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

[crackzyee]

Thats nice!  I know my 1080TI can hit upwards of 800Sol around 1850-1900Mhz (it moves around) at 90% TDP if the temp is sub 50C.  Yeah my efficiency rides around 3.3-3.4 with the card on average.  Still good in my book.  Way better than when i started at only 680 Sols @ 100% TDP  It ends up being more efficient than my 1060's I got but least I topped out what my 2 small rigs can handle for cards and temp.  

Maybe I should look back into water cooling again.  Use to do it ages ago(14+ years), maybe its time to investigate it again.
[/quote]

How do you get 800+ @1900MHz, what card and what OC settings, can you share?

I am able to hit 790-805 @2015Mhz (270W)

[Deathman20]

Thats nice!  I know my 1080TI can hit upwards of 800Sol around 1850-1900Mhz (it moves around) at 90% TDP if the temp is sub 50C.  Yeah my efficiency rides around 3.3-3.4 with the card on average.  Still good in my book.  Way better than when i started at only 680 Sols @ 100% TDP  It ends up being more efficient than my 1060's I got but least I topped out what my 2 small rigs can handle for cards and temp.  

Maybe I should look back into water cooling again.  Use to do it ages ago(14+ years), maybe its time to investigate it again.

How do you get 800+ @1900MHz, what card and what OC settings, can you share?

I am able to hit 790-805 @2015Mhz (270W)

That was with +200 or +230 Core +500 Mem, 100% TDP.  But the card had to be cool, sub 50C for it to get to that high.  It was fluxatuing +1900Mhz, sorry on that one.  Thought I grabbed a screen shot but looks like I didn't.  I'd have to double check tonight if I can get it that low temp again.  I thought it was crazy too when I looked at it but hey I'd take it, if the card could keep cool.  I personally don't run it there but was having fun playing around with it just recently.

I'm guessing only reason it clocks that high with the boost is due to the lower temps.  If I messed with the frequency curve in MSI Afterburner I'm sure it could be tweaked a little more.

[fecker]

well on my 1080ti's i can get 3.4-3.6sol/w. my cards are hot tho, up to 70C. Im getting about 685-695 sols with 80% tdp. If they were running at <50C, I bet you can get close to 4sol/w

[Deathman20]

well on my 1080ti's i can get 3.4-3.6sol/w. my cards are hot tho, up to 70C. Im getting about 685-695 sols with 80% tdp. If they were running at <50C, I bet you can get close to 4sol/w

Solo my card ran at 58-60C, though with the other card in the rig it sits at 64C with the 90% TDP. If I stuck my PC in the garage right now I'm sure I'd get fantastic temps since its about -10C or so out.

[minebomb]

with reference to 12 GPU rigs:

i have noticed using DSTM on my 12 gpu rigs that GPU number 7 and 11 are on noticeably wamer than the others by 5-10degrees.

 I FIND THIS REALLY STRANGE.

Any suggestions?

Thanks in advance

This is nothing unusual, the temperature might differ even if you're using the same vendor/model. You could also check the GPU-utilization like it's reported by 'nvidia-smi' to make sure there are not software issues.

Thank you.
One more problem, miner reduces hash rates until almost zero but does not automatically reload or adjust accordingly as pool switches coin.  Not sure how to explain it. What can I do to resolve this. It only happens with DSTM and if I'm asleep it remains low for hours until I reload  manually.
Also some gpus drop to a far lower intensity than others but do not recover.
Happens on nicehash and miningpoolhub so I'm convinced it's the miner.
I'm fine with paying dev fees as long as I have good product support..

HELP SOMEONE.

[No.15]

Side note: if you want to get rid of the developer fee, then take a look at this tutorial: https://bitcointalk.org/index.php?topic=2773166.0

Happy mining.

Just curious, do you think the developers don't deserve to get paid or do you think the fee is to high?
I guess you could use the --nofee option as provided but it's my understanding the software slows down.

Anyway, I think these guys deserve to get paid for their work, I know I could not create these programs and because they make me money I am ok with it.

Do you think a pay once model would be better?  Apologies if this is not a correct topic for this thread.

[NoOneLt]

Some clarification due to recent fee debate.

This is ofc something I was thinking about - speaking to my colleagues/friends, so called 'consultants' etc. Most of them suggest me to increase it especially on performance improvements. To make it short: I won't change it.

Claymore's Zcash mining software was stated a couple of times - the information wasn't always accurate so that's what Claymore states on his page:
"...current developer fee is 2% if you use secure SSL/TLS connection to mining pool ...If you use unsecure connection to mining pool, current developer fee is 2.5% ... If you don't agree with the dev fee - don't use this miner, or use "-nofee" option."
"-nofee   set "1" to cancel my developer fee at all. In this mode some recent optimizations are disabled so mining speed will be slower by about 5%."

I'm not sure if the statements about zm's accuracy were given by intention or by uncertainty - however it's important to me to clarify this: All numbers reported by zm including the developer fee are exact.

Keep up I believe lot's of us including me is bit jealous seeing that numbers you make We are just people. IDK what i would do being in your place, really...

Just dont abandon software and make it superior, there is some space t improve functionality/stability, and don't loose connection with us, users here.

BR

@dstm - could you please include Manufacturer info of the card, could be only in JSON, if you have mixed rig's it is pretty handy!

[somethingiswrong]

can we use this with suprnova?will someone could give me the bat file for it. thanks

Yes you can use dstm with suprnova:
zm.exe --server zcl.suprnova.cc --port 4042 --user username.workername --pass 1234

[PickAxe131]

I tried my new 10x ASUS STRIX 1080ti OC cards today with 0.58 and got wonderful 740-760 Sol/s at 80% TDP@temperature 48-53 degrees celsius with each card Looking forward for 1080ti optimization!

[Tigel]

I created a Dockerfile for dstm 0.5.8.
I've found that using docker simplifies the management of my Ubuntu mining rig.
No more 'screen' or writing systemd/init scripts!

Requires https://github.com/NVIDIA/nvidia-docker

https://github.com/skinlayers/docker-dstm-zcash-cuda-miner


If you find this useful, feel free to toss some coin my way.

ETH: 0x75F4363b09166FDf8702F41502E90D21A3F2Afb9
ZEN: znY2eiUwdx9Pj4EzRkktZqRWHfEfX7o9gkW
FTC: 6nDBmQE65gy7BbcUYbYPttc9EYA7eNKN3g
ZCL: t1cSvJsEBWCipA4CY9K7oi9sKRQgb3EG76T
VIVO: VVmjio2zqGgfnvmkEU9bZcxXmFpNswusig
TZC: Tn6K9mw6bKc8R447CrpeEMazFcve8vEg1a
LBC: bXp34GVRLnQgTTid9x6nMorHYjq1GzFwz7
GBX: GTeMA2kehXf2E7DqPGEGVSsb1gQWxpg5Lx
BTC: 196h6xSrcjEPMf7jRQZbNBQcAB3twDMHE6

TBH I did not undertand what this is.. what would one gain using this docker?

[Eneen]

Any chances to see intensity option similar to ewbf? I've found dstm faster but this option is must-have for ppl working/gaming during mining.

[somethingiswrong]

DSTM I've got another question regarding the API of your miner.
The fields user and version are included in the response, but they are empty.
"user":     "username",           // username
"version":  "0.5.4",              // zm version

Are you aware of that or is this just an error in my case?

[NameTaken]

I created a Dockerfile for dstm 0.5.8.
I've found that using docker simplifies the management of my Ubuntu mining rig.
No more 'screen' or writing systemd/init scripts!

Requires https://github.com/NVIDIA/nvidia-docker

https://github.com/skinlayers/docker-dstm-zcash-cuda-miner


If you find this useful, feel free to toss some coin my way.

ETH: 0x75F4363b09166FDf8702F41502E90D21A3F2Afb9
ZEN: znY2eiUwdx9Pj4EzRkktZqRWHfEfX7o9gkW
FTC: 6nDBmQE65gy7BbcUYbYPttc9EYA7eNKN3g
ZCL: t1cSvJsEBWCipA4CY9K7oi9sKRQgb3EG76T
VIVO: VVmjio2zqGgfnvmkEU9bZcxXmFpNswusig
TZC: Tn6K9mw6bKc8R447CrpeEMazFcve8vEg1a
LBC: bXp34GVRLnQgTTid9x6nMorHYjq1GzFwz7
GBX: GTeMA2kehXf2E7DqPGEGVSsb1gQWxpg5Lx
BTC: 196h6xSrcjEPMf7jRQZbNBQcAB3twDMHE6

GPU environemt variables is only used by Claymore and sgminer. Use sha1sum from OP. Add BUILD_DEPENDENCIES and RUNTIME_DEPENDENCIES as ARG.

Code:

FROM nvidia/cuda:9.0-base-ubuntu16.04
LABEL maintainer="skinlayers@gmail.com"

ARG ARCHIVE_URL="https://docs.google.com/uc?export=download&id=19fSFYqoeOhOkxQqKnGpNI3n-7TIppHnq"
ARG ARCHIVE_SHA256="827b137e061fe1da8f8fbcea12cec22449d3ae13"
ARG BUILD_DEPENDENCIES="curl ca-certificates"
ARG RUNTIME_DEPENDENCIES="openssl"

COPY ./docker-entrypoint.sh /
RUN useradd -m miner && \
    apt-get update && \
    apt-get -y install --no-install-recommends $BUILD_DEPENDENCIES && \
    curl -L "$ARCHIVE_URL" | tar -xzC /usr/local/bin/ --strip 1 && \
    echo "$ARCHIVE_SHA256 /usr/local/bin/zm" > checksum && \
    sha1sum -c checksum && \
    apt-mark manual $RUNTIME_DEPENDENCIES && \
    apt-get remove --purge -y --auto-remove $BUILD_DEPENDENCIES $(apt-mark showauto) && \
    apt-get -y install $RUNTIME_DEPENDENCIES && \
    rm -r /var/lib/apt/lists/* && \
    chmod 0755 /docker-entrypoint.sh

USER miner
EXPOSE 2222
ENTRYPOINT ["/docker-entrypoint.sh"]
CMD ["zm"]

[Deathman20]

I tried my new 10x ASUS STRIX 1080ti OC cards today with 0.58 and got wonderful 740-760 Sol/s at 80% TDP@temperature 48-53 degrees celsius with each card Looking forward for 1080ti optimization!

80% TDP!  Dang I might have to look a little more at my card a little more if you can can do that.
Water cooled by chance or just really good air cooling?

[dstm]

DSTM, any updates soon ?

I skipped 0.5.8 because nothing good for me and i'm  waiting new version with something new in performance level.

I'm currently focusing on some restructuring and failover pool support.

Should we expect a slight increase in productivity due to restructuring? or reduce CPU load?

Could you look at CUDA 9.0, maybe it will be possible to slightly increase the performance?

Are there any ways to optimize for 1080 and 1080 Ti ?
Thank you very much for your work.

I will look again into performance of GDDR5X GPUs after releasing 0.6.

[dstm]

with reference to 12 GPU rigs:

i have noticed using DSTM on my 12 gpu rigs that GPU number 7 and 11 are on noticeably wamer than the others by 5-10degrees.

 I FIND THIS REALLY STRANGE.

Any suggestions?

Thanks in advance

This is nothing unusual, the temperature might differ even if you're using the same vendor/model. You could also check the GPU-utilization like it's reported by 'nvidia-smi' to make sure there are not software issues.

Thank you.
One more problem, miner reduces hash rates until almost zero but does not automatically reload or adjust accordingly as pool switches coin.  Not sure how to explain it. What can I do to resolve this. It only happens with DSTM and if I'm asleep it remains low for hours until I reload  manually.
Also some gpus drop to a far lower intensity than others but do not recover.
Happens on nicehash and miningpoolhub so I'm convinced it's the miner.
I'm fine with paying dev fees as long as I have good product support..

HELP SOMEONE.

I'm not sure what you're meaning by "as pool switches coin" and "intensity" - however this might help to resolve it.

Are you using some kind of software/script to launch zm? If so try to start zm without it.
The software might work not properly such that at some point you have multiple zm instances running using the same GPUs.

You can also check if your CPU load gets high at some point and what process is responsible for it.
What GPU-utilization does nvidia-smi roport after the drop?

[dstm]

Some clarification due to recent fee debate.

This is ofc something I was thinking about - speaking to my colleagues/friends, so called 'consultants' etc. Most of them suggest me to increase it especially on performance improvements. To make it short: I won't change it.

Claymore's Zcash mining software was stated a couple of times - the information wasn't always accurate so that's what Claymore states on his page:
"...current developer fee is 2% if you use secure SSL/TLS connection to mining pool ...If you use unsecure connection to mining pool, current developer fee is 2.5% ... If you don't agree with the dev fee - don't use this miner, or use "-nofee" option."
"-nofee   set "1" to cancel my developer fee at all. In this mode some recent optimizations are disabled so mining speed will be slower by about 5%."

I'm not sure if the statements about zm's accuracy were given by intention or by uncertainty - however it's important to me to clarify this: All numbers reported by zm including the developer fee are exact.

Keep up I believe lot's of us including me is bit jealous seeing that numbers you make We are just people. IDK what i would do being in your place, really...

Just dont abandon software and make it superior, there is some space t improve functionality/stability, and don't loose connection with us, users here.

BR

@dstm - could you please include Manufacturer info of the card, could be only in JSON, if you have mixed rig's it is pretty handy!

This is already on my todo list, It seems like nvidia doesn't provide this information on linux, I'll look closer into it later.

[dstm]

Any chances to see intensity option similar to ewbf? I've found dstm faster but this option is must-have for ppl working/gaming during mining.

Next release will have it.

[dstm]

DSTM I've got another question regarding the API of your miner.
The fields user and version are included in the response, but they are empty.
"user":     "username",           // username
"version":  "0.5.4",              // zm version

Are you aware of that or is this just an error in my case?

Thx for reporting, will check.

[aidonaks]

DSTM, any updates soon ?

I skipped 0.5.8 because nothing good for me and i'm  waiting new version with something new in performance level.

I'm currently focusing on some restructuring and failover pool support.

Should we expect a slight increase in productivity due to restructuring? or reduce CPU load?

Could you look at CUDA 9.0, maybe it will be possible to slightly increase the performance?

Are there any ways to optimize for 1080 and 1080 Ti ?
Thank you very much for your work.

I will look again into performance of GDDR5X GPUs after releasing 0.6.

@DSTM - I second this request. Many of us are running 1080's and 1080 Ti's and any possible improvements there would really help. Thanks!