hsrminer - Nvidia mining software for various algos by palgin&alexkap - page 9.

MagicSmoker

full member

Activity: 420

Merit: 184

Quote from: MagicSmoker on January 31, 2018, 04:52:55 PM

I just concluded the 24 hour test of ccminer klaust 8.19 and its effective hashrate came in at 3.68 MH/s vs. 4.06 MH/s reported, while hsrminer's effective hashrate was 3.30 MH/s vs. 4.32 MH/s reported. Since these results are so different and I only tested hsrminer for 4.5 hours I am going to give it yet another chance and do a full 24 hour test on the same pool once my existing balance fully clears. This way I can have more confidence that the average difficulty is truly average and eliminate as many variables as possible.

The results are in from my retest of hsminer over a full 24 hour period and its performance was even worse than before: it only earned 91.8 TZC instead of the 130.6 predicted based on its claimed hashrate of 4.3 MH/s and an average difficulty for 24 hours of 66, as reported by minethecoin.com. In other words, hsrminer earned 30% fewer coins because its effective hashrate was closer to 3 MH/s, instead of the 4.3 MH/s it was touting the entire time.

So not only is hsrminer not any faster than the KlausT fork of ccminer, it is actually significantly slower...

The rig used consists of (6) GTX 1060 3GB cards and the pool is the official one for Trezarcoin, pool.trezarcoin.com, and which finds a block every 4 minutes or less on average so a 24 hour test should keep error due to variation in luck below 1%. A 24 hour test also allows the use of the average difficulty reported by minethecoin.com which keeps error due to swings in difficulty to a minimum.

I am definitely curious to see if anyone else can corroborate or refute my results, especially with a different coin and/or pool, but please make sure to follow the same methodology I used so we are all comparing apples to apples here.

EDIT - I am not changing any of the above, but notifying any who come across this that I did another test of ccminer and hsrminer, except run concurrently mining to different addresses on the same pool, and this time hsrminer was 11% faster (though claiming to be 22% faster).

jugger1028

jr. member

Activity: 213

Merit: 3

Quote from: MagicSmoker on February 01, 2018, 07:56:30 AM

Quote from: jp1981 on February 01, 2018, 12:17:55 AM

So I have been reading up on this thread and saw that a lot of people say they are not getting the same hash rates at the pool as what the local miner reports. This has had me worried. I run a small farm and I directed all my cards to use the hsrminer.

I have taken one card out of the batch and created a new Nicehash wallet, to see exactly what hsrminer does.

1. On my side the miner disconnects every hour and mines for about 30-60 seconds to the devs wallet.
2. Below the hash rate reported by the miner. This is a evga 1070 running at 70 power target - 137 core and 495 mem. The miner reports an avg of 1088 kH/s

I don't think this test is trustworthy, not to say completely useless. Poolside average hashrate - such as can be determined from the pool I was using, as it does not provide a simple moving average graph - was roughly the same as what hsrminer reported, but the actual amount of TZC that I earned was substantially less than predicted by minethecoin.com using hsrminer's reported hashrate and the average difficulty for the last 24 hours. I am retesting hsrminer on the same pool now - results available in about 12 hours - but so far at just a little over the halfway mark I have earned 44.5 TZC vs. a predicted earnings of 63.5, or about 30% less, which is even worse than my previous test of hsrminer (24% less than predicted).

EDIT - I should note that this is on the official TZC pool which finds a block every 4 minutes on average.

You will never get the coins whattomine says because the difficulty changes literally every block. The only way to measure which miner you should be using is mining for an extended period of time and taking a look at both the shares submitted at the pool and the hashrate reported at the pool. You can never look at the coins earned because of the difficulty variable and you can never say I made 50 coins today with CCMiner and 75coins the day before with HSRMiner, because that tells you nothing since the difficulty determines coins earned.

I just finished a 7 hour test with CCMiner mining TZC and confirmed the hashrate and shares submitted are lower than Hsrminer. The miner reports about 100-200 kh/s less than hsrminer per card on the miner itself and that's the result I was seeing in shares submitted and hashrate on the pool itself. I was skeptical, but after this test I can't really dispute the reported rates. I will say my hashrate with one card is higher than when I have a bunch of them running. I have this same problem with all miners though..

MagicSmoker

full member

Activity: 420

Merit: 184

Quote from: jp1981 on February 01, 2018, 12:17:55 AM

So I have been reading up on this thread and saw that a lot of people say they are not getting the same hash rates at the pool as what the local miner reports. This has had me worried. I run a small farm and I directed all my cards to use the hsrminer.

I have taken one card out of the batch and created a new Nicehash wallet, to see exactly what hsrminer does.

1. On my side the miner disconnects every hour and mines for about 30-60 seconds to the devs wallet.
2. Below the hash rate reported by the miner. This is a evga 1070 running at 70 power target - 137 core and 495 mem. The miner reports an avg of 1088 kH/s

I don't think this test is trustworthy, not to say completely useless. Poolside average hashrate - such as can be determined from the pool I was using, as it does not provide a simple moving average graph - was roughly the same as what hsrminer reported, but the actual amount of TZC that I earned was substantially less than predicted by minethecoin.com using hsrminer's reported hashrate and the average difficulty for the last 24 hours. I am retesting hsrminer on the same pool now - results available in about 12 hours - but so far at just a little over the halfway mark I have earned 44.5 TZC vs. a predicted earnings of 63.5, or about 30% less, which is even worse than my previous test of hsrminer (24% less than predicted).

EDIT - I should note that this is on the official TZC pool which finds a block every 4 minutes on average.

Dipakdhawani

newbie

Activity: 4

Merit: 0

where can I check my CC version, and what does it means anyways? I am a noob
I have 6 asus dual OC gtx 1070 8GB and 6 giga byte windforce gtx 1070 8gb, will this miner work on this cards?

and can someone explain me this
Example:
I have 6x1080 rig, on each kernel allocates 6260 Mb memory, so sum of RAM + pagefile should be more than 6260*6 = 37560.

Keep that in mind when launching on multi-GPU systems.

I have my virtual ram set to (6(cards)x8(gb)x1024(mb)) 49152.
is this correct.
currently running klaust miner.

thank you

jp1981

newbie

Activity: 34

Merit: 0

So I have been reading up on this thread and saw that a lot of people say they are not getting the same hash rates at the pool as what the local miner reports. This has had me worried. I run a small farm and I directed all my cards to use the hsrminer.

I have taken one card out of the batch and created a new Nicehash wallet, to see exactly what hsrminer does.

1. On my side the miner disconnects every hour and mines for about 30-60 seconds to the devs wallet.
2. Below the hash rate reported by the miner. This is a evga 1070 running at 70 power target - 137 core and 495 mem. The miner reports an avg of 1088 kH/s

https://image.prntscr.com/image/isMeUGeCQ3iM0l7v8WwNRw.png

3. Below the stats from Nicehash

https://image.prntscr.com/image/Phlslh5LRfOd97ShUVQpEQ.png

So to me it looks like it is submitting what it is reporting locally, actually slightly more.

Hope it helps

MagicSmoker

full member

Activity: 420

Merit: 184

Quote from: MagicSmoker on January 30, 2018, 03:25:50 PM

Quote from: MagicSmoker on January 29, 2018, 04:11:39 PM

...
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.

I just concluded the 24 hour test of ccminer klaust 8.19 and its effective hashrate came in at 3.68 MH/s vs. 4.06 MH/s reported, while hsrminer's effective hashrate was 3.30 MH/s vs. 4.32 MH/s reported. Since these results are so different and I only tested hsrminer for 4.5 hours I am going to give it yet another chance and do a full 24 hour test on the same pool once my existing balance fully clears. This way I can have more confidence that the average difficulty is truly average and eliminate as many variables as possible.

Elange

newbie

Activity: 73

Merit: 0

I'm really eager to test out your miner, but I run Linux rigs. Are you planning to release linux binaries?

I can betatest, if needed. I have 1060s, 1050s, 980s, 970s.

Thanks!

Mr.Spider703

member

Activity: 392

Merit: 27

http://radio.r41.ru

Quote from: badfad on January 31, 2018, 04:31:17 AM

I'm having a problem with the miner closing when I use the cpu to mine other coins. 2 x 1080 + 1070ti otherwise running very well.
Any ideeas?

overclocking reduce for 1080 = PL 70-80 mem 150-250 core 100-150, for 1070ti i don't know, i havn't this card.
for 1070ti try at 80% tdp. +135 +700
Or from the overclocking which is now done for -50 for the memory and -10 the core. Run and see how it will work, if we leave it steadily, if it takes off, then we reduce it, you can choose the thresholds yourself, I chose in step of 20

Quote from: jdubya on January 31, 2018, 04:30:21 AM

Are there any updates coming soon for this miner?

Thanks

waiting Palgin

badfad

jr. member

Activity: 186

Merit: 4

I'm having a problem with the miner closing when I use the cpu to mine other coins. 2 x 1080 + 1070ti otherwise running very well.
Any ideeas?

jdubya

newbie

Activity: 24

Merit: 0

Are there any updates coming soon for this miner?

Thanks

rednoW

legendary

Activity: 1510

Merit: 1003

Quote from: Mr.Spider703 on January 31, 2018, 02:47:27 AM

Quote from: skablast on January 31, 2018, 02:44:33 AM

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

Mr.Spider703

member

Activity: 392

Merit: 27

http://radio.r41.ru

Quote from: skablast on January 31, 2018, 02:44:33 AM

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)

skablast

member

Activity: 476

Merit: 19

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

member

Activity: 392

Merit: 27

http://radio.r41.ru

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

GregBitMineMay

newbie

Activity: 3

Merit: 0

Quote from: niehues313 on January 30, 2018, 07:28:19 PM

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!

niehues313

newbie

Activity: 54

Merit: 0

Quote from: Mr.Spider703 on January 30, 2018, 07:57:53 PM

Quote from: niehues313 on January 30, 2018, 07:56:17 PM

Quote from: Mr.Spider703 on January 30, 2018, 07:47:17 PM

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 !

Mr.Spider703

member

Activity: 392

Merit: 27

http://radio.r41.ru

Quote from: niehues313 on January 30, 2018, 07:56:17 PM

Quote from: Mr.Spider703 on January 30, 2018, 07:47:17 PM

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

newbie

Activity: 54

Merit: 0

Quote from: Mr.Spider703 on January 30, 2018, 07:47:17 PM

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

member

Activity: 392

Merit: 27

http://radio.r41.ru

niehues313 try -i 20

niehues313

newbie

Activity: 54

Merit: 0

Quote from: Mr.Spider703 on January 30, 2018, 07:34:13 PM

Quote from: niehues313 on January 30, 2018, 07:28:19 PM

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?

Quote from: EVERYUSERNAMEISTAKEN on January 30, 2018, 07:14:12 PM

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.

Topic: hsrminer - Nvidia mining software for various algos by palgin&alexkap - page 9. (Read 30804 times)