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

[ioglnx]

So in your diagram the legend is missing what is blue what is red..should we guess out of the blue?

..mh

[1715199921]

1715199921

[1715199921]

1715199921

[1715199921]

1715199921

[1715199921]

1715199921

[zorday]

So in your diagram the legend is missing what is blue what is red..should we guess out of the blue?

..mh

yes you can guess or you can read the post

Anyway I fixed legends now in the original post.

or here:
https://ibb.co/cSv9im

[jdubya]

How do I run the miner using only GPUs 10 and higher? -d 10 runs GPU 1


thanks

[EVERYUSERNAMEISTAKEN]

...

Thank you for the testing, always good to have another data point. However, I think 200 minutes is far too short. 12hrs would be much more appropriate, or better yet, 24hrs.

[proteus7]

Still drives me nuts that TitanXP won't work, as there is absolutely no logical reason for it...its just an uncrippled 1080ti. Is there some kind of weird hard coding? every other miner works...but this one.

[malthrax]

Still drives me nuts that TitanXP won't work, as there is absolutely no logical reason for it...its just an uncrippled 1080ti. Is there some kind of weird hard coding? every other miner works...but this one.

Just give it up.  It doesn't work, and the author has abandoned this forum.  Its dead.

[airwalker_]

I believe this is based on Klaust's ccminer fork and there was an issue with Titan XP that he recently fixed. Unfortunately, palgin has been MIA and has not incorporated the fix into hsrminer. There is another member that has forked off of hsrminer and added functionality - maybe you can request it from him: https://bitcointalk.org/index.php?topic=2765610.0

[Gongolo]

Still drives me nuts that TitanXP won't work, as there is absolutely no logical reason for it...its just an uncrippled 1080ti. Is there some kind of weird hard coding? every other miner works...but this one.

Just give it up.  It doesn't work, and the author has abandoned this forum.  Its dead.

Dev is busy for personal reasons, he will be back asap.
Please, some respect.

[MagicSmoker]

There's nothing more difficult in this world than convincing someone of an unfamiliar truth...

I'm very easy to convince if your argument has substance and is based in fact, but so far the only arguments I've seen are merely statements that I am wrong or, even less persuasive, that I am an idiot. Since I've designed entire drive systems for locomotives for my day job I'm unlikely to be convinced I'm an idiot just because some random dude on the internet says so.

Again, I am more than happy to change my testing methodology if a reasonable explanation can be given as to how it could be so flawed it would result in a 30% difference in coins earned on two consecutive days based on reported hashrate and average difficulty on minethecoin.com (not a snapshot of earnings from whattomine.com).
...

Someone took the time to explain what I was missing - that luck varies on a per-share basis, not just on a per-block basis. I still find it hard to believe that luck on a per-share basis could cause one miner to make 30% less than another over a 24 hour period when the pool finds a block every 2-4 minutes, but I can see it is at least possible.

Also, I tried running ccminer-klaust with a fixed difficulty of 512 in preparation for a retest and surprise-surprise, I only made 63% of the expected number of coins based on average difficulty... So that kind of cinched it for me right there.

I'm going to restest the two miners concurrently using the TZC pool on altminer.net as it doesn't require registration, an issue which has prevented me from registering a second payment miner/payment address on the TZC official pool.

[tritonchev89]

UPDATE: made fix for high CPU load, now it's barely loading CPU, added ccminer-style intensity option to play with, but work distribution bug is still there, so work continues...

Please, re-download binary from github, I've uploaded CPU-fix + intensity feature version. Tested intensities are 17.5 for 1080/1070, 16 for 1060 3G.

Thank you!

i have 1060 6gb windforce and should put PL to 100% and then my core clock working. What is the problem. With another miner this does not happen. When i switch to lower PL the core clock goes down. What i Shoud to do to fix it.

[tritonchev89]

GPU#0 - MSI GTX 1060 3GB, Temp is 69C, Fans at 52%, Power consumption 128W

I think we need temp limit on hsrminer - neoscript


I see so big diffirence in power consumption. 124W - 144W
no oc - it is at stock settings.

GTX1060 3GB power needs is 120W.
Why your miner needs more power ?

This is the same with my Gigabyte GTX 1060 6GB, what is the problem?

[jugger1028]

@Zorday , great job on the testing but yes, a much larger testing window would be better. If a baseball batter got up to bat 3times and got 2 hits you'd think he was the greatest baseball player in history by looking at his average. As the batter got more at bats the natural average would kick in and you'd see he's only batting .230 and now we're dumping him back into the minor leagues A slightly higher hashrate with higher shares submitted should produce better results over the long haul but because the increase is so minor, we're talking 5-8% at most, a much longer time period would be best to flesh out this very minor increase.

Definitely on the right path though for testing the two and I appreciate the info.

[zorday]

@Zorday , great job on the testing but yes, a much larger testing window would be better. If a baseball batter got up to bat 3times and got 2 hits you'd think he was the greatest baseball player in history by looking at his average. As the batter got more at bats the natural average would kick in and you'd see he's only batting .230 and now we're dumping him back into the minor leagues A slightly higher hashrate with higher shares submitted should produce better results over the long haul but because the increase is so minor, we're talking 5-8% at most, a much longer time period would be best to flesh out this very minor increase.

Definitely on the right path though for testing the two and I appreciate the info.

Agreed and I will do more tests... I have to restart though as one of the miners got disconnected yesterday and I was not paying attention.
Yes I am well aware that the stats are a bit scarce, but still I would expect to see a slightly different angle on the slope. Isn't the difference more like 20%?
1480kH/s compared to 1790kH/s reported?

[MagicSmoker]

The results are in from the concurrently run comparison between hsrminer and ccminer KlausT v8.19 on the TZC pool at Altminer.net and this time hsrminer won. Average hashrate reported by ccminer was 1010 kH/s while hsrminer claimed 1230 kH/s, or about 22% more. The difference in earnings, however, were not quite as different, with ccminer earning 24.497 TZC while hsrminer brought in 27.264 TZC, or about 11% more.

So my results are similar, but not identical, to what @zorday found. Hsrminer is faster, but it also lies about just how fast it is.

I should also note that hsrminer is much more difficult to stabilize, especially with regards to GPU core overclocking, usually manifesting in a mysterious drop in hashrate of 10-20% after 1-3 hours of continuous operation. That did not occur during this test because I've dialed in the -i and -c settings on each my rigs, but it took a lot of iterations to get there.

[ElMinator]

What W10 build / NVIDIA drivers combo are you using to run this miner?. I keep getting error 43 on some cards

[jugger1028]

@Zorday , great job on the testing but yes, a much larger testing window would be better. If a baseball batter got up to bat 3times and got 2 hits you'd think he was the greatest baseball player in history by looking at his average. As the batter got more at bats the natural average would kick in and you'd see he's only batting .230 and now we're dumping him back into the minor leagues A slightly higher hashrate with higher shares submitted should produce better results over the long haul but because the increase is so minor, we're talking 5-8% at most, a much longer time period would be best to flesh out this very minor increase.

Definitely on the right path though for testing the two and I appreciate the info.

Agreed and I will do more tests... I have to restart though as one of the miners got disconnected yesterday and I was not paying attention.
Yes I am well aware that the stats are a bit scarce, but still I would expect to see a slightly different angle on the slope. Isn't the difference more like 20%?
1480kH/s compared to 1790kH/s reported?

Isn't really 1790 with a rig.. On single cards, yeah, I can get above 1800kh/s with 90%tdp but in a group of cards my range is 1650-1750kh/s. I'd guess it was closer to 5-8% hashrate gains, no where near 20%.

[Gridslack]

Пpи зaпycкe пишeт, чтo нeт дpaйвepoв пoд CUDA.
Кaкиe нyжны вepcии дpaйвepoв?

Bидeoкapты - 1060 3Гб
Oпepaциoнкa - Win 8.1
Дpaйвepa - 22.21.13.8233 oт 17.05.2017
Фaйл пoдкaчки - 50Гб

[ekze]

Пpи зaпycкe пишeт, чтo нeт дpaйвepoв пoд CUDA.
Кaкиe нyжны вepcии дpaйвepoв?

Bидeoкapты - 1060 3Гб
Oпepaциoнкa - Win 8.1
Дpaйвepa - 22.21.13.8233 oт 17.05.2017
Фaйл пoдкaчки - 50Гб

388.71
http://www.nvidia.co.uk/Download/Find.aspx?lang=en-uk

[Ayman84]

for two days testing the miner with GTX 1070
i found its giving fake hashrate number
i get 1300 for neoscrypt with my GTX 1070
CCminer Klaust  1250
but still the profit from nicehash if i use ccminer higher then hsrminer
i test one card for one hour with both software and the winner is ccminer Klaust
three hour also ccminer
two card for one hour and three hours also ccminer better
anyone agree with me ?
come on its close source software

[navarthelol]

So, I did the test I mentioned earlier in the thread.

I created a QBIC wallet and picked up two different send addresses. I picked QBIC as it would generate a fair number of blocks per hour based on the history for that coin.

I have two 1080Ti cards of the same brand/spec (Gigabyte). They are exactly the same.
Both are overclocked with the same settings: +100/+400 80%

I created two batch files one using hsr_miner and one of the addresses, and one using CC-miner-Klaust and the other address. Both were directed at the same port in BSOD pool.

I started both batch jobs at the same time and let it run for 80 minutes roughly then paused for the night and carried on for another 120 minutes this morning.
Both miners behaved well although KlausT had two rejections (99.22% efficiency). There was one pool disconnect (that I noticed) but both miners experienced this and it was only during 10 seconds

The reported hash rates from each miner were :

KlausT :       1480 kH/s
hsrminer:     1790 kH/s


During these periods 58 QBIC blocks were found, and the results in earned coins were:

KlausT:       0.07451398  QBIC
hsrminer:     0.07188344 QBIC


One interesting fact is that the pool for some reason reported a somewhat higher hashrate from hsrminer during the tests, but the results were more or less at par.
I pulled out my google docs skills (quite limited ) and created the graph below.
The red line is KlausT and the blue line is hsrminer.



While luck certainly is a factor in these matters it seems as if the promised edge just isn't there, but perhaps a longer test is needed.
Feel free to post any objections to this test and how it can be improved.

I'm sorry but there is a VITAL piece of information you are missing. QBIC is completely unreliable source to try to benchmark against because it is always under nicehash attacks. Hashrates can go from 1ghs on a pool to 20/30 even 160 I have seen.