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

[car1999]

I can confirm that the CPU load is very low on a 12 gtx1060 rig with Intel G4560.

[1714810813]

1714810813

[Nokia555]

I have released EthControl for ZM also. Linux support available and has remote monitoring support. (https://ethmonitoring.com)

Code:

wget https://github.com/JamesSmith2/EthControlLinux/releases/download/1.0/EthControl.gz && gunzip EthControl.gz && chmod +x EthControl

[smartieee]

Hi guys,

Are there any timeframes for us windows users at all? As mentioned before in this forum EWBF is nice and fast but not that stable, I am sure that there are more peoples than myself itching to try DSTM and hopefully switch over to DSTM as an alternative :>

Thanks in advance

[dstm]

Hi guys,

Are there any timeframes for us windows users at all? As mentioned before in this forum EWBF is nice and fast but not that stable, I am sure that there are more peoples than myself itching to try DSTM and hopefully switch over to DSTM as an alternative :>

Thanks in advance

I've done already some porting work, however I had to interrupt the porting work because of some request/issues some users reported. Sry, I have no eta, but yes, I'm working on it.

[smartieee]

NOT GOOD ENOUGH!!

Um wait...no probbo, we will get it when we get it, thanks for the quick reply!

[kryztoval]

Regarding my network issue, it seems it is a problem with linux and TCP_DEFER_ACCEPT

as explained here:
https://labs.ripe.net/Members/gih/the-curious-case-of-the-crooked-tcp-handshake

I'll keep investigating.

[balkeep]

Also after sorting out overclocking stuff I'm getting this annoying bug that 1 card doesn't get loaded properly (had it with ewfb's under w10 too occasionally).

   GPU0 61°C  Sol/s: 421.3  Sol/W: 3.98  Avg: 417.0  I/s: 227.2  Sh/s: 0.031  +
   GPU1 66°C  Sol/s: 446.8  Sol/W: 3.52  Avg: 439.3  I/s: 235.0  Sh/s: 0.015
   GPU2 64°C  Sol/s: 438.2  Sol/W: 3.52  Avg: 438.6  I/s: 234.8  Sh/s: 0.046
   GPU3 62°C  Sol/s: 305.1  Sol/W: 2.83  Avg: 310.7  I/s: 165.6  Sh/s: 0.015  +
   GPU4 62°C  Sol/s: 314.8  Sol/W: 2.85  Avg: 311.4  I/s: 165.8  Sh/s: 0.038  +
   GPU5 64°C  Sol/s: 492.7  Sol/W: 4.55  Avg: 488.9  I/s: 260.5  Sh/s: 0.038
   ========== Sol/s: 2418.9 Sol/W: 3.54  Avg: 2405.9 I/s: 1289.0 Sh/s: 0.185

Where GPU0 and GPU5 are both identical EVGA 1080 FTW

nvidia-smi also shows gpu utilization of 87% on that card. Anything I can do?

Fri Oct  6 13:32:21 2017
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 384.90                 Driver Version: 384.90                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 1070    On   | 00000000:01:00.0  On |                  N/A |
| 30%   63C    P2   122W / 125W |    569MiB /  8113MiB |    100%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX 1070    On   | 00000000:02:00.0  On |                  N/A |
| 24%   62C    P2   123W / 125W |    528MiB /  8114MiB |    100%      Default |
+-------------------------------+----------------------+----------------------+
|   2  GeForce GTX 106...  On   | 00000000:04:00.0  On |                  N/A |
| 27%   59C    P2   110W / 120W |    506MiB /  6072MiB |    100%      Default |
+-------------------------------+----------------------+----------------------+
|   3  GeForce GTX 106...  On   | 00000000:05:00.0  On |                  N/A |
| 28%   60C    P2   110W / 120W |    506MiB /  6072MiB |    100%      Default |
+-------------------------------+----------------------+----------------------+
|   4  GeForce GTX 1080    On   | 00000000:06:00.0  On |                  N/A |
| 18%   58C    P2    89W / 108W |    555MiB /  8114MiB |     84%      Default |
+-------------------------------+----------------------+----------------------+
|   5  GeForce GTX 1080    On   | 00000000:07:00.0  On |                  N/A |
| 23%   60C    P2   107W / 108W |    555MiB /  8114MiB |    100%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0       992      G   /usr/lib/xorg/Xorg                            33MiB |
|    0      1752      G   compiz                                        13MiB |
|    0      2307      C   /home/leo/mining_soft/zm/zm                  509MiB |
|    1       992      G   /usr/lib/xorg/Xorg                             7MiB |
|    1      2307      C   /home/leo/mining_soft/zm/zm                  509MiB |
|    2       992      G   /usr/lib/xorg/Xorg                             7MiB |
|    2      2307      C   /home/leo/mining_soft/zm/zm                  487MiB |
|    3       992      G   /usr/lib/xorg/Xorg                             7MiB |
|    3      2307      C   /home/leo/mining_soft/zm/zm                  487MiB |
|    4       992      G   /usr/lib/xorg/Xorg                             7MiB |
|    4      2307      C   /home/leo/mining_soft/zm/zm                  535MiB |
|    5       992      G   /usr/lib/xorg/Xorg                             7MiB |
|    5      2307      C   /home/leo/mining_soft/zm/zm                  535MiB |
+-----------------------------------------------------------------------------+

So far I've been using this settings:

nvidia-smi -pm 1
nvidia-smi -i 0 -pl 125
nvidia-smi -i 1 -pl 125
nvidia-smi -i 4 -pl 108
nvidia-smi -i 5 -pl 108
nvidia-smi -i 2 -pl 120
nvidia-smi -i 3 -pl 120

and

sudo nvidia-settings -a "[gpu:0]/GPUGraphicsClockOffset[3]=100" -a "[gpu:0]/GPUMemoryTransferRateOffset[3]=0"
sudo nvidia-settings -a "[gpu:1]/GPUGraphicsClockOffset[3]=100" -a "[gpu:1]/GPUMemoryTransferRateOffset[3]=0"
sudo nvidia-settings -a "[gpu:2]/GPUGraphicsClockOffset[3]=100" -a "[gpu:2]/GPUMemoryTransferRateOffset[3]=0"
sudo nvidia-settings -a "[gpu:3]/GPUGraphicsClockOffset[3]=100" -a "[gpu:3]/GPUMemoryTransferRateOffset[3]=0"
sudo nvidia-settings -a "[gpu:4]/GPUGraphicsClockOffset[3]=100" -a "[gpu:4]/GPUMemoryTransferRateOffset[3]=0"
sudo nvidia-settings -a "[gpu:5]/GPUGraphicsClockOffset[3]=100" -a "[gpu:5]/GPUMemoryTransferRateOffset[3]=0"

[dstm]

Regarding my network issue, it seems it is a problem with linux and TCP_DEFER_ACCEPT

as explained here:
https://labs.ripe.net/Members/gih/the-curious-case-of-the-crooked-tcp-handshake

I'll keep investigating.

I don't think it's because of this, since handshaking/connection-establishment is working fine. What is happening is that at some point zm calls recv but the connection has been dropped, most probably by your router. You might log the network traffic with tcpdump for the stratum ports you're connecting to and pm them to me, I'll take a look if I can find something. You could also check kernels tcp keepalive settings in /proc/sys/net/ipv4/tcp_keepalive_..... . they might help with nat-routers.

[dstm]

Also after sorting out overclocking stuff I'm getting this annoying bug that 1 card doesn't get loaded properly (had it with ewfb's under w10 too occasionally).

Since all the other card are constantly on 100% it seem that it isn't an cpu performance issue - however check your cpu-load to be sure. Most likely it's a pcie-bus issue, the bus that GPU4 is connected to seems to have latencies. If you're unable to find the reason for this, I suggest you to swap the GPUs and put the slower 1060 into the pcie-bus that GPU4 is currently on.

[balkeep]

CPU load is between 30-40 usually so I guess it's fine.

Concerning the card that yields less Sols, I think it's cuz it's the first one, and it has something to do with Xorg running on it. At least on W10 it behaved the same when I had my monitor connected to it, instead of Intel gpu. Quite a few people complained about the same issue always being with the first card.

Also I noticed that after some time I start getting
msg buffer full
Why is that, is it hurting me somehow?

[dstm]

CPU load is between 30-40 usually so I guess it's fine.

Concerning the card that yields less Sols, I think it's cuz it's the first one, and it has something to do with Xorg running on it. At least on W10 it behaved the same when I had my monitor connected to it, instead of Intel gpu. Quite a few people complained about the same issue always being with the first card.

Also I noticed that after some time I start getting
msg buffer full
Why is that, is it hurting me somehow?

This might happen if one of your GPUs crashes/gets unresponsive and stops consuming messages - so check your overclocking. I'll add the gpu-id to this message in the next version, this way it will be easier to track the unresponsive GPU.

[spark.g]

remembered.

[dstm]

New 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

[UnknownX]

New 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

Great!
Maybe some news about windows release? On 24th of September you have told that we can expect it in in 1-2 weeks, any updates of dates?

[dstm]

New 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

Great!
Maybe some news about windows release? On 24th of September you have told that we can expect it in in 1-2 weeks, any updates of dates?

I'm working on it right now. If I'll get new issue reports I have to interrupt the work on it again otherwise it should be ready next week.

[jmayniac]

Version 0.5.1 seems to be working great on my rig.

You should delay the Windows version by a day every time someone asks for it.

Seriously though, I was getting about 2460 Sol/s on my rig under Windows, but I changed over to Linux and I get about 2600 Sol/s now.  Your miner is what convinced me to convert over and now there's no way i'd go back to Windows, unless I really had to.  Even with EWBF's miner, just switching over to Linux gave me about 100 Sol/s more with no other changes.

[dstm]

Version 0.5.1 seems to be working great on my rig.

You should delay the Windows version by a day every time someone asks for it.

Seriously though, I was getting about 2460 Sol/s on my rig under Windows, but I changed over to Linux and I get about 2600 Sol/s now.  Your miner is what convinced me to convert over and now there's no way i'd go back to Windows, unless I really had to.  Even with EWBF's miner, just switching over to Linux gave me about 100 Sol/s more with no other changes.

Nice

There is a reason for it. Apart from technical differences between the operating systems, Nvidia cares a lot about their cuda drivers on Linux, since that's where almost all cuda application are running on.

[zzz.to.the.moon]

Hello dstm,

Great news to hear that you have added so many things to the miner. Can you please tell us if the telemetry json displays the averages or the (semi) instant values? The long time averages are not really useful for status monitoring. For that only the instant or say 15 second upto several minutes averages work. Say if a GPU goes down, there is no way to tell that from the long time averages. This way the monitoring script can restart the miner or the whole machine.
Thank you, and keep up the great work!

[dstm]

Hello dstm,

Great news to hear that you have added so many things to the miner. Can you please tell us if the telemetry json displays the averages or the (semi) instant values? The long time averages are not really useful for status monitoring. For that only the instant or say 15 second upto several minutes averages work. Say if a GPU goes down, there is no way to tell that from the long time averages. This way the monitoring script can restart the miner or the whole machine.
Thank you, and keep up the great work!

This is how json-rpc is looking now, it's described in json-rpc.txt. Suggestions are welcome

Code:

Response:
<--{
    "id": 1,
    "result": [{
        "gpu_id":          0,         // GPU CUDA id
        "temperature":     0,         // current GPU temperature
        "sol_ps":          0.00,      // current Sol/s
        "avg_sol_ps":      0.00,      // average Sol/s
        "sol_pw":          0.00,      // current Sol/Watt
        "avg_sol_pw":      0.00,      // average Sol/Watt
        "power_usage":     0.00,      // current power usage
        "avg_power_usage": 0.00,      // average power usage
        "accepted_shares": 0,         // total amount of accepted shares
        "rejected_shares": 0          // total amount of rejected shares
    }],
    "error": null
}

[zzz.to.the.moon]

This is excellent! I will also give it a try over the weekend, and let you know.
The only thing I would add is the name of the GPU. For people with several types of GPU in one rig, it would really help to know which is the 1070 and 1080.