miniZ v2.0a Equihash 144,5 125,4 210,9 150,5 192,7 BeamHash3 ProgPoW Ethash CFX - page 18.

cryptosize

sr. member

Activity: 1666

Merit: 310

Quote from: miniZ on December 10, 2020, 05:15:52 PM

Quote from: cryptosize on December 09, 2020, 07:30:11 PM

@miniZ

Can you try this driver yourself? https://www.nvidia.com/en-us/drivers/results/167421/

Quote from: joseph32 on December 10, 2020, 02:49:44 AM

Quote from: cryptosize on December 09, 2020, 07:30:11 PM

@miniZ

Can you try this driver yourself? https://www.nvidia.com/en-us/drivers/results/167421/

Right now, none of the popular miners will work with the newest Nvidia driver version. Latest 457.51 is working. The windows version doesnt matter, from 1909 up to the newest 2H20.

Hi cryptosize, joseph32,

We tried the latest NVIDIA 460.79 driver and also got the error "Cannot read health status! (is libnvidia-ml.so installed?): No such file or directory".

We did look for nvml.dll file and found a few versions in distinct locations, a few were from the recent driver, but most from older driver versions.

We located nvml.dll in C:\Windows\LastGood\system32, which seems to be the most recent functional file, and we copied this file into miniZ folder as a temporary fix. Perhaps you can also find this file at this location?
edit: Instead of nvml.dll in LastGood, it is best to use the last version of nvml.dll that you can find @ C:\Windows\System32\DriverStore\FileRepository\
For us it was @ C:\Windows\System32\DriverStore\FileRepository\nvlti.inf_amd64_1b431e1f567ea842
You can copy the file into miniZ folder as a temporary fix.

We'll give some attention to this, and see if we need to make changes to the miner.

Let us know if this works for you.

Cheers

Thanks, this fixed it for now.

I guess you plan to support NVAPI in the future for temp/watt measurements? Not everyone will know what to do with nvml.dll.

miniZ

member

Activity: 690

Merit: 17

Quote from: tymeksm on December 10, 2020, 02:50:36 AM

Thanks. It works ok ;-)
I didnt know about this curl change. Thanks for info.
Other miners (API) seems to work fine under Ub20.
I had similar problem when migrating from Ub16 to Ub18 - need add "-q" to all nc requests to get work on Ub18

Now I find another method to get API JSON data from your miner (miniz), and seems to work well on all of Ubuntu 16/18/20 ;-)
echo '{"id":"0", "method":"getstat"}' | timeout 5 nc -q 0 127.0.0.1 20000

Thanks for help

Hi tymeksm,

Good that it worked for you.

Thanks for the alternative way to get the API JSON data, it might well be useful to others

Cheers

miniZ

member

Activity: 690

Merit: 17

Quote from: cryptosize on December 09, 2020, 07:30:11 PM

@miniZ

Can you try this driver yourself? https://www.nvidia.com/en-us/drivers/results/167421/

Quote from: joseph32 on December 10, 2020, 02:49:44 AM

Quote from: cryptosize on December 09, 2020, 07:30:11 PM

@miniZ

Can you try this driver yourself? https://www.nvidia.com/en-us/drivers/results/167421/

Right now, none of the popular miners will work with the newest Nvidia driver version. Latest 457.51 is working. The windows version doesnt matter, from 1909 up to the newest 2H20.

Hi cryptosize, joseph32,

We tried the latest NVIDIA 460.79 driver and also got the error "Cannot read health status! (is libnvidia-ml.so installed?): No such file or directory".

We did look for nvml.dll file and found a few versions in distinct locations, a few were from the recent driver, but most from older driver versions.

We located nvml.dll in C:\Windows\LastGood\system32, which seems to be the most recent functional file, and we copied this file into miniZ folder as a temporary fix. Perhaps you can also find this file at this location?
edit: Instead of nvml.dll in LastGood, it is best to use the last version of nvml.dll that you can find @ C:\Windows\System32\DriverStore\FileRepository\
For us it was @ C:\Windows\System32\DriverStore\FileRepository\nvlti.inf_amd64_1b431e1f567ea842
You can copy the file into miniZ folder as a temporary fix.

We'll give some attention to this, and see if we need to make changes to the miner.

Let us know if this works for you.

Cheers

tymeksm

member

Activity: 707

Merit: 12

Quote from: miniZ on December 09, 2020, 02:01:17 PM

Quote from: tymeksm on December 09, 2020, 09:14:12 AM

Hi again

I found the problem with API..but do not know whats the reason.

On Ubuntu 16 and 18 works ok, but under Ubuntu 20 there is a problem...but only sometimes - not stable API

I use this method to retrive JSON data:
curl -s http://127.0.0.1:20000/getstat

but under ubuntu20 this act like this:
- very often give just no output (empty string)
in this same time respond with HTML page if request with:
curl -s http://127.0.0.1:20000/
gives full html page
- after some period of time api freezes and hangs every connection (both of avobe). Must enter Ctrl+C to back to bash
- only sometimes after few miner restarts work ok, and respond with JSON data.

Could you try the following :
curl http://localhost:20000/getstat -s --http0.9

Curl has been modified in Ubuntu 20.

Thanks. It works ok ;-)
I didnt know about this curl change. Thanks for info.
Other miners (API) seems to work fine under Ub20.
I had similar problem when migrating from Ub16 to Ub18 - need add "-q" to all nc requests to get work on Ub18

Now I find another method to get API JSON data from your miner (miniz), and seems to work well on all of Ubuntu 16/18/20 ;-)
echo '{"id":"0", "method":"getstat"}' | timeout 5 nc -q 0 127.0.0.1 20000

Thanks for help

joseph32

member

Activity: 418

Merit: 21

Quote from: cryptosize on December 09, 2020, 07:30:11 PM

@miniZ

Can you try this driver yourself? https://www.nvidia.com/en-us/drivers/results/167421/

Right now, none of the popular miners will work with the newest Nvidia driver version. Latest 457.51 is working. The windows version doesnt matter, from 1909 up to the newest 2H20.

cryptosize

sr. member

Activity: 1666

Merit: 310

@miniZ

Can you try this driver yourself? https://www.nvidia.com/en-us/drivers/results/167421/

cryptosize

sr. member

Activity: 1666

Merit: 310

Quote from: miniZ on December 09, 2020, 05:38:37 PM

Quote from: cryptosize on December 09, 2020, 02:23:26 PM

With the latest nVidia driver (460.79) I get this on Windows 10 and latest miniZ:

"Cannot read health status! (is libnvidia-ml.so installed?): No such file or directory"

"Sol/W=inf"

457.51 worked fine.

What gives?

Hi cryptosize,

You get that error when miniZ cannot find nvml.dll.

The file should be on the standard Windows locations, in \PROGRAM FILES\\NVIDIA Corporation\\NVSMI\\nvml.dll.
Or in the miniZ folder.

Where do you have this file?

Cheers

I'm not sure, maybe nVidia changed the dll location in 460 branch?

miniZ

member

Activity: 690

Merit: 17

Quote from: cryptosize on December 09, 2020, 02:23:26 PM

With the latest nVidia driver (460.79) I get this on Windows 10 and latest miniZ:

"Cannot read health status! (is libnvidia-ml.so installed?): No such file or directory"

"Sol/W=inf"

457.51 worked fine.

What gives?

Hi cryptosize,

You get that error when miniZ cannot find nvml.dll.

The file should be on the standard Windows locations, in \PROGRAM FILES\\NVIDIA Corporation\\NVSMI\\nvml.dll.
Or in the miniZ folder.

Where do you have this file?

Cheers

cryptosize

sr. member

Activity: 1666

Merit: 310

With the latest nVidia driver (460.79) I get this on Windows 10 and latest miniZ:

"Cannot read health status! (is libnvidia-ml.so installed?): No such file or directory"

"Sol/W=inf"

457.51 worked fine.

What gives?

miniZ

member

Activity: 690

Merit: 17

Quote from: tymeksm on December 09, 2020, 09:14:12 AM

Hi again

I found the problem with API..but do not know whats the reason.

On Ubuntu 16 and 18 works ok, but under Ubuntu 20 there is a problem...but only sometimes - not stable API

I use this method to retrive JSON data:
curl -s http://127.0.0.1:20000/getstat

but under ubuntu20 this act like this:
- very often give just no output (empty string)
in this same time respond with HTML page if request with:
curl -s http://127.0.0.1:20000/
gives full html page
- after some period of time api freezes and hangs every connection (both of avobe). Must enter Ctrl+C to back to bash
- only sometimes after few miner restarts work ok, and respond with JSON data.

can you help?

Hi tymeksm,

Could you try the following :
curl http://localhost:20000/getstat -s --http0.9

Curl has been modified in Ubuntu 20.

Let us know if it works for you.

Thank you for your question.
This may help other miners too.

Cheers

tymeksm

member

Activity: 707

Merit: 12

Hi again

I found the problem with API..but do not know whats the reason.

On Ubuntu 16 and 18 works ok, but under Ubuntu 20 there is a problem...but only sometimes - not stable API

I use this method to retrive JSON data:
curl -s http://127.0.0.1:20000/getstat

but under ubuntu20 this act like this:
- very often give just no output (empty string)
in this same time respond with HTML page if request with:
curl -s http://127.0.0.1:20000/
gives full html page
- after some period of time api freezes and hangs every connection (both of avobe). Must enter Ctrl+C to back to bash
- only sometimes after few miner restarts work ok, and respond with JSON data.

can you help?

tymeksm

member

Activity: 707

Merit: 12

Quote from: miniZ on December 08, 2020, 12:33:36 PM

In the next version there will be a more clear message.
Something like this:
Number of miniZ CUDA>=[8.0] using driver CUDA[10.20] devices found: 1

Thank you. Now it's clear ;-)

UselessGuru

jr. member

Activity: 200

Merit: 3

Quote from: miniZ on December 07, 2020, 08:23:16 PM

Quote from: UselessGuru on December 05, 2020, 10:41:56 AM

Thank you for your feedback..
>> miniZ v1.6w4 is only to be used with RTX 30XX.
Cannot load deviceZ code.
It is just odd that it sometimes works... Huh

Quote from: platinum4 on December 05, 2020, 10:34:03 PM

Can you put this on beta page? I'd like to test it on a rig of 1070s that are rented for 2 days.

@UselessGuru Hi again -- I am using 16w4 on CUDA10 for Pascal/Turing just fine. Don't use the CUDA11.1 version but use the CUDA10 from the beta page.

Hi UselessGuru, platinum4,
We've just released a new version v1.6x

Could you try it out and see if it works for you?
Cheers

The issues I had have been fixed.
1.6x is working fine here - thank you!

miniZ

member

Activity: 690

Merit: 17

Quote from: sxemini on December 08, 2020, 04:23:37 AM

Thanks --avg1 works. I need this for Awesomeminer. Maybe you can change your usage list on your page? So many commands are missing there.
--mode
--avg1

and more you explained some times in this thread, but didn´t find it in your usage commandline list.

Hi sxemini,
Good that it worked for you.

You're right, we'll review a few things on the website (and --help) in the next few days.
For information on --mode see FAQ

Thanks!
Cheers

miniZ

member

Activity: 690

Merit: 17

Quote from: tymeksm on December 08, 2020, 02:56:08 AM

Hello

Question about new cuda8/10/11 support:

Example output:

Code:

$ ./miniZ -ci
Number of miniZ CUDA[8.0] driver[11.10] devices found: 6
  #0 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:01:00.0
  #1 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:03:00.0
  #2 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:06:00.0
  #3 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:07:00.0
  #4 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:08:00.0
  #5 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:09:00.0

tells me, that cuda8 libraries are used, yes?

I have Ubuntu20 with no any cuda libraries installed. Just NVIDIA driver 455.45.01 installed.
As I understand the miniz binary from 1.6x has integrated 3 versions of cuda libraries. How to make the binary use cuda10 or 11 libraries?
Or is it only used for newer NV GPUs that required newer cuda libraries?

Hi tymeksm,
Yes, miniZ integrates cuda 8/10/11. You only need to have the driver installed.
Cuda 8 is the minimum Cuda version that miniZ supports.

In the next version there will be a more clear message.
Something like this:
Number of miniZ CUDA>=[8.0] using driver CUDA[10.20] devices found: 1

Thank you for the feedback!
Cheers

sxemini

member

Activity: 1558

Merit: 69

Quote from: miniZ on December 07, 2020, 08:26:35 PM

Quote from: sxemini on December 07, 2020, 09:03:55 AM

@miniZ

The Solution per second in the API is the average speed right?
Can you add Solution Average and Solution per second for the realtime hashrate.

Makes more sense to see the realtime hashrate over API.

Because a RIG goes down or slow down after 50 Day´s, i will recognize the slow down after days, because the average speed will go down very slow.

Hi sxemini,

The api speed_sps is a 3h average.

If you want you can read speed_is (iteration rate, 3 minutes average) and multiply by 2.

You could also add --avg1=period_in_seconds to the miniZ command line. This way you can adjust as you wish.

Let us know if this helps.
Cheers

Thanks --avg1 works. I need this for Awesomeminer. Maybe you can change your usage list on your page? So many commands are missing there.
--mode
--avg1

and more you explained some times in this thread, but didn´t find it in your usage commandline list.

tymeksm

member

Activity: 707

Merit: 12

Hello

Question about new cuda8/10/11 support:

Example output:

Code:

$ ./miniZ -ci
Number of miniZ CUDA[8.0] driver[11.10] devices found: 6
  #0 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:01:00.0
  #1 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:03:00.0
  #2 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:06:00.0
  #3 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:07:00.0
  #4 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:08:00.0
  #5 GeForce GTX 1070    | SM version: 6.1 | SM count: 15 | cores: 1920 | Memory:  3.767GB | BusID: 0000:09:00.0

tells me, that cuda8 libraries are used, yes?

I have Ubuntu20 with no any cuda libraries installed. Just NVIDIA driver 455.45.01 installed.
As I understand the miniz binary from 1.6x has integrated 3 versions of cuda libraries. How to make the binary use cuda10 or 11 libraries?
Or is it only used for newer NV GPUs that required newer cuda libraries?

miniZ

member

Activity: 690

Merit: 17

Quote from: sxemini on December 07, 2020, 09:03:55 AM

@miniZ

The Solution per second in the API is the average speed right?
Can you add Solution Average and Solution per second for the realtime hashrate.

Makes more sense to see the realtime hashrate over API.

Because a RIG goes down or slow down after 50 Day´s, i will recognize the slow down after days, because the average speed will go down very slow.

Hi sxemini,

The api speed_sps is a 3h average.

If you want you can read speed_is (iteration rate, 3 minutes average) and multiply by 2.

You could also add --avg1=period_in_seconds to the miniZ command line. This way you can adjust as you wish.

Let us know if this helps.
Cheers

miniZ

member

Activity: 690

Merit: 17

Quote from: UselessGuru on December 05, 2020, 10:41:56 AM

Thank you for your feedback..
>> miniZ v1.6w4 is only to be used with RTX 30XX.
Cannot load deviceZ code.
It is just odd that it sometimes works... Huh

Quote from: platinum4 on December 05, 2020, 10:34:03 PM

Can you put this on beta page? I'd like to test it on a rig of 1070s that are rented for 2 days.

@UselessGuru Hi again -- I am using 16w4 on CUDA10 for Pascal/Turing just fine. Don't use the CUDA11.1 version but use the CUDA10 from the beta page.

Hi UselessGuru, platinum4,
We've just released a new version v1.6x

Could you try it out and see if it works for you?
Cheers

miniZ

member

Activity: 690

Merit: 17

Hi everyone,

Thank you for all your feedback and for using miniZ miner!

A new miniZ v1.6x version is out with major improvements for 210,9 algorithm - AION.

Please find miniZ v1.6x version @ Download page.

Changelog:

* 210,9 major improvements: up to 8% depending on GPU.
* 144,5 improved hashrate for the RTX 30XX GPUs.
* Added support for SM_30 (Kepler) devices.
* Fixed issues with --pers on MiningRigRentals
* Combined CUDA 8/10/11 versions into one (same) executable.

You can check some of our 210,9 testing conditions here. See also FAQ - How to mine AION?

Remember to try --oc1--oc2--ocX options for optimum performance.

Download miniZ version v1.6x here.

For additional information check our Usage or FAQ pages.

Follow us
On Twitter

Happy miniZ mining!

Topic: miniZ v2.0a Equihash 144,5 125,4 210,9 150,5 192,7 BeamHash3 ProgPoW Ethash CFX - page 18. (Read 60004 times)