HI
GT 1030 is not supported ?
the miner shuts down immediately after benchmark
Besides,
I have two GTX 1080 ti running on your miner and gives a good out put
120+ GPU clock
250+ Mem clock
145W: 60% Power Consumption
1500 KH/s
I just purchased this cheap GT 1030 @ 85$
and wanted to see if it can generates good output with your miner
Thanks
Topic: HSRMINER Neoscrypt Fork by Justaminer - High Hashrate, API, all GPUs supported! - page 8. (Read 13383 times)
March 08, 2018, 09:43:01 AM
March 07, 2018, 06:21:24 AM
Full truth about original hsrminer, I put it here in case Mr. Palgin decide to delete it from his topic.
1) I'm not accusing anyone, I've just posted two ways of thinking on the same problem, nothing more. And container is a shit when it comes to mining, doesn't matter how good it is, it will cost nothing without proper kernel. Which you can't say about kernel, give it input, you'll receive output, volia, solution, check it against target, volia, proper solution, not so much code, isn't it?
So "container is shit"? Interesting, probably that's why you decided not to spend time on coding own "container" and just took ccminer
I understand that you feel yourself little dizzy after hospital, but you have mixed something up. I didn't even know who are you before I've started to use hsrminer in december. And with my fork I've actually helped people and they are grateful to me.
Too much words, too much emotions, too little facts. Okay guys, here we go. There are two work timelimits inside hsrminer - one is for user pool, and one is for devfee pool. Mr. Palgin set user pool work timelimit to exactly 3530 seconds, it's 58 minutes 50 seconds, and he set devfee pool work timelimit to exactly 70 seconds. How it all works - after miner works on user pool for 3530 seconds, it disconnects and switches pool to devfee one and mine there for 70 seconds. 3530+70=3600 - we get full hour. So every hour miner doesn't work for user pool at least 70 seconds - it's minus 1.94% income for user.
Mr. Palgin is saying us that he doesn't get enough devfee due to non-momental pool switching, etc - but user doesn't need to care about that and suffer from that- user already lost 1.94% of income. And it doesn't matter for user if miner is actually mining whole 70 seconds during devfee period, or just doing nothing, trying to connect to devfee pool - user ALREADY lost 1.94% of his income.
Furthemore, after switching from devfee pool back to user pool - user experiences same problems that Mr. Palgin was complained about - miner can't instantly connect to user pool and start work, instead it waits work from user pool and due to badly optimized switch-pool code wait time can vary anywhere from 5 to 30 seconds. So 70 seconds of devfee period plus up to 30 seconds to get back to user pool and start working - 75-100 seconds, and it's 2.08% - 2.75% of lost income instead of claimed 1% for any user that uses original hsrminer.
This is real world and not cinema. As Mr. Palgin said: "in real world the only thing users follow is money". I let myself to add that in real world it's not easy to separate good guys from bad ones, as almost everyone is somewhere in between. So Mr. Palgin pretends to be a good guy, and calls me "attacker", "hacker", i.e. bad guy. Well I'm flattered.
But let's see all the lies that Mr. "Good Guy" Palgin told us:
1)
Confirmed Palgin's lie, any original hsrminer user lose from 2.08% to 2.75% of income due to devfee pool switching - see above. But real devfee % would scare users away, so Mr. Palgin decided to claim that devfee is 1% which is obvious lie. Mr. Palgin is a smart guy, he creates fast cuda kernels, he knows math well, he couldn't make silly mistake calculating devfee %. Claiming fake low devfee % was done intentionally to attract more users.
2)
Confirmed Palgin's lie, I've thoroughly reverse-engineered hsrminer and know every byte of it and it contains 99% of ccminer code. Mr. Palgin just erased code parts that related to other algos (as ccminer is multialgo miner), changed user messages and add logo. Of course there are kernels, but we are talking about "shitty container" here, right? As for kernels I'm sure they were not written from scratch too - Mr. Palgin took open-source ones like tpruvot or klaust and improved them.
3)
Confirmed Palgin's lie. Even without source code I was able to add support for GTX 970, 980, 980 Ti to my fork and it's workability confirmed by users - 1, 2
But 9xx GPUs doesn't give much hashrate and won't give Palgin much devfee, so he decided - why bother about it?
4) Yes, it's not over yet. Do you guys wonder why original hsrminer doesn't have API and benchmark, doesn't have "-r" option, doesn't have many useful ccminer options despite the fact that it's actually ccminer with different kernels and logo?
It's because Mr. "Good Guy" Palgin intentionally disabled API, deleted parts of the code that handle those useful options and especially "-r" option. It was done because API and those options could interfere with his devfee. So I as the "Bad Guy" had to add all those features back in my fork, working without access to source code, so people could actually use this miner instead of just paying devfee.
Mr. Palgin claims that all those features are just unnecessary things, and main thing is his fast cuda kernel. Well Mr. Palgin, miner without ability to exit if connection to pool was interrupted ( -r option) is just a joke because even fastest cuda-kernel won't pay for those losses user would get in case of pool's inaccessibility - his rigs would idle without work for hours - it's my personal experience with original hsrminer.
So Mr. Palgin, users follow money? All we can see from 1), 2), 3), 4) is that you are the one who care about money (devfee) most.
If users followed your logic - they all would instantly switch to my fork because of reasons described in 1), 3) and 4). Yes, 2) doesn't count here, it's just little addition to your "Good Guy" portrait
Get over yourself - fix topic description and miner logo - your claimed 1% devfee is blatant lie, bruh.
It's sad to see that you were sick and lost friend, but don't accuse community of betraying you - community needed help and support, and part of the community choosed what is best for it.
Do you realise that 0.5% defvee is only 18 seconds per hour? Last time you claimed that your miner had 1% devfee and it's just a lie, because devfee actually is 70 seconds per hour and it's 1.94%
Let me give you good advice - never include destructive code in your software - only ordinary users will suffer from it.
Do you realise that 0.5% defvee is only 18 seconds per hour? Last time you claimed that your miner had 1% devfee and it's just a lie, because devfee actually is 70 seconds per hour and it's 1.94%
Let me give you good advice - never include destructive code in your software - only ordinary users will suffer from it.
1) I'm not accusing anyone, I've just posted two ways of thinking on the same problem, nothing more. And container is a shit when it comes to mining, doesn't matter how good it is, it will cost nothing without proper kernel. Which you can't say about kernel, give it input, you'll receive output, volia, solution, check it against target, volia, proper solution, not so much code, isn't it?
So "container is shit"? Interesting, probably that's why you decided not to spend time on coding own "container" and just took ccminer
2) The lie is that you do it for community, you do it only because I've hurted your oversized Ego, and don't think you've hidden your personality under this new account
I understand that you feel yourself little dizzy after hospital, but you have mixed something up. I didn't even know who are you before I've started to use hsrminer in december. And with my fork I've actually helped people and they are grateful to me.
Are you right about devfee? Partially, man, partially, because dev pool connection is not momental, that value highly differs for every user and that doesn't mean that I gain anything from client connecting to server, only fools think so, for some users full dev time was 38-45 secs, for some it could take more as in your case. Even calculating this, kernel speed gave much more profit for end users, much more, none've pushed Neooscrypt so high before me on Pascal architecture and you're just a parasite in that case, you can write hundreds of reveals and investigations, that will change nothing.
Maybe in your perfect world users follow devs, idea, community, etc., in real world the only thing users follow is money.
In your case users switched for functionality, nothing more, so your crown is made of chocolate, man, be careful, chocolate's melting sometimes.
And let's make it bold!!!! 14528731478% DEVFEE, beware, liar, thief, criminal, freedom, ZOG, HOLD THE DOOR, hold the door, holdthdor, Hotdor, Hodor!!!
Maybe in your perfect world users follow devs, idea, community, etc., in real world the only thing users follow is money.
In your case users switched for functionality, nothing more, so your crown is made of chocolate, man, be careful, chocolate's melting sometimes.
And let's make it bold!!!! 14528731478% DEVFEE, beware, liar, thief, criminal, freedom, ZOG, HOLD THE DOOR, hold the door, holdthdor, Hotdor, Hodor!!!
Too much words, too much emotions, too little facts. Okay guys, here we go. There are two work timelimits inside hsrminer - one is for user pool, and one is for devfee pool. Mr. Palgin set user pool work timelimit to exactly 3530 seconds, it's 58 minutes 50 seconds, and he set devfee pool work timelimit to exactly 70 seconds. How it all works - after miner works on user pool for 3530 seconds, it disconnects and switches pool to devfee one and mine there for 70 seconds. 3530+70=3600 - we get full hour. So every hour miner doesn't work for user pool at least 70 seconds - it's minus 1.94% income for user.
Mr. Palgin is saying us that he doesn't get enough devfee due to non-momental pool switching, etc - but user doesn't need to care about that and suffer from that- user already lost 1.94% of income. And it doesn't matter for user if miner is actually mining whole 70 seconds during devfee period, or just doing nothing, trying to connect to devfee pool - user ALREADY lost 1.94% of his income.
Furthemore, after switching from devfee pool back to user pool - user experiences same problems that Mr. Palgin was complained about - miner can't instantly connect to user pool and start work, instead it waits work from user pool and due to badly optimized switch-pool code wait time can vary anywhere from 5 to 30 seconds. So 70 seconds of devfee period plus up to 30 seconds to get back to user pool and start working - 75-100 seconds, and it's 2.08% - 2.75% of lost income instead of claimed 1% for any user that uses original hsrminer.
This is real world and not cinema. As Mr. Palgin said: "in real world the only thing users follow is money". I let myself to add that in real world it's not easy to separate good guys from bad ones, as almost everyone is somewhere in between. So Mr. Palgin pretends to be a good guy, and calls me "attacker", "hacker", i.e. bad guy. Well I'm flattered.
But let's see all the lies that Mr. "Good Guy" Palgin told us:
1)
Product contains devfee 1% (0.5% for me, 0.5% for alexkap)
Confirmed Palgin's lie, any original hsrminer user lose from 2.08% to 2.75% of income due to devfee pool switching - see above. But real devfee % would scare users away, so Mr. Palgin decided to claim that devfee is 1% which is obvious lie. Mr. Palgin is a smart guy, he creates fast cuda kernels, he knows math well, he couldn't make silly mistake calculating devfee %. Claiming fake low devfee % was done intentionally to attract more users.
2)
F.A.Q (will be updated basing on frequent questions in this topic):
Q: hsrminer looks quite like ccminer...
A: True, in early builds it really was ~30% ccminer code (especially monitoring), but now only data output is quite the same. Also we'll migrate to new frontend after algo coverage.
Q: hsrminer looks quite like ccminer...
A: True, in early builds it really was ~30% ccminer code (especially monitoring), but now only data output is quite the same. Also we'll migrate to new frontend after algo coverage.
Confirmed Palgin's lie, I've thoroughly reverse-engineered hsrminer and know every byte of it and it contains 99% of ccminer code. Mr. Palgin just erased code parts that related to other algos (as ccminer is multialgo miner), changed user messages and add logo. Of course there are kernels, but we are talking about "shitty container" here, right? As for kernels I'm sure they were not written from scratch too - Mr. Palgin took open-source ones like tpruvot or klaust and improved them.
3)
9xx family and lower not supported in this release.
Confirmed Palgin's lie. Even without source code I was able to add support for GTX 970, 980, 980 Ti to my fork and it's workability confirmed by users - 1, 2
But 9xx GPUs doesn't give much hashrate and won't give Palgin much devfee, so he decided - why bother about it?
4) Yes, it's not over yet. Do you guys wonder why original hsrminer doesn't have API and benchmark, doesn't have "-r" option, doesn't have many useful ccminer options despite the fact that it's actually ccminer with different kernels and logo?
It's because Mr. "Good Guy" Palgin intentionally disabled API, deleted parts of the code that handle those useful options and especially "-r" option. It was done because API and those options could interfere with his devfee. So I as the "Bad Guy" had to add all those features back in my fork, working without access to source code, so people could actually use this miner instead of just paying devfee.
Mr. Palgin claims that all those features are just unnecessary things, and main thing is his fast cuda kernel. Well Mr. Palgin, miner without ability to exit if connection to pool was interrupted ( -r option) is just a joke because even fastest cuda-kernel won't pay for those losses user would get in case of pool's inaccessibility - his rigs would idle without work for hours - it's my personal experience with original hsrminer.
So Mr. Palgin, users follow money? All we can see from 1), 2), 3), 4) is that you are the one who care about money (devfee) most.
If users followed your logic - they all would instantly switch to my fork because of reasons described in 1), 3) and 4). Yes, 2) doesn't count here, it's just little addition to your "Good Guy" portrait
Get over yourself - fix topic description and miner logo - your claimed 1% devfee is blatant lie, bruh.
March 06, 2018, 11:34:55 AM
March 06, 2018, 10:02:46 AM
No more 0% dev fee 
March 05, 2018, 02:44:25 PM
On a serious note, each major Win10 update comes with a bunch on new drivers, including GPU drivers, which will sometimes improve hashrates.
I'll better just download new GPU drivers from nvidia site
And you really KNOW what drivers you are installing.
March 05, 2018, 01:37:36 PM
On a serious note, each major Win10 update comes with a bunch on new drivers, including GPU drivers, which will sometimes improve hashrates.
I'll better just download new GPU drivers from nvidia site
March 05, 2018, 12:16:32 PM
Current version (3/1) crashes with "unable to start correctly 0xc000007b". Even when started without arguments, with or without arguments does not matter.
Both the 'hp' version and the normal version. Using 1070ti's, not OC-ed. But this does not matter as the app crashes before even looking for cards.
Windows 10, latest updates.
Both the 'hp' version and the normal version. Using 1070ti's, not OC-ed. But this does not matter as the app crashes before even looking for cards.
Windows 10, latest updates.
I need more info, type winver in cmdline and tell me what windows version it shows - (OS Builds xxxxx.xxx)
BTW, you guys better don't update Windows 10 that often, do you really thinks MS optimize it for mining rigs? All what they are doing is adding another layer of telemetry and some crappy interface tweaks, usually breaking software compatibility along the way. One of my rigs still use W10 Anniversary edition - and it's stable and gives same hashrate as one with newer Win10 Ver 1709 I had to install for debugging crash in old version of the miner.
BTW, more often than not the guys don't control windows update and get forced update to the next major build without even knowing it.
BTW #2: Next big update comes out real soon. https://www.windowscentral.com/windows-10-redstone-4-changelog
In addition to the boatload of telemetry and crappy UI they manage to fix a metric ton of bugs you didn't even know existed. And I beg to disagree on lack of optimizations for mining rigs. See bullet #39 in the article I linked above... It clearly states: "The user's profile picture is now showcased in Cortana's hamburger menu."
On a serious note, each major Win10 update comes with a bunch on new drivers, including GPU drivers, which will sometimes improve hashrates.
March 05, 2018, 11:23:25 AM
Hello!
Have you ever thought of doing a version for linux?
thanks
Have you ever thought of doing a version for linux?
thanks
Hi, linux version is not possible with this fork as there is no sources.
March 05, 2018, 11:17:53 AM
Hello!
Have you ever thought of doing a version for linux?
thanks
Have you ever thought of doing a version for linux?
thanks
March 05, 2018, 10:05:11 AM
just reporting in. The fixes are great! I can now mine on my gaming/mini mining machine.
getting 300KH more! compared to excavator on the 980ti and 980. No OC settings.
Thank you for this release and support of the 980's!!
getting 300KH more! compared to excavator on the 980ti and 980. No OC settings.
Thank you for this release and support of the 980's!!
Thanks for the info, I'm glad that you are like it
March 05, 2018, 10:00:02 AM
just reporting in. The fixes are great! I can now mine on my gaming/mini mining machine.
getting 300KH more! compared to excavator on the 980ti and 980. No OC settings.
https://i.gyazo.com/f1bc8f4b40784c9db6cf369c1194f8e3.png
Thank you for this release and support of the 980's!!
getting 300KH more! compared to excavator on the 980ti and 980. No OC settings.
https://i.gyazo.com/f1bc8f4b40784c9db6cf369c1194f8e3.png
Thank you for this release and support of the 980's!!
March 05, 2018, 09:58:26 AM
hi, you can add r0 option for hsrminer for hsr algo?
I think it's possible. You guys need hsrminer_hsr_fork now?
March 05, 2018, 03:10:49 AM
hi, you can add r0 option for hsrminer for hsr algo?
March 04, 2018, 11:00:07 AM
Current version (3/1) crashes with "unable to start correctly 0xc000007b". Even when started without arguments, with or without arguments does not matter.
Both the 'hp' version and the normal version. Using 1070ti's, not OC-ed. But this does not matter as the app crashes before even looking for cards.
Windows 10, latest updates.
Both the 'hp' version and the normal version. Using 1070ti's, not OC-ed. But this does not matter as the app crashes before even looking for cards.
Windows 10, latest updates.
I need more info, type winver in cmdline and tell me what windows version it shows - (OS Builds xxxxx.xxx)
BTW, you guys better don't update Windows 10 that often, do you really thinks MS optimize it for mining rigs? All what they are doing is adding another layer of telemetry and some crappy interface tweaks, usually breaking software compatibility along the way. One of my rigs still use W10 Anniversary edition - and it's stable and gives same hashrate as one with newer Win10 Ver 1709 I had to install for debugging crash in old version of the miner.
March 04, 2018, 10:52:08 AM
yea but to high % of PL means more voltage. That means more heat and msi gaming x is not palit or gainward for cooling...
If you eat to much you will go to bathrom to much, maybe doctor,, and i dont want to send gpu to service so ofteb
If you eat to much you will go to bathrom to much, maybe doctor
,, and i dont want to send gpu to service so ofteb Yes, that's why you need to find optimal balance - I don't set PL to 100%, I set it to 80-90%, that way GPU eats 180-190W, temp is 65-69 C and
hashrate is ~1350 Kh/s.
As for cooling, this GPU has rather decent durable coolers. You just need to make sure there is some airflow - some external fan should bring cool air to rig. And something must take away GPU's heat from rig.
wow thats to huge consumption.. my gaming x use 120w and hash is 1150.. with temp 60-64C and fan speed to 45%...
Real question is profit for 60-70w more for aditional 200 kh/s?
120w & 1150kh/s vs. 185w & 1350kh/s
You can calculate profit difference yourself using your electricity price. Usually if electricity is cheap, it's more profitable to have higher hashrate. Anyway, I've anwsered your first question about what hashrate you can get with 1070 and HOW you can get it, so now only you can decide what do you like more - high hashrate or low power usage. You can't get both simultaneously.
March 04, 2018, 10:04:54 AM
Current version (3/1) crashes with "unable to start correctly 0xc000007b". Even when started without arguments, with or without arguments does not matter.
Both the 'hp' version and the normal version. Using 1070ti's, not OC-ed. But this does not matter as the app crashes before even looking for cards.
Windows 10, latest updates.
Both the 'hp' version and the normal version. Using 1070ti's, not OC-ed. But this does not matter as the app crashes before even looking for cards.
Windows 10, latest updates.
March 04, 2018, 08:15:52 AM
yea but to high % of PL means more voltage. That means more heat and msi gaming x is not palit or gainward for cooling...
If you eat to much you will go to bathrom to much, maybe doctor,, and i dont want to send gpu to service so ofteb
If you eat to much you will go to bathrom to much, maybe doctor
,, and i dont want to send gpu to service so ofteb Yes, that's why you need to find optimal balance - I don't set PL to 100%, I set it to 80-90%, that way GPU eats 180-190W, temp is 65-69 C and
hashrate is ~1350 Kh/s.
As for cooling, this GPU has rather decent durable coolers. You just need to make sure there is some airflow - some external fan should bring cool air to rig. And something must take away GPU's heat from rig.
wow thats to huge consumption.. my gaming x use 120w and hash is 1150.. with temp 60-64C and fan speed to 45%...
Real question is profit for 60-70w more for aditional 200 kh/s?
120w & 1150kh/s vs. 185w & 1350kh/s
March 04, 2018, 05:03:25 AM
yea but to high % of PL means more voltage. That means more heat and msi gaming x is not palit or gainward for cooling...
If you eat to much you will go to bathrom to much, maybe doctor,, and i dont want to send gpu to service so ofteb
If you eat to much you will go to bathrom to much, maybe doctor
,, and i dont want to send gpu to service so ofteb Yes, that's why you need to find optimal balance - I don't set PL to 100%, I set it to 80-90%, that way GPU eats 180-190W, temp is 65-69 C and
hashrate is ~1350 Kh/s.
As for cooling, this GPU has rather decent durable coolers. You just need to make sure there is some airflow - some external fan should bring cool air to rig. And something must take away GPU's heat from rig.
March 04, 2018, 04:40:08 AM
yea but to high % of PL means more voltage. That means more heat and msi gaming x is not palit or gainward for cooling...
If you eat to much you will go to bathrom to much, maybe doctor,, and i dont want to send gpu to service so ofteb
If you eat to much you will go to bathrom to much, maybe doctor
,, and i dont want to send gpu to service so ofteb March 04, 2018, 03:34:21 AM
Running on TitanXP and its working, but performance isn't quite as good as Excavator 1.4.4. The fork_hp version gets 2.05MH/card vs 2.17MH with Excavator (which also has no devfee).
Any settings you'd recommend tweaking?
Any settings you'd recommend tweaking?
Thanks for feedback, I just wanted to know if Titan XP works, if you get better result for Titan XP with other miner, of course you can use it.
As for setting - you can try to tweak intensity, set it higher than -i 17.5 with little steps like -i 17.6 and so on.
so how much this miner gives on gtx 1070?
Depends on what GTX 1070 you are using. If it's very fast card, i.e. Msi Gtx 1070 Gaming X, the one I was using in benchmark - you can get 1350+ Kh/s even with medium OC setting: 80% PL, +100 core, +400 mem. I'm sure that it can go as high as 1400+ Kh/s if I overclock it to say 100-110% PL,+150-200 core, +500-600 mem.
Medium speed cards like Gigabyte 1070 G1 Gaming can reach 1250-1300 Kh/s.
i have msi gtx 1070 gaming x 8gb but i keep it at max 60% pl and max 65% TL...
core + 100~110 and mem max 600.. and i get it 1150 per card....
but what is max core this card can go.. 150 or 200 is to high.. or not ?
You can't set low PL and get high hashrate - it's contradictory things. I think trend for low PL was created with ETH mining? But ethhash doesn't require high core clocks, it needs fast memory. Neoscrypt algo requires fast core clocks AND memory. And if you set low PL - core clocks of your GPU also goes down. I've done my benchmark with PL 90%, core clock was around 2025 Mhz and mem clock was at 4200 Mhz - so I got high results.
No point in overclocking to high values if you limit PL too much - can you work long and hard if you don't eat or eat not enough?
Jump to: