilar algorithms.

I'm not sure what I'll do or when but it seems the end is in sight.

Bring Your Experience ...

And Join us at CWI for our Developments in this area

#crysx #cwi

[JayDDee]

I have seen increasing instances of cpuminer-opt source code being used as a decoy by other github users to give an appearance of
an open source cpu miner. The binary releases don't match the posted source code and could be infested with malware.

It seems ironic that in the worst mining market in years this would be a growing problem.

This new problem goes beyond the frequent practice of closing the source code to collect a fee, it's downright fraud. It misleads the user into
thinking they are downloading a binary built from open source.

As always stay alert. cpuminer-opt never needs to run with elevated priviledges, doing so with any untrusted miner exposes the system to attack.
Only use trusted software.

[sandy25007]

Hi All,

I am trying to understand working of bitcoin mining and going through the source code of cpuminer-opt. The source code is really helpful, Thanks!

I have a basic question, my understanding is that in a coinbase transaction miner adds Coinbase_reward + fees (total fees from all the transactions included in the block being mined).

I can see the cpu-miner.c code adding Coinbase_reward by taking coinbasevalue, but I could not find code for fees getting added. Can someone please point me to the code that does that?

Thanks & Regards,
Sandy

PS
Referring : https://github.com/JayDDee/cpuminer-opt/blob/master/cpu-miner.c

[JayDDee]

Hi All,

I am trying to understand working of bitcoin mining and going through the source code of cpuminer-opt. The source code is really helpful, Thanks!

I have a basic question, my understanding is that in a coinbase transaction miner adds Coinbase_reward + fees (total fees from all the transactions included in the block being mined).

I can see the cpu-miner.c code adding Coinbase_reward by taking coinbasevalue, but I could not find code for fees getting added. Can someone please point me to the code that does that?

Thanks & Regards,
Sandy

PS
Referring : https://github.com/JayDDee/cpuminer-opt/blob/master/cpu-miner.c

I think you're looking for the merkle tree but that's not my area of expertise. You might get better advice elsewhere.

[sandy25007]

Hi All,

I am trying to understand working of bitcoin mining and going through the source code of cpuminer-opt. The source code is really helpful, Thanks!

I have a basic question, my understanding is that in a coinbase transaction miner adds Coinbase_reward + fees (total fees from all the transactions included in the block being mined).

I can see the cpu-miner.c code adding Coinbase_reward by taking coinbasevalue, but I could not find code for fees getting added. Can someone please point me to the code that does that?

Thanks & Regards,
Sandy

PS
Referring : https://github.com/JayDDee/cpuminer-opt/blob/master/cpu-miner.c

I think you're looking for the merkle tree but that's not my area of expertise. You might get better advice elsewhere.

No, actually I am looking at coinbase transaction generation, code under section   /* build coinbase transaction */

Code:

      tmp = json_object_get( val, "coinbasevalue" );
      if ( !tmp || !json_is_number( tmp ) )
      {
         applog( LOG_ERR, "JSON invalid coinbasevalue" );
         goto out;
      }
      cbvalue = (int64_t) ( json_is_integer( tmp ) ? json_integer_value( tmp )
                                                   : json_number_value( tmp ) );

I feel it is missing the fees addition, below is the code I was expecting to compute and add fees.


      tmp = json_object_get( val, "coinbasevalue" );
      if ( !tmp || !json_is_number( tmp ) )
      {
         applog( LOG_ERR, "JSON invalid coinbasevalue" );
         goto out;
      }
      cbvalue = (int64_t) ( json_is_integer( tmp ) ? json_integer_value( tmp )
                                                   : json_number_value( tmp ) );

       /* add fee */
        int64_t total_fees=0;
        int64_t fee=0;

        for (i = 0; i < tx_count; i++) {
        const json_t *tx = json_array_get(txa, i);            
        tmp = json_object_get(tx, "fee");      
        if (!tmp || !json_is_number(tmp)) {
           applog(LOG_ERR, "JSON invalid Fee");
          goto out;
        }
      fee = json_is_integer(tmp) ? json_integer_value(tmp) : json_number_value(tmp);      
      total_fees = total_fees + fee;
       }

   cbvalue = cbvalue + total_fees;



Thanks & Regards,
Sandy

[hbouma]

The future of cpuminer-opt.

A number of things are converging that require some decisions to be made.

After 7 years I think I've squeezed about as much juice from cpuminer as I can. I have a couple more small optimisations so there will be,
at least, one more release.

cpuminer-opt's usefulness is diminishing fast, and falling further behind GPUs. CPUs can't provide large generational performance increases
that GPUs do. The crypto bear market isn't helping.

I'm not sure what I'll do or when but it seems the end is in sight.

Just wanted to say thanks for writing this miner! Good luck with your future projects if this one remains on the backburner indefinitely.

Hal

[punkyshungry]

I read this miner supports the Xeon Phi Coprossers? I was thinking about getting one, is this true?

[JayDDee]

I read this miner supports the Xeon Phi Coprossers? I was thinking about getting one, is this true?

Never supported.

[WayneBunbury]

dose it support all intel cpus  old and new ?