dstm (OP)
|
|
December 04, 2017, 11:02:33 AM |
|
Thx for the report. Are you thinking about a unix daemon? This isn't available on windows, I don't want to diverge the versions without a good reason, I'll do ofc if there is one. If you're mainly interested in disabling the output you can redirect it to /dev/null or/and start it in your init system (e.g. using systemd).
What colors and where would improve readability?
Hi, I'm using dstm on Windows on my mining PC. Unfortunately Linux is not an option. I think it shouldn't be a big problem but I understand if you don't want to implement it. It was an idea which could be useful because it won't take a space on taskbar. Also thanks for commands. It would be helpful inthe future. And about colors. This is my idea for your miner look: What do you thing? To my mind, currently there are too many informations in the console.
Thx Atexor, suggestions are welcome. The idea is to have a compact presentation of the state, such that anytime you look at the output you'll able to see the current state and the history (how things evolve) without scrolling. The most interesting values (Sol/s, Sol/W) are intentionally located to the left because of this. Just ignore the remaining numbers if they aren't important for you. In my version summary for GPU(s) will appear every X minutes or shares .
In this case you would have to scroll to see the summary, it would also make it hard to see how things evolve with time, currently you simply have to pick a column an check the values top down to see how things evolve. "+++++" instead of number is strange ideaso I changed it. More info would be avaible via telemetry.
Thee '++++' are there to make the representation compact, if you put every single submitted share/received job on a separate line, you will waste a lot of screen space, so you probably have to scroll to see the summary it also makes it hard to see how thing evolve. In addition the '++++' make it easy to check how the share distribution looks like, so if something goes wrong it's easy to see.
|
|
|
|
dstm (OP)
|
|
December 04, 2017, 11:06:47 AM |
|
+ 1 for this idea. I just think you should add 3 colors for the temperatures. Below 65 - green. 65 to 79 - yellow, and 80+ - red. Also it will be good to see how many accepted shares, and how many rejected, also green for accepted and red for rejected. Per GPU and total. It would be much much better than this current mess of information. I cannot understand half of them, and really don't need them as an average miner. I though about colors for GPU temperaturs, hovewer for different models there is different level of "hot" and could be difficult do maintain and change (in config) fpr each of the cards. Also using too much colors could make it too unreadable. Rejected shares appears sometimes for me but their form Recently I had one. I think it could has red color and show as normal information. If everything works normally, there is no reason to color it at green every line. Take a look: https://i.imgur.com/5xjthIu.pngI don't know what "21" means there - number of rejected share? Also I've bringed back "*" to show where was devfee but in different place. It's the error message like it was returned by the server. For me showing year in date is unpractible (month rather too but i kept it), like that accepted shares ratio too (always equal 1.0). It's not necessary to show every line Avg. Sol/s because it doesn't change so often. On summary it would be great. Of course this is my proposals and for someone other ideas could be better. Nonetheless I hope that dstm will like my suggestions.
The accepted share ratio and the Avg. Sol/s change - especially if something goes wrong. By the way - dstm. Miner started to works better to me and it haspractically a bit more better score than EWBF. Hoverwer it needs much time to spin up. I am wondering why with ewbf/dstm miner I'm able to overclock core only to 60/75mhz max, while other people on 1070/1080ti could overclock it further than 100mhz.
I'm not sure what you mean by 'spin up'. ZM uses all available GPU 'power' right from the start.
|
|
|
|
Atexor
Member
Offline
Activity: 155
Merit: 10
|
|
December 04, 2017, 11:52:52 AM |
|
Rejected shares appears sometimes for me but their form Recently I had one. I think it could has red color and show as normal information. If everything works normally, there is no reason to color it at green every line. Take a look: https://i.imgur.com/5xjthIu.pngI don't know what "21" means there - number of rejected share? Also I've bringed back "*" to show where was devfee but in different place. It's the error message like it was returned by the server. Okay thank you. And if I may ask - what do you think about my proposal of your miner looks? By the way - dstm. Miner started to works better to me and it haspractically a bit more better score than EWBF. Hoverwer it needs much time to spin up. I am wondering why with ewbf/dstm miner I'm able to overclock core only to 60/75mhz max, while other people on 1070/1080ti could overclock it further than 100mhz.
I'm not sure what you mean by 'spin up'. ZM uses all available GPU 'power' right from the start. For me your miner starts from 520-530 Sols/s and then hashrate (sol/s?) increases ~550. Avg/s takes probably counts entire sols/s since start (reconnect to pool) and it works a bit strange because it takes too long term to detect average hashrate ans it increases slowly entire day. Avg from 530 to 550. Take a look on these timestamps: 2017-12-03 12:53:55| GPU0 64C Sol/s: 529.5 Sol/W: 3.22 Avg: 529.1 I/s: 283.9 Sh: 5.67 0.97 50 +++ 2017-12-03 15:44:39| GPU0 65C Sol/s: 535.0 Sol/W: 3.23 Avg: 530.1 I/s: 283.7 Sh: 4.22 1.00 53 2017-12-04 12:46:12| GPU0 65C Sol/s: 555.2 Sol/W: 3.43 Avg: 550.1 I/s: 296.0 Sh: 4.22 1.00 53 ++ Or gif: https://imgur.com/a/sNlNYThere were one disconnect on suprnova so everything started from the beginning.
|
|
|
|
dstm (OP)
|
|
December 04, 2017, 12:43:34 PM |
|
Rejected shares appears sometimes for me but their form Recently I had one. I think it could has red color and show as normal information. If everything works normally, there is no reason to color it at green every line. Take a look: https://i.imgur.com/5xjthIu.pngI don't know what "21" means there - number of rejected share? Also I've bringed back "*" to show where was devfee but in different place. It's the error message like it was returned by the server. Okay thank you. And if I may ask - what do you think about my proposal of your miner looks? I've tried to explain the disadvantages it has in my previous post. They are especially visible in your scrolling example. https://bitcointalk.org/index.php?topic=2021765.msg25716313#msg25716313By the way - dstm. Miner started to works better to me and it haspractically a bit more better score than EWBF. Hoverwer it needs much time to spin up. I am wondering why with ewbf/dstm miner I'm able to overclock core only to 60/75mhz max, while other people on 1070/1080ti could overclock it further than 100mhz.
I'm not sure what you mean by 'spin up'. ZM uses all available GPU 'power' right from the start. For me your miner starts from 520-530 Sols/s and then hashrate (sol/s?) increases ~550. Avg/s takes probably counts entire sols/s since start (reconnect to pool) and it works a bit strange because it takes too long term to detect average hashrate ans it increases slowly entire day. Avg from 530 to 550.
This might happen if something consumes GPU-time when you start the miner, e.g. if it's also used to render the screen. This is how an average value works mathematically, it doesn't detect something there is also no way to speed up it, since it wouldn't be an average anymore. Take a look on these timestamps: 2017-12-03 12:53:55| GPU0 64C Sol/s: 529.5 Sol/W: 3.22 Avg: 529.1 I/s: 283.9 Sh: 5.67 0.97 50 +++ 2017-12-03 15:44:39| GPU0 65C Sol/s: 535.0 Sol/W: 3.23 Avg: 530.1 I/s: 283.7 Sh: 4.22 1.00 53 2017-12-04 12:46:12| GPU0 65C Sol/s: 555.2 Sol/W: 3.43 Avg: 550.1 I/s: 296.0 Sh: 4.22 1.00 53 ++ Or gif: https://imgur.com/a/sNlNYThere were one disconnect on suprnova so everything started from the beginning.
|
|
|
|
m1n1ngP4d4w4n
Full Member
Offline
Activity: 224
Merit: 100
CryptoLearner
|
|
December 04, 2017, 03:27:41 PM |
|
It would be great to have like SGminer a console where you have a fixed dynamic top that show all the summary information like on the web interface (pool, uptime, gpu's, efficiency and so on), then under this make the scrolling log. I find this interface to be the best. What do you think ?
|
|
|
|
marco.ruiz
Newbie
Offline
Activity: 29
Merit: 0
|
|
December 04, 2017, 05:55:27 PM |
|
on nicehash 1 gpu = 1 worker? $rigName notworking?
Yup, same here. Also a lot of disconnection problems with NH. Same here, a little annoying to have multiple workers on NH stats page but not a deal breaker, it still has better performance than NH Excavator and EWBF. When i start the miner i get a lot of "Share above target" but after a few minutes it works like a charm. Just upgraded to 0.5.6 a few minutes ago, will report back later.
|
|
|
|
marco.ruiz
Newbie
Offline
Activity: 29
Merit: 0
|
|
December 04, 2017, 06:27:27 PM |
|
You can always use the telemetry server implemented. It has really nice clean interface. --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] dstm, here's my suggestion for the telemtry server i wanted to post a while ago: a way to sort columns so we can identify quicker the less efficient GPU, lowest hashrate, hottest, etc. Maybe using datatables.net plugin? I know you tried to keep it as light as possible but using datatables.net would save you a lot of time. If you want i could help, send me a private message.
|
|
|
|
m1n1ngP4d4w4n
Full Member
Offline
Activity: 224
Merit: 100
CryptoLearner
|
|
December 04, 2017, 06:29:53 PM |
|
--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] dstm, here's my suggestion for the telemtry server i wanted to post a while ago: a way to sort columns so we can identify quicker the less efficient GPU, lowest hashrate, hottest, etc. Maybe using datatables.net plugin? I know you tried to keep it as light as possible but using datatables.net would save you a lot of time. If you want i could help, send me a private message. That's why i stated above "like on the web interface", i was speaking about how the console actually look. I obviously use the telemetry already, but just if we want a "cleaner" console.
|
|
|
|
NameTaken
|
|
December 04, 2017, 06:35:38 PM |
|
ncurses console does not play well with running under systemd.
|
|
|
|
akuci
|
|
December 04, 2017, 09:58:44 PM |
|
It would be great to have like SGminer a console where you have a fixed dynamic top that show all the summary information like on the web interface (pool, uptime, gpu's, efficiency and so on), then under this make the scrolling log. I find this interface to be the best. What do you think ? Really want to hear what dstm think about this. Maybe the best suggestion so far.
|
|
|
|
akuci
|
|
December 04, 2017, 10:24:39 PM |
|
2017-12-04 18:06:35|# connection closed by server r:0 2017-12-04 18:06:35|# reconnecting 2017-12-04 18:07:01|# server not responding - retrying... 2017-12-04 18:07:16|# server not responding - retrying...
I'm getting these all the time on NH pool. Sometimes it's every 24 hours, but sometimes it's few times a day.
My friend on different setup, different ISP, and still have the same problem at the same time as me.
|
|
|
|
Atexor
Member
Offline
Activity: 155
Merit: 10
|
|
December 04, 2017, 11:41:45 PM Last edit: December 05, 2017, 08:12:50 AM by Atexor |
|
Sorry, I didn't see your reply. I was focused on second post. On the one hand you've right to see it in a compact form, hovewer on the other hand it doesn't look clearly becuse there are too much "compressed" informations. Like --time parameter, maybe could you think about disabling few of printfs such as iterations and shares? Later I will think on other proposal how to get it look more clearly if u don't mind. This might happen if something consumes GPU-time when you start the miner, e.g. if it's also used to render the screen.
It is logical, hovewer I'm connected to iGPU (integrated graphics). So it is strange because no one program - except miner - interfere with graphics card. This is how an average value works mathematically, it doesn't detect something there is also no way to speed up it, since it wouldn't be an average anymore.
I know, but as I've shown on my gif the average looks a bit strange in my case because it's increasing and increasing for many hours until it will achieve it's operating point. In my opinion taking into consideration latest 100-200 records (circular buffer) should be fine. --------------------------------------------- edit: This is what I said before: I had been disconnected from suprnova pool and few minutes after reconnect my hashrate started from ~530: https://i.imgur.com/egeP7La.png
|
|
|
|
dstm (OP)
|
|
December 05, 2017, 12:08:13 PM |
|
It would be great to have like SGminer a console where you have a fixed dynamic top that show all the summary information like on the web interface (pool, uptime, gpu's, efficiency and so on), then under this make the scrolling log. I find this interface to be the best. What do you think ? Thx for this example. It has the advantage of having a static region for stats so it's easy to read - however it has also disadvantages - you have no history (you can't scroll) so you can't see how things evolve, there is also no correlation between events and performance values, which are very useful if something goes wrong - using curses has also it's own disadvantages.
|
|
|
|
dstm (OP)
|
|
December 05, 2017, 12:10:58 PM |
|
It would be great to have like SGminer a console where you have a fixed dynamic top that show all the summary information like on the web interface (pool, uptime, gpu's, efficiency and so on), then under this make the scrolling log. I find this interface to be the best.
What do you think ?
You can always use the telemetry server implemented. It has really nice clean interface. --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] dstm, here's my suggestion for the telemtry server i wanted to post a while ago: a way to sort columns so we can identify quicker the less efficient GPU, lowest hashrate, hottest, etc. Maybe using datatables.net plugin? I know you tried to keep it as light as possible but using datatables.net would save you a lot of time. If you want i could help, send me a private message. Thx for the suggestion. I don't want to depend on javascript unless it provides something very useful. The webui works currently pretty well even on terminal based browsers which don't support javascript. There are not that many lines you have to check also you have usually to check each GPU as a whole separately - so I don't think this provides real benefits.
|
|
|
|
dstm (OP)
|
|
December 05, 2017, 12:23:36 PM |
|
Sorry, I didn't see your reply. I was focused on second post. On the one hand you've right to see it in a compact form, hovewer on the other hand it doesn't look clearly becuse there are too much "compressed" informations. Like --time parameter, maybe could you think about disabling few of printfs such as iterations and shares? Iteration are very useful since it's a good way to measure the performance of your GPUs. The solution rate isn't a good measure for this since it's probabilistic. Later I will think on other proposal how to get it look more clearly if u don't mind.
Thx, all suggestions are welcome. This is how an average value works mathematically, it doesn't detect something there is also no way to speed up it, since it wouldn't be an average anymore.
I know, but as I've shown on my gif the average looks a bit strange in my case because it's increasing and increasing for many hours until it will achieve it's operating point. In my opinion taking into consideration latest 100-200 records (circular buffer) should be fine. The average has manly two purposes, it's useful for people who are running their systems for long periods: the performance may vary during this time because of thermal conditions or CPU-load caused by other processes - so an overall average is very useul. It also there to be able to check the solution rate reported by the pool. --------------------------------------------- edit: This is what I said before: I had been disconnected from suprnova pool and few minutes after reconnect my hashrate started from ~530: https://i.imgur.com/egeP7La.pngThis looks strange, it's a pretty significant difference, I've never seen such a behavior on my systems. Do you have other systems to check if this is reproducible?
|
|
|
|
shibob
Full Member
Offline
Activity: 672
Merit: 154
Blockchain Evangelist.
|
|
December 05, 2017, 01:52:19 PM |
|
Anyone got frequently hashrate drop problem recently (Drop to 0 in about 20 mins, but when I checked miner, it still hashing as normal)? Look like after few days, it happen again to all of my miners. I mined in flypool, asia1 server, port 3333.
|
|
|
|
Dylmine
Newbie
Offline
Activity: 4
Merit: 0
|
|
December 05, 2017, 04:02:50 PM |
|
Simple question: does DSTM support 1070 Ti? Thinking of building a new mining rig based on exclusively these GPUs.
|
|
|
|
VoskCoin
|
|
December 05, 2017, 04:34:35 PM Last edit: December 05, 2017, 04:48:51 PM by VoskCoin |
|
anyone having issues with 12 cards running in DSTM on linux? works fine with ewbf // anything I need to add or change?
edit had to reboot and just seems slow to load the cards, but its running now hope its stable pickedup ~200 hashes over ewbf ~8070-8270
|
|
|
|
Darkrage
Newbie
Offline
Activity: 3
Merit: 0
|
|
December 05, 2017, 04:36:42 PM |
|
I downloaded the Winx64 version Version 0.5.6 and the SHA-1 hash does not match.
Any reason why?
|
|
|
|
NameTaken
|
|
December 05, 2017, 04:43:58 PM |
|
I downloaded the Winx64 version Version 0.5.6 and the SHA-1 hash does not match.
Any reason why?
Matches with the binary. $ sha1sum zm.exe b9c818fedd53898cf010f129acefbd7be6fe8812 zm.exe
|
|
|
|
|