How is going testing for Windows version?
If there are no major issues I'll start working on it next week.
Topic: [ANN] dstm's ZCash / Equihash Nvidia Miner v0.6.2 (Linux / Windows) - page 162. (Read 225062 times)
September 17, 2017, 04:35:10 PM
If there are no major issues I'll start working on it next week.
September 17, 2017, 04:12:39 PM
How is going testing for Windows version?
September 17, 2017, 02:58:21 PM
Im switching my ZEC rig to Linux this week (will have more than 8 cards for that rig)
Im glad I found this miner, thanks ypsi for the info.
Edit: Thanks to ypsi
Im glad I found this miner, thanks ypsi for the info.
Edit: Thanks to ypsi
September 17, 2017, 01:56:11 PM
New Version 0.4.3
stratum: support more pools
ui: flush stdout stream
con: improve reconnections
performance improvements 0.3-0.5 %
For people who used stdbuf to disable stream buffering - there is no need for it anymore.
stratum: support more pools
ui: flush stdout stream
con: improve reconnections
performance improvements 0.3-0.5 %
For people who used stdbuf to disable stream buffering - there is no need for it anymore.
September 16, 2017, 09:58:58 AM
miner does not work with luckpool.org, error message:
#protocol version 00000020 not supported
#protocol version 00000020 not supported
Done, it will work in the next release.
September 16, 2017, 04:34:08 AM
Feature request: It'd be nice if the miner can restart itself/reboot the machine (maybe make it like -r option in Claymore miner) when something goes wrong (GPU crashes, etc.)
Thx, suggestions are always welcome.
I don't think this is a good one. The miner should be stable, it's of high priority for me. If it isn't report it pls. Your hardware should also be stable. Doing restarts all the time wont help. Restarting of tasks should be done at the OS level for various reasons, that's how it's usually done. Does someone really put this option into his miner? Does it really reboot the system? That's crazy - you might end up in a reboot loop, it might damage your hardware.
September 15, 2017, 06:41:42 PM
Feature request: It'd be nice if the miner can restart itself/reboot the machine (maybe make it like -r option in Claymore miner) when something goes wrong (GPU crashes, etc.)
September 15, 2017, 05:08:19 AM
Just been mining a few hours with 0.4.2 and noticed it had stopped with a "msg buffer full" error.
I had been running 0.4.1 for longer and didn't have the error so not sure if it's something caused by the changes.
Rig is seven 1070s running on nanopool.
Performance for me is matching EWBF and seems a bit more consistent over time according to the stats from the pool.
I had been running 0.4.1 for longer and didn't have the error so not sure if it's something caused by the changes.
Rig is seven 1070s running on nanopool.
Performance for me is matching EWBF and seems a bit more consistent over time according to the stats from the pool.
Thx, will check.
September 14, 2017, 10:29:34 PM
Just been mining a few hours with 0.4.2 and noticed it had stopped with a "msg buffer full" error.
I had been running 0.4.1 for longer and didn't have the error so not sure if it's something caused by the changes.
Rig is seven 1070s running on nanopool.
Performance for me is matching EWBF and seems a bit more consistent over time according to the stats from the pool.
I had been running 0.4.1 for longer and didn't have the error so not sure if it's something caused by the changes.
Rig is seven 1070s running on nanopool.
Performance for me is matching EWBF and seems a bit more consistent over time according to the stats from the pool.
September 14, 2017, 05:05:43 PM
New Version 0.4.2
mp: more accurate time measurements
ui: calculate efficiency
Nvidia states that power usage readings have an accuracy to within +/- 5% of current power draw, be aware of this. I've balanced the sample rate, since reading the power usage too often consumes a significant amount of cpu time, I'm not sure if it also produces latencies on the PCIE-Bus. I'm using the average solution rate to calculate this value since that's the most accurate way to do it, be aware of this if you compare the values with other mining software. First values might be inaccurate since the solution rate is probabilistic, however it converges after some time, i.e. it gets more accurate the longer it runs.
LOL, so much text for a such small change
However that's how the things are.
mp: more accurate time measurements
ui: calculate efficiency
Nvidia states that power usage readings have an accuracy to within +/- 5% of current power draw, be aware of this. I've balanced the sample rate, since reading the power usage too often consumes a significant amount of cpu time, I'm not sure if it also produces latencies on the PCIE-Bus. I'm using the average solution rate to calculate this value since that's the most accurate way to do it, be aware of this if you compare the values with other mining software. First values might be inaccurate since the solution rate is probabilistic, however it converges after some time, i.e. it gets more accurate the longer it runs.
LOL, so much text for a such small change
However that's how the things are. September 14, 2017, 02:31:23 PM
Tried v0.4.1 on my 1080ti, cannot see significant performance improvement at 200 watts.
If I set the power limit to 250watts, the miner said "cudaMemcpy 1 failed" and hanged.
If I set the power limit to 250watts, the miner said "cudaMemcpy 1 failed" and hanged.
I checked the this part again. I'm not sure why it fails for you while doing PCIE data transfers. I'll include more debugging output into my next version. This should help to pin down the problem.
Btw. if you compare the numbers ShowMeCoins posted for version 0.4 and version 0.4.1 - he has an speedup of about 1% for 0.4.1 on his system, that's about what I expected.
September 14, 2017, 10:50:48 AM
Tried v0.4.1 on my 1080ti, cannot see significant performance improvement at 200 watts.
If I set the power limit to 250watts, the miner said "cudaMemcpy 1 failed" and hanged.
If I set the power limit to 250watts, the miner said "cudaMemcpy 1 failed" and hanged.
It's the second time someone has reported a failure on memcpy. It seemed it was due to a bad riser. I'll check this part again, thx.
September 14, 2017, 10:35:43 AM
Tried v0.4.1 on my 1080ti, cannot see significant performance improvement at 200 watts.
If I set the power limit to 250watts, the miner said "cudaMemcpy 1 failed" and hanged.
If I set the power limit to 250watts, the miner said "cudaMemcpy 1 failed" and hanged.
September 13, 2017, 06:12:24 PM
Thx, I'll take a look why they're using an old protocol version.
September 13, 2017, 06:00:23 PM
miner does not work with luckpool.org, error message:
#protocol version 00000020 not supported
#protocol version 00000020 not supported
September 13, 2017, 05:51:06 PM
1080ti : 2-3% faster than ewbf with same overclock settings
September 13, 2017, 03:30:08 PM
Windows version, please ....
If there are no major issues I'll start working on it next week.
September 13, 2017, 03:22:39 PM
Windows version, please ....
September 13, 2017, 03:19:14 PM
Quote
so what makes your mining program better than any others? program is still pretty lacking for a 2 percent dev fee.
I haven't done any comparisons beside performance. It seems to perform pretty well in comparison, however you should use whatever is more useful for you. Quote
how do we know your not going to do just like ewbf and grab the money and run and leave the miner up with a fee while still raking in the money
You can't, however I'm not planning to do this and I'll let you know ofc. if I won't have enough time for development in the future.Quote
why did you choose a 2 percent dev fee? just cause that's what everyone else is doing?
I haven't done any calculations, yes it's that simple.Quote
compared to claymores miners, not speed wise but as a whole, your program is way off from being worth 2 percent
I'm a unix guy. "Do one thing, and do it well".If you like fancy features you should probably use something else.
September 13, 2017, 02:28:14 PM
@dtsm so what makes your mining program better than any others? program is still pretty lacking for a 2 percent dev fee. how do we know your not going to do just like ewbf and grab the money and run and leave the miner up with a fee while still raking in the money. why did you choose a 2 percent dev fee? just cause that's what everyone else is doing? compared to claymores miners, not speed wise but as a whole, your program is way off from being worth 2 percent
Jump to: