[ANN] dstm's ZCash / Equihash Nvidia Miner v0.6.2 (Linux / Windows)

[dstm]

#################################################
#  dstm's ZCash Cuda miner 
#################################################

I've written it initially for my own purpose. It performs pretty well - so here it is.
Compatible with devices having Compute Capability 5.0 and greater i.e. it should run fine on Maxwell/Pascal GPUs.
Supports every server using the stratum protocol. If you have issues with certain pools pls. report them here.
Contains 2% devfee.
Suggestions are welcome.


Performance:
1080Ti 725-740 Sol/s
1070   440-450 Sol/s
stock settings


UI:
Sample output mining on a rig with 8 GPUs.
GPU0-7: 1070Ti

Code:

>  GPU0  65C  75% |  507.9 Sol/s   504.9 Avg   269.8 I/s | 4.52 S/W  112 W |  3.74  100  39 ++++++++
>  GPU1  64C  70% |  508.4 Sol/s   509.1 Avg   273.0 I/s | 4.57 S/W  111 W |  4.11  100  38 +++++++++
>  GPU2  62C  70% |  512.9 Sol/s   514.0 Avg   274.0 I/s | 4.60 S/W  110 W |  2.62  100  37 +++++
>  GPU3  61C  70% |  502.4 Sol/s   500.8 Avg   266.9 I/s | 4.48 S/W  113 W |  2.24  100  38 ++++++++*
>  GPU4  64C  70% |  508.6 Sol/s   508.2 Avg   272.9 I/s | 4.53 S/W  111 W |  1.49  100  38 ++++++++++++
>  GPU5  57C  70% |  506.7 Sol/s   504.7 Avg   270.0 I/s | 4.53 S/W  110 W |  1.94  100  38 ++++++++++
>  GPU6  59C  70% |  514.5 Sol/s   506.3 Avg   270.4 I/s | 4.55 S/W  112 W |  2.36  100  38 ++++++
>  GPU7  64C  75% |  511.1 Sol/s   515.1 Avg   275.2 I/s | 4.62 S/W  109 W |  1.12  100  37 ++++++++
>  ============== | 4072.6 Sol/s  4063.3 Avg  2172.1 I/s | 4.55 S/W  891 W | 19.63  100  37 ++++++++++

Sol/s: solutions per second
Avg: average solutions per second
I/s: iterations per second done by the GPU
S/W: efficiency - average Sol/s per Watt
W: power consuption in Watt
last 3 colums: <shares per minute> <accepted shares ratio> <network latency in ms>


> : indicates that a new job was received
+ : indicates one submitted share
* : indicates one submitted dev fee share
= : sum/average if mining on multiple GPUs



Features:
To get help just start zm without any arguments

Code:

$ ./zm
ZM 0.6.1, dstm's ZCASH/Equihash Cuda Miner

Usage:
 zm --server hostname --port port_nr --user user_name
    [--pass password] [options]...

 zm --cfg-file[=path]

 Stratum:
    --server         Stratum server hostname
                     prefix hostname with 'ssl://' for encrypted
                     connections - e.g. ssl://mypool.com
    --port           Stratum server port number
    --user           Username
    --pass           Worker password

 Options:
    --help           Print this help
    --list-devices   List available cuda devices

    --dev            Space separated list of cuda devices to use.
                     If this option is not given all available devices
                     are used.

    --time           Enable output of timestamps
    --color          colorize the output

    --logfile        [=path] Append logs to the file named by 'path'
                     If 'path' is not given append to 'zm.log' in
                     current working directory.
    --noreconnect    Disable automatic reconnection on network errors.

    --temp-target    =dev_id:temp-target[,dev_id:temp-target] ...
                     In C - If set, enables temperature controller.
                     The workload of each GPU will be continuously
                     adjusted such that the temperature stays around
                     this value. It is recommended to set your fan speed
                     to a constant value when using this setting.
                     Example: --temp-target=0:65,2:70

    --intensity      =dev_id:intensity[,dev_id:intensity] ...
                     Reduce the load which is put on the GPU - valid
                     intensity range ]1.0-0.0[.
                     Example: --intensity=0:0.893,2:0.8

    --telemetry      [=ip:port]. Starts telemetry server. Telemetry data
                     can be accessed using a web browser(http) or by json-rpc.
                     If no arguments are given the server listens on
                     127.0.0.1:2222 - Example: --telemetry=0.0.0.0:2222
                     Valid port range [1025-65535]

    --cfg-file       [=path] Use configuration file. All additional command
                     line options are ignored - configuration is done only
                     through configuration file. If 'path' is not given
                     use 'zm.cfg' in current working directory.

    --pool           =hostname,port_nr,user_name[,pass]
                     Setup additional failover pools.

 Example:
    zm --server servername.com --port 1234 --user username

BIG THANKS TO SWISSTRADER for providing access to hardware.



Dependencies                                                                                                                  
Linux:
 openssl 1.0 (for versions <= 0.5.7)

Windows:
 Visual C++ Redistributable for visual studio 2015 (VCRUNTIME140.dll)



Version 0.6.2

- fix ssl handshake failures
- fix device selection bug introduced in 0.6.1
- make linux performance improvements introduced in 0.6.1 optional
  via 'mq-solver' parameter (due to issues on some systems)
- improve device initialization on large systems

Linux x64:
executable sha1 87cd45ad538382d0bcad8ecc1077e759e3e9a95
https://mega.nz/#!vqgwhIoD!vH85qJ3aGJkgXrIqPefnhU219oi_EBGJfCXomiFoJ3k

Win x64:
executable sha1 e1da12a3af603f19c69df0c674a2668947776797
https://mega.nz/#!CioUkKAK!iDcjULjgO19gKzI0IVgLVQ1B-ZwrCHZ90ldnZSdbqL0


Version 0.6.1
- fix failover-pool not switching back on ssl errors
- fix failover-pool not switching back on some servers
- add support for stratum's client.reconnect rpc
- support configuration of 'temp-target', 'intensity', 'pool' via cmd-line parameters
- report current difficulty and target on term-ui
- sort output by gpu_id on term-ui
- colorize temperature above 70°C on term-ui
- report intended fan speed on term-ui, web-ui, json-rpc
- report power consumption on term-ui
- improve performance on linux systems by ~2%

Linux x64:
executable sha1 7a6f0eb858d8da18116b115a9b46f21187741f8a
https://drive.google.com/file/d/1JKeBTJshILqYpHiu7qRhjvcRhaa4NC5W
https://mega.nz/#!rXpjAC7D!1BLUk0PVwI9BMKLzk0HHUQLx4UBMEZL_1IXh1sJa6uU

Win x64:
executable sha1 c2c5314ca29830044ff26bcfb2dd575f625b825d
https://drive.google.com/file/d/10CQy7Tx63YiP9MbsWmqWMSfCxoZuP3LL
https://mega.nz/#!zLxnhCCS!FSq1L-NNXV7Bg21XTw7ksXPI4DfFDI927Iq1iijyJc0


Version 0.6
- failover pool support
- introduce configuration file
- introduce intensity option
- support per gpu temp-target via cfg-file
- support per gpu intensity via cfg-file
- json-rpc report gpu_name, pci_bus_id, pci_device_id
  gpu_uuid
- improve error reporting
- improve handling of invalid server resposes
- improve reconnection in cases where servers
  become reachable but respond with malformed data
- ui: add colorized output
- web-ui: use dark colors

Linux x64:
executable sha1 c5951521699e8f58f339a3a194a8896b27ebfba8
https://drive.google.com/file/d/1Q8kCklgXS9SctNARYyg48RXv4qEyUcQs
https://mega.nz/#!14IS2AbI!Zl9z9ZrWoNRGPQ59Fog15f4tIdPcT40u3fOMkVavHlo

Win x64:
https://drive.google.com/file/d/1kv9tqmrxsedidhws7OtFnEI7j3sJT1sR
https://mega.nz/#!1kRxQRSD!I3ryiEI5eT7datW842QNESyBQpZY6PILYS4HNIEHpYY
executable sha1 4bd604a5de330dae96a5df0525fb44363b1b22ec


Version 0.5.8
con: add support for protocol v.01000020                                                                                      
link openssl statically

Linux x64:
sha1 827b137e061fe1da8f8fbcea12cec22449d3ae13
https://drive.google.com/file/d/19fSFYqoeOhOkxQqKnGpNI3n-7TIppHnq
https://mega.nz/#!vx0QgQ7K!dG2hbzh1KINYsqL9v75YlxpKwq0U7KaualMN6uFW2JY

Win x64:
sha1 bbdab3f21bbce114d972045b9056c64f299ef7a7
https://drive.google.com/file/d/1PrFBWLZmdcoqyCOGiKcy0qO4mKrl7Tm_
https://mega.nz/#!v8lBWYoT!8gRWHeDU9dbZALQfkLUhgi9YuPAIJOyR2gXZxnPpgao


Version 0.5.7
reduce cpu load
minor performance improvements
con: use single pool connection
con: ssl: clear session data before reconnect
nvml: handle invalid values

Linux x64:
sha1 e7a38b2771e2b8d976435a6b569053d103bc9465
https://drive.google.com/file/d/11e1fSA2cl14xRL8sMA9BA3AhBGQNn4e4

Win x64:
sha1 808dcfc440beb6af98c84420e1299b211033c8ce
https://drive.google.com/file/d/1WTJylr2uJYy_7-5isuja34ONu0WEBhin


Version 0.5.6
reduce cpu load
minor performance improvements

Linux x64:
sha1 ec26a36c131dcc28aaf747714b55a71ff0167836
https://drive.google.com/file/d/1Ql-hwlI0IsgOGZdOZQe45SfSMdpmgzE_

Win x64:
sha1 b9c818fedd53898cf010f129acefbd7be6fe8812
https://drive.google.com/file/d/1H6u2ZWKaQ4lGcvg2X-cqA0gP966e2-Bk


Version 0.5.5
con: handle failure of dev share pool
con: disable reconnect limit
logfile: disable buffering

Linux x64:
sha1 43e52e9494510f98d28152bf2a2e162b7381c379
https://drive.google.com/file/d/1USpxkXUd0Xs29XizG63OGf9w60UGl4Gf

Win x64:
sha1 4bf1cd44c1bfce6b95e0d6fd26139dcde5d818b2
https://drive.google.com/file/d/1Zmq6ZuOStO-bNjK-4XKacKpe66hk1Qwt


Version 0.5.4
fix: device selection
fix: latency reporting
add logfile support
add option to disable auto reconnect
handle disconnect periods in web/json stats
add basic GPU responsiveness infrastructure
ui: change timestamps to ISO 8601 format
webui: report: uptime contime server user port
json: report: uptime contime server user port version

Linux x64:
sha1 02816754ab46191109bed9f89f17c073000b0950
https://drive.google.com/file/d/12MmjEm-qIpdK7zc22jatQWD9WmAFIOLy

Win x64:
sha1 e4312b348a04348a6229b0098b7862a1306b0145
https://drive.google.com/file/d/1B7d6Qp9WzSPmCC6Yp0BgeyFQ9osdBTqq


Version 0.5.3
cpu: reduce cpu load
con: fix recon loop
con: network latency measurement
ui:  add accepted/rejected shares ratio
ui:  add network latency
ui:  add information about selected devices

Linux x64:
sha1 0583d59930a6160f80db612eae089c84712b5f44
https://drive.google.com/file/d/0B5QcL4Z6nYEPcHRuZk5LV3NXOUE

Win x64:
sha1 d20670a2f41834dfc0828946e187197c1c260992
https://drive.google.com/file/d/0B5QcL4Z6nYEPSm9yTmhDV1gwZ3c


Version 0.5.2
fix a race that might have corrupted shares
net: tcp: disable the Nagle algorithm
  which reduces network latency in some circumstances
net: lower keepalive for nat-networks

Linux x64:
sha1 6d07aee8134dcd122d68b2b35cf10886dccbbfb8
https://drive.google.com/file/d/0B5QcL4Z6nYEPSzRBU1hoQXp3elk

Win x64:
sha1 b57d667e0a4552d2a1a833b33f6aab94a43ecd8e
https://drive.google.com/file/d/0B5QcL4Z6nYEPWVl4bFVhbmhfQjg


Version 0.5.1
con: fix reconnection might have give up too early
con: enable TCP Keepalive to prevent nat-routers from
dropping the connection.
ui: output rejected shares
ui: optional output of timestamps
telemetry: make web ui mobile friendly + add share stats
telemetry: json: add more GPU stats

Linux x64:
sha1 98e52d1357187397f201a5379b4e42919e82c7a8
https://drive.google.com/file/d/0B5QcL4Z6nYEPNTdVUEYtbzA4Yjg


Version 0.5
con: support set_extranonce rpc
con: improve handling of temporary slow network conditions
con: add monitoring support using web browser
con: add monitoring support using json-rpc
mp: rebalance queue sizes - this improves the solution rate as
seen by the pools, especially on pools that submit new jobs often

Linux x64:
sha1 d4debdcf80b2b50c0ed06b55bcaa72262afe41bc
https://drive.google.com/file/d/0B5QcL4Z6nYEPQWgwbVl0Z0xOcGc


Version 0.4.5
fix: bad miner performance when using --temp-target
on systems with many GPUs
reduce temperature controller cpu usage

Linux x64:
sha1 e568c6d4353b35c45d25a00b3d172eda7c11ba11
https://drive.google.com/file/d/0B5QcL4Z6nYEPVGpuTkNCMXo2T2M


Version 0.4.4
fix: Sol/W might have produced wrong values
fix: Sum not shown if --dev was used
fix: memory leak on reconnection
improve stability

Linux x64:
sha1 60e0cad05e93f57330bf8271799d2a5417d904bc
https://drive.google.com/file/d/0B5QcL4Z6nYEPQjdYS2N5R1UyTkU


Version 0.4.3
stratum: support more pools
ui: flush stdout stream
con: improve reconnections
performance improvements 0.3-0.5 %

Linux x64:
sha1 d98792ad93b85f85df66d6e7ab9c2f00aec27cfd
https://drive.google.com/file/d/0B5QcL4Z6nYEPVlBrZmdndHE0VXc


Version 0.4.2
mp: more accurate time measurements
ui: calculate efficiency

Linux x64:
sha1 e53545c72c7a378a637962352b63498693f31f9e
https://drive.google.com/file/d/0B5QcL4Z6nYEPTGh5YUFfYUpGUjA

Version 0.4.1
improve performance

Linux x64:
sha1 4973ab1b90d2c31cc26671c71e71535727cc9e63
https://drive.google.com/file/d/0B5QcL4Z6nYEPaWlTa2VuTklMejQ


Version 0.4.0
temperature controller added
cpu: reduce CPU usage
con: improve pool support
minor performance improvement

Linux x64:
sha1 db0a9bcb7784686f0582d2c3f79be1bbc5be5be5
https://drive.google.com/file/d/0B5QcL4Z6nYEPRk5ITDMwck5jaWs


Version 0.3.2
mp:  improve stability
con: support more pools
con: more robust reconnection support for unreliable
  servers which often drop the connection                                                                                  
ui:  report GPU temperature

Linux x64:
sha1 cc8b83289055bfb659c57937f08f520e928c7fdc
https://drive.google.com/file/d/0B5QcL4Z6nYEPZFpHSzhDR1FSYnM


Version 0.3.1
cpu: reduce CPU usage

Linux x64:
sha1 7077d093b36b1183a80c53abc7b7268e
https://drive.google.com/file/d/0B5QcL4Z6nYEPWVNoS2dfcDdYdEU


Version 0.3
con: support encrypted connections

Linux x64:
sha1  58a36a30ea10d58731c5304a4941744de310be5d
https://drive.google.com/file/d/0B5QcL4Z6nYEPckdSYjF3QlJHN0U


Version 0.2.2
con: Miner reconnects automatically on connection loss now,
 e.g. on DSL-disconnects / public IP-changes
UI: report submitted devfee-shares

Linux x64:
sha1 04551d3f7c9369bccb8cec4bfe404bedee35cf45
https://drive.google.com/file/d/0B5QcL4Z6nYEPQUlVZUJ5LUVNOHM


Version 0.2.1
Performance improvements:                                                                                                      
- about 4% on Maxwell                                                                                                            
- about 1% on Pascal

Linux x64:
sha1 b7ae3a4b1f3567f3855f1b040fc48c253969a6f3
https://drive.google.com/file/d/0B5QcL4Z6nYEPeFVqLWxtNTVvS3M


Version 0.2
- Improved pascal perfromance.                                                                      
- Added support for concurrent mining on multiple GPUs.
- Be asynchronous where possible.
- Security work

Linux x64:
sha1 0ebdcce71fc3ed10f7c70b1d8818f051e2d91613
https://drive.google.com/file/d/0B5QcL4Z6nYEPOVJZd3JrVnR4aEU

[1713582112]

1713582112

[1713582112]

1713582112

[CjMapope]

Scans clean with ESET, but dammit there's alot of new malware thats undetectable too... :/

no source you would share just for quick perusal eh?

why the brand new account? you wrote a zcash miner but NEVER made an account here ever? :/

[dstm]

It's clean. It does only one thing, it mines ZEC.                                                                                                                  
Launch it in a VM and monitor everything u can to get confident.

[dstm]

Ah, I see. Malware seem to be a huge problem on this forum. I took a look for a description to setup a VM quickly. It seems it not that easy. PCI/GPU-Passthrough seems to be very hardware dependent - not that easy to setup. I've really no good idea how to provide security for you, without opening the source. Well I guess someone has to try it out and report. Sry, but I have no better ideas for this.

[Subw]

sure will try it out as soon as you publish sources

[dstm]

I've done some comparisons now. It seems there are not that many cuda miners out there.                                        
The fastest miner that I'm aware of averages at about 72 Sol/s on the GPU I've mentioned in the OP.

[Quarx2k]

dstm, can you please compile for windows? I can test with 1080 TI  

[dstm]

Quarx2k  That would be really a great benchmark.

Unfortunately it's currently written specifically for linux and uses direct OperatingSystem calls to be as fast as possible, to have very low cpu usage and to consume energy as little as possible. However if people are interested in a Windows version I'll port it. First we have to check how it performs on something faster then a 750ti However my test are very promising.

[doktor83]

Why do you always compare with the second fastest?
Which one is the fastest one and how much sol does it give on a 750ti?

[dstm]

Oh I wrote it a bit confusing. The second fastest is actually the fastest I could find. I called it second fastest because it's slower then mine miner. I've edited my posts now.

[laik2]

Just finished benching(Couldn't find out how exactly to use --dev to enable all gpus, tried 012345 then 0,1,2,3,4,5 then 0 1 2 3 4 5 and still only one enabled)

GTX 1070 MSI Armor(Micron)
core +125 , mem+750

your miner: 340Sol/s
ewbf: 410Sol/s
ccminer/nicehash miner: 370Sol/s


A source code would be greatly appreciated.

[dstm]

Thx a lot.
                                                                                                                    
Actually that looks good, since I've never tested it on pascal class GPUs so there is zero optimization for them. So it looks really good.

What column in the output did u use when reading my Sol/s? "Avg" or "Sol/s"?                                                                        
Concerning the --dev option. Currently only one GPU per process is supported, however u could start different processes for different GPUs.

[LuaKT]

Tested on a 1080 ti.

Code:

GeForce GTX 1080 Ti 28
Connected
>  Sol/s: 428.340577 Avg: 428.340577 Shares/s: 0.500000 luck: 0.001167  +++++
   Sol/s: 408.492722 Avg: 418.416650 Shares/s: 0.450000 luck: 0.001075  ++++
   Sol/s: 420.814390 Avg: 419.215897 Shares/s: 0.433333 luck: 0.001034  ++++
>  Sol/s: 417.192518 Avg: 418.710052 Shares/s: 0.425000 luck: 0.001015  ++++
   Sol/s: 427.623539 Avg: 420.492749 Shares/s: 0.400000 luck: 0.000951  +++

I get about 740 Sol/s using EWBF

[dstm]

Thx, finally 1080ti                                                                                                          

1.76 times faster. That's a much bigger difference then I expected. Not sure if it's due to design or due to optimization. I'll have to take a closer look at pascal manuals.
Also it seems not to scale compared to a 1070.

[reb0rn21]

If you want fee you must be faster then ewbf, else open source it

[dstm]

I'll do it if I can't get enough performance.
Give me some tries first it's the first time it runs on pascal GPUs.

[dstm]

LuaKT

Could u please run another test. Set your cpu-governor to performance mode this time. The miner is highly multithreaded. Since it scales so bad compared to a 1070 - it could be due to scheduling latencies. So if u have some time pls test with cpu-scaling governor set to performance.

[LuaKT]

Setting it to performance does provide a small increase

Code:

GeForce GTX 1080 Ti 28
Connected
>  Sol/s: 454.747435 Avg: 454.747435 Shares/s: 0.500000 luck: 0.001100  +++++
   Sol/s: 460.647203 Avg: 457.697319 Shares/s: 0.550000 luck: 0.001202  ++++++
>  Sol/s: 463.059314 Avg: 459.484650 Shares/s: 0.500000 luck: 0.001088  ++++
   Sol/s: 444.676640 Avg: 455.782648 Shares/s: 0.475000 luck: 0.001042  ++++
   Sol/s: 455.486465 Avg: 455.723411 Shares/s: 0.440000 luck: 0.000965  +++
   Sol/s: 457.690007 Avg: 456.051177 Shares/s: 0.500000 luck: 0.001096  ++++++++

[laik2]

Fee or nofee, open or closed source it wont make any difference because there is already open source cuda kernel, just noone is interested as of devs to improve it for free
Open source miners are dead.

[dstm]

LuaKT

Nice,                                                                                                                         
                                                                                                                               
so it's latency dependent. I'm not sure how much time u have. If u have some, I would compile some versions that should show where the bottleneck is. Also could u pls check the cpu usage. If u have some time for all this stuff ofc.