ethminer-0.9.41-genoil-1.1

[giagge]

I want to try with my GTX 950 OC  .

what i write in the .bat ? The best mining pool ?

Thanks! .

[Rimtu]

that would be awsome

lololol, only if computers have a REALLY GOOD videocards

Gtx 670.. they hash around 13.7MHs... and there are 8 of them...

That is very good performance. 670 is a very old card. Maybe their memory bandwidth is quite high to have such high speed.

Apparently it can do even better with slight OC.
https://bitcointalk.org/index.php?topic=1368785.msg15236264#msg15236264

Each PC is around 160w at the wall when mining. But my electric cost is included in rent... so I doesn't bother me....

[Genoil]

1.1.5 is out. Adds support for nicehash stratum (add -ES flag), as well as a third option to load DAGs. This is only supported on CUDA for the time being. It allows you to generate the DAG on 1 GPU and then distribute it via system RAM to the other GPUs. Usage:  -L single <n>, where <n> is the GPU ID. As I don't have a dev system with multiple OpenCL GPU's on the same platform, I can't currently port this to OpenCL.

This work has all been done by nicehash themselves, thank you!

It kind of paves the way for supporting more stratum variants, so -ES wil probably change into -SM <s> in an upcoming release.

[sorry2xs]

so how can i download it

[kenshirothefist]

so how can i download it

As always: https://github.com/Genoil/cpp-ethereum/tree/110/releases

Regarding this:

generate the DAG on 1 GPU and then distribute it via system RAM to the other GPUs. Usage:  -L single <n>, where <n> is the GPU ID

Please note that if you have multiple different GPUs, choose the fastest one with -L <n> ... this will make sure DAG will be generated as fast as possible and will be even faster distributed to other GPUs via system memory.

Keep on mining

[sorry2xs]

so how can i download it

As always: https://github.com/Genoil/cpp-ethereum/tree/110/releases

Regarding this:

generate the DAG on 1 GPU and then distribute it via system RAM to the other GPUs. Usage:  -L single <n>, where <n> is the GPU ID

Please note that if you have multiple different GPUs, choose the fastest one with -L <n> ... this will make sure DAG will be generated as fast as possible and will be even faster distributed to other GPUs via system memory.

Keep on mining

Thanks

[Amph]

mmh experiencing a crash on the first gpu with the last release, when the gpu are creating the dag file

[Genoil]

mmh experiencing a crash on the first gpu with the last release, when the gpu are creating the dag file

With -L single 0? Or with defaults?

[drhash]

what about gpu overheat command?when you implement this command?Genoil?

[Amph]

mmh experiencing a crash on the first gpu with the last release, when the gpu are creating the dag file

With -L single 0? Or with defaults?

defaults setting

[Genoil]

mmh experiencing a crash on the first gpu with the last release, when the gpu are creating the dag file

With -L single 0? Or with defaults?

defaults setting

CUDA or CL? Can you consistently crash it?

[Amph]

mmh experiencing a crash on the first gpu with the last release, when the gpu are creating the dag file

With -L single 0? Or with defaults?

defaults setting

CUDA or CL? Can you consistently crash it?

cuda, yes it happen each time unless i reduce the overclock, so maybe is that the issue? but it was not the case with the previous version

[Genoil]

mmh experiencing a crash on the first gpu with the last release, when the gpu are creating the dag file

With -L single 0? Or with defaults?

defaults setting

CUDA or CL? Can you consistently crash it?

cuda, yes it happen each time unless i reduce the overclock, so maybe is that the issue? but it was not the case with the previous version

That's weird because nothing changed in the default DAG load code. Does it help when you set --cuda-grid-size 2048?

[bensam1231]

1.1.5 is out. Adds support for nicehash stratum (add -ES flag), as well as a third option to load DAGs. This is only supported on CUDA for the time being. It allows you to generate the DAG on 1 GPU and then distribute it via system RAM to the other GPUs. Usage:  -L single <n>, where <n> is the GPU ID. As I don't have a dev system with multiple OpenCL GPU's on the same platform, I can't currently port this to OpenCL.

This work has all been done by nicehash themselves, thank you!

It kind of paves the way for supporting more stratum variants, so -ES wil probably change into -SM <s> in an upcoming release.

That sounds neat. Guessing it's a better version of the one dag per GPU sequential load that was in place before.

Is there anyway to add a verify after generation for those of us with OC'd systems so it can check for corruption in the DAG generation phase?

Also curiously why is it loaded to system memory instead of just transferring right from one GPU to another?

[Genoil]

1.1.5 is out. Adds support for nicehash stratum (add -ES flag), as well as a third option to load DAGs. This is only supported on CUDA for the time being. It allows you to generate the DAG on 1 GPU and then distribute it via system RAM to the other GPUs. Usage:  -L single <n>, where <n> is the GPU ID. As I don't have a dev system with multiple OpenCL GPU's on the same platform, I can't currently port this to OpenCL.

This work has all been done by nicehash themselves, thank you!

It kind of paves the way for supporting more stratum variants, so -ES wil probably change into -SM <s> in an upcoming release.

That sounds neat. Guessing it's a better version of the one dag per GPU sequential load that was in place before.

Is there anyway to add a verify after generation for those of us with OC'd systems so it can check for corruption in the DAG generation phase?

Also curiously why is it loaded to system memory instead of just transferring right from one GPU to another?

Device 2 device would be nice indeed. Noted

A DAG verification check wouldn't be too difficult to add. Sounds useful too.

[clintar2]

1.1.5 is out. Adds support for nicehash stratum (add -ES flag), as well as a third option to load DAGs. This is only supported on CUDA for the time being. It allows you to generate the DAG on 1 GPU and then distribute it via system RAM to the other GPUs. Usage:  -L single <n>, where <n> is the GPU ID. As I don't have a dev system with multiple OpenCL GPU's on the same platform, I can't currently port this to OpenCL.

This work has all been done by nicehash themselves, thank you!

It kind of paves the way for supporting more stratum variants, so -ES wil probably change into -SM <s> in an upcoming release.

Using the miner WITHOUT -ES gives

Code:

  ℹ  13:24:25|openclminer0  Solution found; Submitting to blah ...
  ✘  13:24:25|openclminer0  Exception thrown in Worker thread: basic_string::substr

which is caused by the

Code:

string minernonce = solution.nonce.hex().substr(m_extraNonceHexSize, 16 - m_extraNonceHexSize);

in libstratum/EthStratumClient.cpp

I just put that code behind the if (!m_ethereumStratum) block like so

Code:

        string minernonce;
        if (!m_ethereumStratum)
                cnote << "  Nonce:" << "0x" + solution.nonce.hex();
        else
                minernonce = solution.nonce.hex().substr(m_extraNonceHexSize, 16 - m_extraNonceHexSize);

to get it to work, fyi.

[Genoil]

Thanks, I didn't encounter that during my own tests.

[clintar2]

Thanks, I didn't encounter that during my own tests.

The same code appears in the EthStratumClientV2.cpp file, too, but I don't think the problem showed in there, but my memory might be backwards on that.

[jstefanop]

Thanks, I didn't encounter that during my own tests.

can you push the fix to that to your repo?

[davembg]

Thanks, I didn't encounter that during my own tests.

I'm getting a segfault when trying to use the new NiceHash stratum feature.

Code:

ethminer -G -ES -SV 2 -S daggerhashimoto.usa.nicehash.com:3353 -O 1HtkaiWKeKZ6MJgag6a3fUQo3BxcT2gAKv.amdo:x

Genoil's ethminer 0.9.41-genoil-1.1.5
=====================================================================
Forked from github.com/ethereum/cpp-ethereum
CUDA kernel ported from Tim Hughes' OpenCL kernel
With contributions from nicehash, nerdralph, RoBiK and sp_

Please consider a donation to:
ETH: 0xeb9310b185455f863f526dab3d245809f6854b4d

[OPENCL]:Found suitable OpenCL device [Hawaii] with 8443591744 bytes of GPU memory
  ℹ  14:58:29|stratum  Connecting to stratum server daggerhashimoto.usa.nicehash.com:3353
  ℹ  14:58:29|stratum  Connected!
  ℹ  14:58:29|stratum  Starting farm
  ℹ  14:58:30|stratum  Subscribed to stratum server
  ℹ  14:58:30|stratum  Extranonce set to 02dd70
  ℹ  14:58:30|stratum  Difficulty set to 0.05
  ℹ  14:58:30|stratum  Received new job #000000070aefbb6f
  ℹ  14:58:30|openclminer0  set work; seed: #a1cf1239, target:  #00000013ffec
  ℹ  14:58:30|openclminer0  Initialising miner...
  ℹ  14:58:30|openclminer1  set work; seed: #a1cf1239, target:  #00000013ffec
  ℹ  14:58:30|openclminer1  Initialising miner...
  ℹ  14:58:30|stratum  Authorized worker 1HtkaiWKeKZ6MJgag6a3fUQo3BxcT2gAKv.amdo
  ℹ  14:58:30|openclminer2  set work; seed: #a1cf1239, target:  #00000013ffec
  ℹ  14:58:30|openclminer2  Initialising miner...
  m  14:58:31|ethminer  Mining on PoWhash #d59c2ff3 : 0.00MH/s [A0+0:R0+0:F0]
[OPENCL]:Using platform: AMD Accelerated Parallel Processing
[OPENCL]:Using platform: AMD Accelerated Parallel Processing
[OPENCL]:[OPENCL]:Using device: Hawaii(OpenCL 2.0 AMD-APP (1912.5))
[OPENCL]:Using device: Hawaii(OpenCL 2.0 AMD-APP (1912.5))
Using platform: AMD Accelerated Parallel Processing
[OPENCL]:Using device: Hawaii(OpenCL 2.0 AMD-APP (1912.5))
  m  14:58:33|ethminer  Mining on PoWhash #d59c2ff3 : 0.00MH/s [A0+0:R0+0:F0]
Segmentation fault (core dumped)