Topic: hsrminer - Nvidia mining software for various algos by palgin&alexkap - page 7. (Read 30774 times)

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.

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.

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

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://bitcointalksearch.org/topic/hsrminer-neoscrypt-fork-by-justaminer-high-hashrate-api-all-gpus-supported-2765610

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

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.

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.

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


thanks

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

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

..mh

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.

https://ibb.co/cSv9im

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 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).

That said, the official Trezarcoin pool does allow mining with a fixed difficulty but adding a second account from the same IP address appears to be blocked by default - I've already sent an email to the support address saying I want to add a second account so I can test miners concurrently but no response so far. Thus I can at least try jugger1028's suggestion to count the number of shares earned with a fixed difficulty for x period of time with hsrminer, then compare that by mining with ccminer for the same period of time/same difficulty; the length of time wasn't specified, but based on the average time to find a block on pool.trezarcoin.com it seems that 12 hours per miner should be sufficient.

So, if that is acceptable I will proceed with the test, and if the results contradict my earlier findings I will edit my previous posts to warn that I got different results with a different test methodology because I don't have a particular axe to grind here.

I noticed that on the front page there are three different links to HSR miner (neoscrypy).

There's a link to the Win version, the ZIP and to the test version. Both the Win version and the test version link to a file last changed on 30 Dec and is stable for me on 1050Ti. The link to the ZIP file contains an older .EXE from Dec 25th and is not stable for me.

Maybe this helps other people having trouble with HSR miner. Check if your .exe is the most recent one with file size 7.065.600 bytes.

Further I found that there a speed differences between different GPU models. Therefore I think you can't easily say if a speed score is high or low, it depends also on your GPU model and OC settings.

Here are my scores:
- Gigabyte 1050Ti 4GB OC (490kH/s withour overclocking and 530 kH/s using +144 core clock +170 memory clock, power limit 100)
- MSI 1050Ti 4GB Gaming X (490kH/s withour overclocking and 566 kH/s using +144 core clock +170 memory clock, power limit 100)

I'm mining to Nicehash and with the default software I get around 380 kH/s. When mining with HSR miner I see much higher speed and also much higher earnings so I'm very happy with this miner.

I want linux to ,but can add keyblind like h= show actual hashrate ?

Any news on linux version?

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

Thank you for your answer, my cards are rock stable with this speeds (i'm quiet lucky)

So I've only changed the power limit from 70% to 90% and I can obtain 1800 khs...

Now I have to check the power consumption to find the best balance but i'm not at home.

Don't set anything for intensity, hsrminer does it itself.. Increase your virtual memory above 11gb per card. your overclock seems pretty high, I can't get anything above +100 core stable on HSRminer and +600 mem... I can go +150/+625 on my evga hybrids. my TDP is 90%.. I can get near 1800kh/s on some cards, 1700 on others.. hsrminer is not as stable as the other miners when you have higher overclock, so lower that plus increase the virtual memory by 35gb for those 3 cards and see how that works for you.

I use 3 x 1080 ti and I can't specify an intensity higher than 7, if i put more the program crash.

If I change the intensity (5, 6, 7 or nothing) the performance doesn't change and the memory usage as well (only 6700 Mb)

My perfs : 1600 per card at best.

I have the last drivers.

My cards are oc with +150 GPU, +800 RAM and 70 power limit, no problem with other miners...

Bruh, you wanna pass along bad info, that's up to you.. concurrent on same pool+coin or hashrate/share rate is the only way to go. Let go of the coins/difficulty idea, it's not useful.. I'm not picking on you, you just won't let that bad measuring method die. Let me kill it for you.