hsrminer - Nvidia mining software for various algos by palgin&alexkap

[MagicSmoker]

Since you are very difficult, I have made this!


...
No need further explanation ? ^_^

En Francais! Sorry, no cedilla on the US keyboard. Also, my French is really, really rusty.

It looks like what is holding me back is not lowering the core voltage. I've been trying to convince MSI AB to let me do that but haven't had success yet...

[katak]

Since you are very difficult, I have made this!


...
No need further explanation ? ^_^

En Francais! Sorry, no cedilla on the US keyboard. Also, my French is really, really rusty.

It looks like what is holding me back is not lowering the core voltage. I've been trying to convince MSI AB to let me do that but haven't had success yet...

Si je commence à parler en Français, je crains que certains me sautent dessus!   

[g01dt00th]

When GTX Titan Xp support will be added?

[MagicSmoker]

Si je commence à parler en Français, je crains que certains me sautent dessus!  

If you speak French you fear some will jump over you?!?  

I suspect this an idiom that did not survive my poor translation skills (I suppose I could plug the sentence into Google Translate, but where's the fun in that?)

On a more relevant note, I changed skins in MSI AB - I was using the old v3 skin  - and now I see that my core voltage occasionally drops even lower than yours at times so I am guessing you didn't manually tweak it lower. If my guess is incorrect then please tell me how to lower core voltage because it seems all I am able to do is raise it.

[MagicSmoker]

...
I just concluded another test of hsrminer on the Trezarcoin official pool and so far the results don't look good but I need all the blocks to confirm before I bury the hatchet on this one. Let's just say that with average difficulty over the last 24 hours (which would be conservative, because current difficulty spent most of the day below that) I should have earned 38 TZC but so far the sum of confirmed and unconfirmed is around 29.

Difficulty on ZEN has been really high today so it's as good a time as any to have the 6x GTX 1060 rig do another neoscrypt mining test. The results reported in the quote above were with hsrminer claiming it was hashing at 4.32 MH/s. The KlausT fork of ccminer (v8.19) is reporting a hashrate of 4.06 MH/s after 10 minutes of operation (it does take awhile for neoscrypt to get up to speed for some reason), so about 6.4% less than what hsrminer was claiming. However, I only got credited for 29 coins instead of the 38 that whattomine said I should have received given the hashrate and average difficulty, which comes out to a 24% haircut... If I get screwed by the same percentage with ccminer then either both are lying at exactly the same magnitude or else I can't trust whattomine and/or the pool.

[katak]

Si je commence à parler en Français, je crains que certains me sautent dessus!  

If you speak French you fear some will jump over you?!?  

I suspect this an idiom that did not survive my poor translation skills (I suppose I could plug the sentence into Google Translate, but where's the fun in that?)

On a more relevant note, I changed skins in MSI AB - I was using the old v3 skin  - and now I see that my core voltage occasionally drops even lower than yours at times so I am guessing you didn't manually tweak it lower. If my guess is incorrect then please tell me how to lower core voltage because it seems all I am able to do is raise it.

Hey you do get skills here! ;-)
It's just that I have seen so many people shouting at other ones when they did not speak english but still I get the feeling...

Anyway I did not tweak core (100% in both case), I think it's due to activity handling from miner, as you can see on my snapshot, Hsrminer is less steady than ccminer, certianly it makes drop on core speed.

I also mine ZEN, on another rig... here it's for GBX, imagining that it might pump someday, Nvidia is better actually on Neoscrypt than Equihash.

[MagicSmoker]

Hey you do get skills here! ;-)
It's just that I have seen so many people shouting at other ones when they did not speak english but still I get the feeling...

Anyway I did not tweak core (100% in both case), I think it's due to activity handling from miner, as you can see on my snapshot, Hsrminer is less steady than ccminer, certianly it makes drop on core speed.

I also mine ZEN, on another rig... here it's for GBX, imagining that it might pump someday, Nvidia is better actually on Neoscrypt than Equihash.

I think most of the shouting at damn foreigners is for the Russians that write in Cyrillic!

I am still puzzled by how well hsrminer works for you and how poorly it works for me. First you get 1.38 MH/s while I only got 1.23 MH/s - that's 12% more already - and then my actual earnings at the pool were only 76.3% of what I should have made based on average difficulty for the time period.

My test with ccminer-klaust has only been running for 1.75 hours but earnings are pretty much exactly as predicted by whattomine so far.

[EVERYUSERNAMEISTAKEN]

https://i.imgur.com/nJWAbno.jpg

~11hr (~6.75h to ~16.25h on the graph) test for hsrminer on Nemosminer.

Ignore the rollercoaster at the 5h mark, I was playing around with settings. The drop at 17h is a result of a 30min internet outage, this screws up the following moving average.

As you might be able to see, the relevant moving average is about 2820-2880kh. I'd say it's closer to a 2850-2860kh average if you hover above the data points. This is server-side on AHP24h (pure Neoscrypt). Client-side, I'm averaging 2930-2950kh.

So where'd the extra 100kh go? Take into account a 1% fee and that leaves us at 2900-2920kh. From here we're still missing a bit. I'm going to assume AHP takes the 1% from your actual profits, and doesn't just cut your hash rate accordingly. The only logical explanation is my 99% efficiency, which would take us to 2871-2890kh. Still missing 10-20kh, but in the grand scheme of things, that's pretty negligible. I've seen a lot of people claim hsrminer doesn't deliver the promised client-side numbers, but my experience differs. I'm not trying to discredit any claims, but I would encourage everyone to try it and see whether it works for them.

Stats in the Powershell window are fairly accurate, may be the 1070Ti should be closer to ~142w.

Specifics:

1070Ti @ 85% TDP, 80c target, +200 core, +400 memory (may be I should clock this down too), ~58c @ 65% fan
1080Ti @ 85% TDP, 80c target, +190 core, +100 memory (+300 mem didn't really help for me in Neo), ~52c @ 85% fan

Ambient temp is ~21-22c.

All in all, I'm much more concerned about my Blake2s hashrates being quite different from what I'm getting client-side.

[niehues313]

Hello.

Any ideia what this error can be:

Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1460 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1460 : the launch timed out and was terminated.

Closed program.

Thanks !

[Mr.Spider703]

Hello.

Any ideia what this error can be:

Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1460 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1460 : the launch timed out and was terminated.

Closed program.

Thanks !

overclock what is it worth? PL?

~11hr (~6.75h to ~16.25h on the graph) test for hsrminer on Nemosminer.

Ignore the rollercoaster at the 5h mark, I was playing around with settings. The drop at 17h is a result of a 30min internet outage, this screws up the following moving average.

As you might be able to see, the relevant moving average is about 2820-2880kh. I'd say it's closer to a 2850-2860kh average if you hover above the data points. This is server-side on AHP24h (pure Neoscrypt). Client-side, I'm averaging 2930-2950kh.

So where'd the extra 100kh go? Take into account a 1% fee and that leaves us at 2900-2920kh. From here we're still missing a bit. I'm going to assume AHP takes the 1% from your actual profits, and doesn't just cut your hash rate accordingly. The only logical explanation is my 99% efficiency, which would take us to 2871-2890kh. Still missing 10-20kh, but in the grand scheme of things, that's pretty negligible. I've seen a lot of people claim hsrminer doesn't deliver the promised client-side numbers, but my experience differs. I'm not trying to discredit any claims, but I would encourage everyone to try it and see whether it works for them.

Stats in the Powershell window are fairly accurate, may be the 1070Ti should be closer to ~142w.

Specifics:

1070Ti @ 85% TDP, 80c target, +200 core, +400 memory (may be I should clock this down too), ~58c @ 65% fan
1080Ti @ 85% TDP, 80c target, +190 core, +100 memory (+300 mem didn't really help for me in Neo), ~52c @ 85% fan

Ambient temp is ~21-22c.

All in all, I'm much more concerned about my Blake2s hashrates being quite different from what I'm getting client-side.

in blake2s 12 Gh/s on 3*1060 (PL 75, mem 550, core 150)  & 1*1080 (PL 75, mem 0, core 0) 10 min working and closed

[niehues313]

Hello.

Any ideia what this error can be:

Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1460 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1460 : the launch timed out and was terminated.

Closed program.

Thanks !

overclock what is it worth? PL?

https://i.imgur.com/nJWAbno.jpg

~11hr (~6.75h to ~16.25h on the graph) test for hsrminer on Nemosminer.

Ignore the rollercoaster at the 5h mark, I was playing around with settings. The drop at 17h is a result of a 30min internet outage, this screws up the following moving average.

As you might be able to see, the relevant moving average is about 2820-2880kh. I'd say it's closer to a 2850-2860kh average if you hover above the data points. This is server-side on AHP24h (pure Neoscrypt). Client-side, I'm averaging 2930-2950kh.

So where'd the extra 100kh go? Take into account a 1% fee and that leaves us at 2900-2920kh. From here we're still missing a bit. I'm going to assume AHP takes the 1% from your actual profits, and doesn't just cut your hash rate accordingly. The only logical explanation is my 99% efficiency, which would take us to 2871-2890kh. Still missing 10-20kh, but in the grand scheme of things, that's pretty negligible. I've seen a lot of people claim hsrminer doesn't deliver the promised client-side numbers, but my experience differs. I'm not trying to discredit any claims, but I would encourage everyone to try it and see whether it works for them.

Stats in the Powershell window are fairly accurate, may be the 1070Ti should be closer to ~142w.

Specifics:

1070Ti @ 85% TDP, 80c target, +200 core, +400 memory (may be I should clock this down too), ~58c @ 65% fan
1080Ti @ 85% TDP, 80c target, +190 core, +100 memory (+300 mem didn't really help for me in Neo), ~52c @ 85% fan

Ambient temp is ~21-22c.

All in all, I'm much more concerned about my Blake2s hashrates being quite different from what I'm getting client-side.

in blake2s 12 Gh/s on 3*1060 (PL 75, mem 550, core 150)  & 1*1080 (PL 75, mem 0, core 0) 10 min working and closed

I was doing 100% PL, 0 overclock, 90% fan speed.
Now i'm trying 90% PL, 0 overclock, 90% fan speed.

It ran for about an hour before the reported crash.

[Mr.Spider703]

niehues313 try -i 20

[niehues313]

niehues313 try -i 20

You mean such as example: -p x -i 20 ?

What would this do exactly? Intensity 20?

Sorry if dumb question.

Thanks alot.

[Mr.Spider703]

niehues313 try -i 20

You mean such as example: -p x -i 20 ?

What would this do exactly? Intensity 20?

Sorry if dumb question.

Thanks alot.

yes it's intensity "-p x -i 20"

[niehues313]

niehues313 try -i 20

You mean such as example: -p x -i 20 ?

What would this do exactly? Intensity 20?

Sorry if dumb question.

Thanks alot.

yes it's intensity "-p x -i 20"

Ok will try it if error appears again, thanks man !

[GregBitMineMay]

Hello.

Any ideia what this error can be:

Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1460 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1464 : the launch timed out and was terminated.
Cuda error in func 'neoscrypt_cpu_hash_1070' at line 1460 : the launch timed out and was terminated.

Closed program.

Thanks !

I have been hitting this issue several times.   It shows up with 4 or more 1070's in my rig.   If I only mine in 1 HSR Miner process with 3 devices (1070 gpus), then it doesnt seem to happen.   So, I have been trying two HSR Miner instances with 3 cards each instance, so far thats been working.   But I am lazy so I run one CC miner with 4 cards and one HSR miner with 3 1070's and its working ok.   I have my Virtual Memory up to 68 GB for 7 1070's ... so I dont think thats the issue, but could be wrong.    My rig only has 4 GB, so may a memory issue, but not sure.

When I account for the increased rejected shares and the ramp up time every time the HSR miner switches to the Dev pool, then I think its not too much better than CC Miner.   But, HSR miner does some magic with GPU 0 that CCminer cant do (for some reason GPU0 is about 50-75% slower on CC Miner).   So, I will keep it mining away!

[Mr.Spider703]

whether it is used in the mining, try to disable mining on the CPU if used

[skablast]

Hi, I have a mixed rig with 4 GTX 1070 and 4 Rx 580 currently mining with claymore dual

Its possible to have two mining software one for AMD and this one for NVDA, mining at the same time neoscrypt ?

Anyone have tried ?

 

[Mr.Spider703]

Hi, I have a mixed rig with 4 GTX 1070 and 4 Rx 580 currently mining with claymore dual

Its possible to have two mining software one for AMD and this one for NVDA, mining at the same time neoscrypt ?

Anyone have tried ?

 

make two bat files
first for amd
second for nvidia -d 3,4,5 (3,4,5 - name of card in system)

[rednoW]

Hi, I have a mixed rig with 4 GTX 1070 and 4 Rx 580 currently mining with claymore dual

Its possible to have two mining software one for AMD and this one for NVDA, mining at the same time neoscrypt ?

Anyone have tried ?

 

make two bat files
first for amd
second for nvidia -d 3,4,5 (3,4,5 - name of card in system)

No problem at all. Choose proper gpus in claymore for eth (only amd). Use hsrminer for neoscrypt with default number of gpus - it will use all cuda supported (nvidia) cards