fullzero (OP)
Legendary
Offline
Activity: 1260
Merit: 1009
|
|
August 19, 2017, 05:24:12 PM |
|
use the download link from the OP: In your picture you are hashing the img; the hash to compare is from the zip file itself (as this is what you download; this is what I generate a hash to test against for). Let me know if the zip produces the correct hash.
|
|
|
|
fullzero (OP)
Legendary
Offline
Activity: 1260
Merit: 1009
|
|
August 19, 2017, 05:32:16 PM |
|
Hi Fullzero, I really love your work, I got my gtx 1070 working and mining around 31~mhs each. My system boots up using the thumbdrive, but I would like to use my hdd instead. My hard disk drive has Windows 10 installed on it, I'm aware that nvOC will zero it out due to "dd if=/dev/zero of=/dev/sda" https://bitcointalk.org/index.php?topic=1854250.msg20290398#msg20290398. What I would like to do is to dual boot Windows 10 together with nvOC on it. Is there possible way to do that? If not, is it possible to install the nvOC in hdd instead of thumbdrive? Thanks in advance, it might be helpful to some. Not sure whether dual boot is possible but you can install nvOC on separate HDD as well, it still works like normal. To All Ubuntu / *nix guru outhere,
I read the 2unix script and I still don't understand why it's removing the "Windows partition data". I'm pretty sure the thumbdrive is mounted under m1 alias, but why do we need to zero the sda partition. That doesn't make sense to me. (Is it a preparation in case I need to copy the mounted m1 to sda?)
My aim is to dual boot Windows 10 with it.
You can do just about any multi boot if you are good with manually editing partitions / grub loader. However; I don't recommend doing this. I will test not zeroing the first partition and only removing 1bash from it (after transfer) at first boot for v0019. Note; it is only the initial boot of v0018 that does this action. I also don't recommend using the standard partition set when/if installing windows but that is another topic I don't want to get into.
|
|
|
|
fullzero (OP)
Legendary
Offline
Activity: 1260
Merit: 1009
|
|
August 19, 2017, 05:34:32 PM |
|
hi fullzero so much thanks to all your works, just wondering on your next release is it possible for you to make the whole nvoc smaller on size so it can run on a 16gb flash drive? just my 2 cents
thanks
God Bless
I am assuming you mean a 15.5 gb USB key marketed as a 16gb. I will make the v0019 image 15.5 gb.
|
|
|
|
fullzero (OP)
Legendary
Offline
Activity: 1260
Merit: 1009
|
|
August 19, 2017, 05:41:42 PM |
|
Thanks for posting this info - I banged my head against a wall for quite a while trying to figure this error out before giving up on it. This worked like a charm for me.
Though I think I'm still getting really low hashrates for skunk - whattomine thinks I should be getting 450mh/s with 5 1080 Tis and 7 1070s but I was only getting around 300mh/s. With the new krnlx miner I'm still only getting about 370Mh/s, which still seems pretty low. My miner also restarts with the watchdog reporting under utilization. My 1080 Tis are only running at about 60% and 1070s at 80-85% capacity. Can't figure out what's causing that.
EDIT: I added the '-i 25' flag to the command line and boosted core clock on the 1070s to +125 now I'm up at 480mh/s =)
Hum just stay away from krnlx release ... Low report share is a "know" problem ... I lost one week of mining on this stupid release ... Just dont USE krlnx this dev drop a brocken build and vanish ... First and last one i use something from him ... So a basic; krlnx mod boost your hashrate but you loose 30% expected share ... Use new ccminer release from tpru u get same hashrate and REAL SHARE. For 1070 u can push -i 25, Coreclock + 100/140 depend on card, MemoryClock -1500. I dont use 1080, but i bet u can push them to -i 26, to grab a solid 50mh/s+ per 1080ti https://github.com/tpruvot/ccminer/releases link to tpruvot version. Yes indeed... How do you lose 30% of expected shares? You mean rejected shares? Not rejected shares, is a weird thing and i dont analyze deeper on this. I get 70% of expected share with this hashrate report, so remove 30% of your hashrate when u use krnlx Where are you getting your expected shares tally from? I think that varies based on difficulty of the shares calculated which is determined by the stratum server, not your machine, so i don't know how you'd reliably determine expected shares in order to reason that you're losing 30%. Right now with 460-480Mh/s I'm getting about 22 coins an hour, which seems to be in line with the predictions of whattomine.com's mining calculator profit wise, ~ 525-550 coins a day @ $.07 per coin = ~$37-38.00 /day. I haven't mined at length with TPruvot ccminer 2.2 with the current configuration on my system, so I don't know if it would result in higher profit, but krnlx seems to be on point for profitability estimates now that I've got it properly configured and my clock settings are dialed in. FWIW, I'm not sure why you're downclocking your memory so much. I overclock mine +100 on 1070s and +800 on 1080 Tis. I see no benefit from underclocking memory on Nvidia cards really since power usage comes from core clock speed primarily, and it could result in unstable mining to underclock memory too much. EDIT: I had the under use problem with Watchdog as well, but I took care of that by raising my clock speed on my GPUs from +100 to +125 and setting intensity to -i 25, now I don't have watchdog restarts at all. I get same result on suprnova or pool.mn, for 2 days on each after i install krnlx, i lost 30% income and lost 30% on wathtomine prediction. I give a try to push my MemoryClock to +100 and i lost hashrate instant, prolly depend from the card. Maybe try to lower your MemoryClock u gona gain more hashrate. I use 8 x ASUS 1070 8OG, i have stable for more them 24h running and get 242+mh/s After reading this: I don't want to include the krnlx fork. Instead I will update the tpruvot to the newest version for v0019. Is this the optimal client right now for skunk?
|
|
|
|
Bibi187
Full Member
Offline
Activity: 420
Merit: 106
https://steemit.com/@bibi187
|
|
August 19, 2017, 06:40:27 PM |
|
Thanks for posting this info - I banged my head against a wall for quite a while trying to figure this error out before giving up on it. This worked like a charm for me.
Though I think I'm still getting really low hashrates for skunk - whattomine thinks I should be getting 450mh/s with 5 1080 Tis and 7 1070s but I was only getting around 300mh/s. With the new krnlx miner I'm still only getting about 370Mh/s, which still seems pretty low. My miner also restarts with the watchdog reporting under utilization. My 1080 Tis are only running at about 60% and 1070s at 80-85% capacity. Can't figure out what's causing that.
EDIT: I added the '-i 25' flag to the command line and boosted core clock on the 1070s to +125 now I'm up at 480mh/s =)
Hum just stay away from krnlx release ... Low report share is a "know" problem ... I lost one week of mining on this stupid release ... Just dont USE krlnx this dev drop a brocken build and vanish ... First and last one i use something from him ... So a basic; krlnx mod boost your hashrate but you loose 30% expected share ... Use new ccminer release from tpru u get same hashrate and REAL SHARE. For 1070 u can push -i 25, Coreclock + 100/140 depend on card, MemoryClock -1500. I dont use 1080, but i bet u can push them to -i 26, to grab a solid 50mh/s+ per 1080ti https://github.com/tpruvot/ccminer/releases link to tpruvot version. Yes indeed... How do you lose 30% of expected shares? You mean rejected shares? Not rejected shares, is a weird thing and i dont analyze deeper on this. I get 70% of expected share with this hashrate report, so remove 30% of your hashrate when u use krnlx Where are you getting your expected shares tally from? I think that varies based on difficulty of the shares calculated which is determined by the stratum server, not your machine, so i don't know how you'd reliably determine expected shares in order to reason that you're losing 30%. Right now with 460-480Mh/s I'm getting about 22 coins an hour, which seems to be in line with the predictions of whattomine.com's mining calculator profit wise, ~ 525-550 coins a day @ $.07 per coin = ~$37-38.00 /day. I haven't mined at length with TPruvot ccminer 2.2 with the current configuration on my system, so I don't know if it would result in higher profit, but krnlx seems to be on point for profitability estimates now that I've got it properly configured and my clock settings are dialed in. FWIW, I'm not sure why you're downclocking your memory so much. I overclock mine +100 on 1070s and +800 on 1080 Tis. I see no benefit from underclocking memory on Nvidia cards really since power usage comes from core clock speed primarily, and it could result in unstable mining to underclock memory too much. EDIT: I had the under use problem with Watchdog as well, but I took care of that by raising my clock speed on my GPUs from +100 to +125 and setting intensity to -i 25, now I don't have watchdog restarts at all. I get same result on suprnova or pool.mn, for 2 days on each after i install krnlx, i lost 30% income and lost 30% on wathtomine prediction. I give a try to push my MemoryClock to +100 and i lost hashrate instant, prolly depend from the card. Maybe try to lower your MemoryClock u gona gain more hashrate. I use 8 x ASUS 1070 8OG, i have stable for more them 24h running and get 242+mh/s After reading this: I don't want to include the krnlx fork. Instead I will update the tpruvot to the newest version for v0019. Is this the optimal client right now for skunk? So far i know trpuvot fork is the best for skunk algo, i dont test new palginmod release https://github.com/palginpav/ccminer/releases cause is build for window. SP_mod give good hashrate but u have to pay for his mod
|
|
|
|
BigSmurf
Newbie
Offline
Activity: 15
Merit: 0
|
|
August 19, 2017, 10:41:37 PM |
|
Hi,
can you explain to me how to set Power limit for GPU-z ? also when i try to OC on bash file no effect, i must do it in powermizer to work.
On another note I have Asrock H110 BTC+ PRO with 13*P106-100 6GB GPU-z, they all work fine, and i see ppl asking about CPU, G3900 i not enough minimum you should tell them is i3 7100 3.9ghz 2-Cor/4-Thread i had G3900 but system froze a lot as CPU usage was 70+% now with the i3 everything is good, CPU usage avg. 20-30%
The problem is with the P106-100 not being properly recognized in nvOC. Until they are; you will not be able to use powerlimit or OC with them. Hi, i have successfully reduced power-limit (via 1bash) as was instructed by the program its just trial and error, as there is not a tutorial for nvOC. ( set maximus =YES and Individual Power set =YES then set preferred values and restart system.) I confirm that it works.for OC powermiser works but tomorrow we will test if it will work with 1bash file. we are running 1*gtx1060 for vga(display) and 12*p106-100 all cards are running and working. Until you get hold of P106-100 to make it possible to run 13*P106-100 ppl must use one normal gaming GPU.
|
|
|
|
EvgeniX
Newbie
Offline
Activity: 18
Merit: 0
|
|
August 20, 2017, 12:00:51 AM |
|
How to add Claymore 9.8 Im keep getting error with screen attach?
|
|
|
|
car1999
|
|
August 20, 2017, 12:06:27 AM |
|
Hi,
can you explain to me how to set Power limit for GPU-z ? also when i try to OC on bash file no effect, i must do it in powermizer to work.
On another note I have Asrock H110 BTC+ PRO with 13*P106-100 6GB GPU-z, they all work fine, and i see ppl asking about CPU, G3900 i not enough minimum you should tell them is i3 7100 3.9ghz 2-Cor/4-Thread i had G3900 but system froze a lot as CPU usage was 70+% now with the i3 everything is good, CPU usage avg. 20-30%
The problem is with the P106-100 not being properly recognized in nvOC. Until they are; you will not be able to use powerlimit or OC with them. Hi, i have successfully reduced power-limit (via 1bash) as was instructed by the program its just trial and error, as there is not a tutorial for nvOC. ( set maximus =YES and Individual Power set =YES then set preferred values and restart system.) I confirm that it works.for OC powermiser works but tomorrow we will test if it will work with 1bash file. we are running 1*gtx1060 for vga(display) and 12*p106-100 all cards are running and working. Until you get hold of P106-100 to make it possible to run 13*P106-100 ppl must use one normal gaming GPU. i can confirm that one normal GPU with many p106 works well out of the box.
|
|
|
|
car1999
|
|
August 20, 2017, 12:08:10 AM |
|
How to add Claymore 9.8 Im keep getting error with screen attach?
download 9.8, extract to ~/eth/9_8, replace 9_7 with 9_8 in 1bash.
|
|
|
|
car1999
|
|
August 20, 2017, 12:13:13 AM |
|
Thanks for posting this info - I banged my head against a wall for quite a while trying to figure this error out before giving up on it. This worked like a charm for me.
Though I think I'm still getting really low hashrates for skunk - whattomine thinks I should be getting 450mh/s with 5 1080 Tis and 7 1070s but I was only getting around 300mh/s. With the new krnlx miner I'm still only getting about 370Mh/s, which still seems pretty low. My miner also restarts with the watchdog reporting under utilization. My 1080 Tis are only running at about 60% and 1070s at 80-85% capacity. Can't figure out what's causing that.
EDIT: I added the '-i 25' flag to the command line and boosted core clock on the 1070s to +125 now I'm up at 480mh/s =)
Hum just stay away from krnlx release ... Low report share is a "know" problem ... I lost one week of mining on this stupid release ... Just dont USE krlnx this dev drop a brocken build and vanish ... First and last one i use something from him ... So a basic; krlnx mod boost your hashrate but you loose 30% expected share ... Use new ccminer release from tpru u get same hashrate and REAL SHARE. For 1070 u can push -i 25, Coreclock + 100/140 depend on card, MemoryClock -1500. I dont use 1080, but i bet u can push them to -i 26, to grab a solid 50mh/s+ per 1080ti https://github.com/tpruvot/ccminer/releases link to tpruvot version. Yes indeed... How do you lose 30% of expected shares? You mean rejected shares? Not rejected shares, is a weird thing and i dont analyze deeper on this. I get 70% of expected share with this hashrate report, so remove 30% of your hashrate when u use krnlx Where are you getting your expected shares tally from? I think that varies based on difficulty of the shares calculated which is determined by the stratum server, not your machine, so i don't know how you'd reliably determine expected shares in order to reason that you're losing 30%. Right now with 460-480Mh/s I'm getting about 22 coins an hour, which seems to be in line with the predictions of whattomine.com's mining calculator profit wise, ~ 525-550 coins a day @ $.07 per coin = ~$37-38.00 /day. I haven't mined at length with TPruvot ccminer 2.2 with the current configuration on my system, so I don't know if it would result in higher profit, but krnlx seems to be on point for profitability estimates now that I've got it properly configured and my clock settings are dialed in. FWIW, I'm not sure why you're downclocking your memory so much. I overclock mine +100 on 1070s and +800 on 1080 Tis. I see no benefit from underclocking memory on Nvidia cards really since power usage comes from core clock speed primarily, and it could result in unstable mining to underclock memory too much. EDIT: I had the under use problem with Watchdog as well, but I took care of that by raising my clock speed on my GPUs from +100 to +125 and setting intensity to -i 25, now I don't have watchdog restarts at all. I get same result on suprnova or pool.mn, for 2 days on each after i install krnlx, i lost 30% income and lost 30% on wathtomine prediction. I give a try to push my MemoryClock to +100 and i lost hashrate instant, prolly depend from the card. Maybe try to lower your MemoryClock u gona gain more hashrate. I use 8 x ASUS 1070 8OG, i have stable for more them 24h running and get 242+mh/s After reading this: I don't want to include the krnlx fork. Instead I will update the tpruvot to the newest version for v0019. Is this the optimal client right now for skunk? So far i know trpuvot fork is the best for skunk algo, i dont test new palginmod release https://github.com/palginpav/ccminer/releases cause is build for window. SP_mod give good hashrate but u have to pay for his mod I'm using trpuvot too, looks great.
|
|
|
|
Bigf00t
Newbie
Offline
Activity: 60
Merit: 0
|
|
August 20, 2017, 01:23:00 AM |
|
Bigf00t, did you mean latest release of ethminer?? fullzero will add it in v19 hopefully (you can also do it mean while, if you wish to have it)
If not etminer already exists like Bibi187 saying.
I am a Linux newbie, but I was thinking too complicated. I simply replaced the file in the Genoil folder with the ethminer 0.12.0.dev2 file and it works like I wanted
|
|
|
|
TenaciousJ
|
|
August 20, 2017, 02:58:11 AM Last edit: August 20, 2017, 05:20:23 AM by TenaciousJ |
|
Thanks for posting this info - I banged my head against a wall for quite a while trying to figure this error out before giving up on it. This worked like a charm for me.
Though I think I'm still getting really low hashrates for skunk - whattomine thinks I should be getting 450mh/s with 5 1080 Tis and 7 1070s but I was only getting around 300mh/s. With the new krnlx miner I'm still only getting about 370Mh/s, which still seems pretty low. My miner also restarts with the watchdog reporting under utilization. My 1080 Tis are only running at about 60% and 1070s at 80-85% capacity. Can't figure out what's causing that.
EDIT: I added the '-i 25' flag to the command line and boosted core clock on the 1070s to +125 now I'm up at 480mh/s =)
Hum just stay away from krnlx release ... Low report share is a "know" problem ... I lost one week of mining on this stupid release ... Just dont USE krlnx this dev drop a brocken build and vanish ... First and last one i use something from him ... So a basic; krlnx mod boost your hashrate but you loose 30% expected share ... Use new ccminer release from tpru u get same hashrate and REAL SHARE. For 1070 u can push -i 25, Coreclock + 100/140 depend on card, MemoryClock -1500. I dont use 1080, but i bet u can push them to -i 26, to grab a solid 50mh/s+ per 1080ti https://github.com/tpruvot/ccminer/releases link to tpruvot version. Yes indeed... How do you lose 30% of expected shares? You mean rejected shares? Not rejected shares, is a weird thing and i dont analyze deeper on this. I get 70% of expected share with this hashrate report, so remove 30% of your hashrate when u use krnlx Where are you getting your expected shares tally from? I think that varies based on difficulty of the shares calculated which is determined by the stratum server, not your machine, so i don't know how you'd reliably determine expected shares in order to reason that you're losing 30%. Right now with 460-480Mh/s I'm getting about 22 coins an hour, which seems to be in line with the predictions of whattomine.com's mining calculator profit wise, ~ 525-550 coins a day @ $.07 per coin = ~$37-38.00 /day. I haven't mined at length with TPruvot ccminer 2.2 with the current configuration on my system, so I don't know if it would result in higher profit, but krnlx seems to be on point for profitability estimates now that I've got it properly configured and my clock settings are dialed in. FWIW, I'm not sure why you're downclocking your memory so much. I overclock mine +100 on 1070s and +800 on 1080 Tis. I see no benefit from underclocking memory on Nvidia cards really since power usage comes from core clock speed primarily, and it could result in unstable mining to underclock memory too much. EDIT: I had the under use problem with Watchdog as well, but I took care of that by raising my clock speed on my GPUs from +100 to +125 and setting intensity to -i 25, now I don't have watchdog restarts at all. I get same result on suprnova or pool.mn, for 2 days on each after i install krnlx, i lost 30% income and lost 30% on wathtomine prediction. I give a try to push my MemoryClock to +100 and i lost hashrate instant, prolly depend from the card. Maybe try to lower your MemoryClock u gona gain more hashrate. I use 8 x ASUS 1070 8OG, i have stable for more them 24h running and get 242+mh/s After reading this: I don't want to include the krnlx fork. Instead I will update the tpruvot to the newest version for v0019. Is this the optimal client right now for skunk? I ran the 2.2 version of TPruvot ccminer w/ skunk and got horrible hashrates. I will recompile and test it again, but for now, krnlx is the fastest for me. I'm not sure why there is a perceived drop of 30% in his setup, but I'm not seeing that at all. I'm mining exactly at the expected coin rate per day based on both whattomine.com and crypto-economy.com calculators (450mh/s nets me around 550 coins a day +/- 25). I will note that when I tried changing clock settings a little higher it really affected hashrates, dropping them over time considerably. There is definitely a 'sweet spot' clock rate for skunk, and deviation appears to create greatly reduced production. Also mining at -i 26 seems to have reduced my hashrates, so -i 25 is the proper setting for 1070 / 1080 Ti at least. Lower settings or leaving -i blank greatly reduce hashrates, as is to be expected. I'll update with the results of a recompile with tpruvot's most recent ccminer release on 0018. Thanks for keeping an eye on this. EDIT: the latest tpruvot ccminer release runs at just about the same hashrate as krnlx within 1-2mh/s (463mh/s vs. 462mh/s) and in a 2 hour test it mined 26-27 coins an hour on sigt.pool.mn which is a small number more than I was mining before, but it's a shjort test length so I'll update again after 24 hours to get a better average. I was able to run tpruvot with skunk from command line in guake, but it crashes for some reason when running from the script almost immediately after it starts mining (before watchdog kicks in). Not sure why that's happening, the command line is the same in both as I have it configured. In the end, I don't see a dramatic difference between the two miners' performance except that I know krnlx runs from the script properly, whereas TPccminer 2.2 does not work in the script yet. I'll keep playing around with it to see if some code in the script I added has an error. It's strange because the miner actually starts mining properly, but then errors out, rather than just not running at all which is normally how that goes.
|
|
|
|
ruanyu
Newbie
Offline
Activity: 1
Merit: 0
|
|
August 20, 2017, 05:24:15 AM |
|
Hi fullzero, Thank for your awesome distro. It help me alot!!!
If u dont mind, i think here are some improvements for your 1 bash file:
1. Use "eval" instead of very very long repeatedly redundancy source code, For example:
for j in $TI do for (( i=0; i<GPUS; i++ )) ; do eval \${NVD} -a [gpu:\${i}]/GPUGraphicsClockOffset[\${j}]=\${__CORE_OVERCLOCK_$i} eval \${NVD} -a [gpu:\${i}]/GPUMemoryTransferRateOffset[\${j}]=\${MEMORY_OVERCLOCK_$i} done done
or
if [[ $MANUAL_FAN == "YES" && $Maxximus007_AUTO_TEMPERATURE_CONTROL == "NO" ]] then for (( i=0; i<GPUS; i++ )) ; do eval \${NVD} -a [gpu:\${i}]/GPUFanControlState=1 eval \${NVD} -a [fan:\${i}]/GPUTargetFanSpeed=\${FAN_SPEED_$i} done fi
2. Move repeatedly source code to a function For example: function SLOW_USB_KEY_DELAY(){ if [ $SLOW_USB_KEY_MODE == "YES" ] then sleep 6 fi }
3. Use directly $CLAYMORE_VERSION to create claymore path instead of using if...fi For example: $CLAYMORE_VERSION="9_8" HCD="/home/m1/eth/$CLAYMORE_VERSION/ethdcrminer64"
Sorry for my bad English but hope that it will help u Cheers,
|
|
|
|
Thagodeus
Newbie
Offline
Activity: 4
Merit: 0
|
|
August 20, 2017, 06:22:27 AM Last edit: August 21, 2017, 06:03:52 AM by Thagodeus |
|
So I'm pretty new to mining, have 2 rigs. 1 running 2 1080ti's and 1 1060 mini. And then the other rig has 1 1080ti and one 1060 mini (other 1080ti is bunk) I tried downloading nvoc18 but it didn't cooperate and the copy just crashed upon boot. So I imaged back to nvoc17 and it's running well other than this. Stuff about the individual power settings and teamviewer etc those show up every time I've run 17, even with multiple re image attempts. My buddy voskcoin has been helping me troubleshoot my problems and getting me started, but he wasn't even sure about this. http://i999.photobucket.com/albums/af112/Ben_klink/E411BAAB-938C-42A3-BDE6-0B6B5610CF6A_zpskweq0qgp.jpg
|
|
|
|
damNmad
Full Member
Offline
Activity: 378
Merit: 104
nvOC forever
|
|
August 20, 2017, 07:50:26 AM |
|
How to add Claymore 9.8 Im keep getting error with screen attach?
download 9.8, extract to ~/eth/9_8, replace 9_7 with 9_8 in 1bash. Like the above post, download it, move it to /eth/ rename the 9_7 to 9_7_old or something and rename the 9_8xxxxx version to 9_7, should work normally. If you want to use 9_8 as is, you need to change add the directory in lot of places on 1bash.
|
|
|
|
Bibi187
Full Member
Offline
Activity: 420
Merit: 106
https://steemit.com/@bibi187
|
|
August 20, 2017, 08:59:11 AM |
|
Thanks for posting this info - I banged my head against a wall for quite a while trying to figure this error out before giving up on it. This worked like a charm for me.
Though I think I'm still getting really low hashrates for skunk - whattomine thinks I should be getting 450mh/s with 5 1080 Tis and 7 1070s but I was only getting around 300mh/s. With the new krnlx miner I'm still only getting about 370Mh/s, which still seems pretty low. My miner also restarts with the watchdog reporting under utilization. My 1080 Tis are only running at about 60% and 1070s at 80-85% capacity. Can't figure out what's causing that.
EDIT: I added the '-i 25' flag to the command line and boosted core clock on the 1070s to +125 now I'm up at 480mh/s =)
Hum just stay away from krnlx release ... Low report share is a "know" problem ... I lost one week of mining on this stupid release ... Just dont USE krlnx this dev drop a brocken build and vanish ... First and last one i use something from him ... So a basic; krlnx mod boost your hashrate but you loose 30% expected share ... Use new ccminer release from tpru u get same hashrate and REAL SHARE. For 1070 u can push -i 25, Coreclock + 100/140 depend on card, MemoryClock -1500. I dont use 1080, but i bet u can push them to -i 26, to grab a solid 50mh/s+ per 1080ti https://github.com/tpruvot/ccminer/releases link to tpruvot version. Yes indeed... How do you lose 30% of expected shares? You mean rejected shares? Not rejected shares, is a weird thing and i dont analyze deeper on this. I get 70% of expected share with this hashrate report, so remove 30% of your hashrate when u use krnlx Where are you getting your expected shares tally from? I think that varies based on difficulty of the shares calculated which is determined by the stratum server, not your machine, so i don't know how you'd reliably determine expected shares in order to reason that you're losing 30%. Right now with 460-480Mh/s I'm getting about 22 coins an hour, which seems to be in line with the predictions of whattomine.com's mining calculator profit wise, ~ 525-550 coins a day @ $.07 per coin = ~$37-38.00 /day. I haven't mined at length with TPruvot ccminer 2.2 with the current configuration on my system, so I don't know if it would result in higher profit, but krnlx seems to be on point for profitability estimates now that I've got it properly configured and my clock settings are dialed in. FWIW, I'm not sure why you're downclocking your memory so much. I overclock mine +100 on 1070s and +800 on 1080 Tis. I see no benefit from underclocking memory on Nvidia cards really since power usage comes from core clock speed primarily, and it could result in unstable mining to underclock memory too much. EDIT: I had the under use problem with Watchdog as well, but I took care of that by raising my clock speed on my GPUs from +100 to +125 and setting intensity to -i 25, now I don't have watchdog restarts at all. I get same result on suprnova or pool.mn, for 2 days on each after i install krnlx, i lost 30% income and lost 30% on wathtomine prediction. I give a try to push my MemoryClock to +100 and i lost hashrate instant, prolly depend from the card. Maybe try to lower your MemoryClock u gona gain more hashrate. I use 8 x ASUS 1070 8OG, i have stable for more them 24h running and get 242+mh/s After reading this: I don't want to include the krnlx fork. Instead I will update the tpruvot to the newest version for v0019. Is this the optimal client right now for skunk? I ran the 2.2 version of TPruvot ccminer w/ skunk and got horrible hashrates. I will recompile and test it again, but for now, krnlx is the fastest for me. I'm not sure why there is a perceived drop of 30% in his setup, but I'm not seeing that at all. I'm mining exactly at the expected coin rate per day based on both whattomine.com and crypto-economy.com calculators (450mh/s nets me around 550 coins a day +/- 25). I will note that when I tried changing clock settings a little higher it really affected hashrates, dropping them over time considerably. There is definitely a 'sweet spot' clock rate for skunk, and deviation appears to create greatly reduced production. Also mining at -i 26 seems to have reduced my hashrates, so -i 25 is the proper setting for 1070 / 1080 Ti at least. Lower settings or leaving -i blank greatly reduce hashrates, as is to be expected. I'll update with the results of a recompile with tpruvot's most recent ccminer release on 0018. Thanks for keeping an eye on this. EDIT: the latest tpruvot ccminer release runs at just about the same hashrate as krnlx within 1-2mh/s (463mh/s vs. 462mh/s) and in a 2 hour test it mined 26-27 coins an hour on sigt.pool.mn which is a small number more than I was mining before, but it's a shjort test length so I'll update again after 24 hours to get a better average. I was able to run tpruvot with skunk from command line in guake, but it crashes for some reason when running from the script almost immediately after it starts mining (before watchdog kicks in). Not sure why that's happening, the command line is the same in both as I have it configured. In the end, I don't see a dramatic difference between the two miners' performance except that I know krnlx runs from the script properly, whereas TPccminer 2.2 does not work in the script yet. I'll keep playing around with it to see if some code in the script I added has an error. It's strange because the miner actually starts mining properly, but then errors out, rather than just not running at all which is normally how that goes. Yeap skunk have sweet spot for every brand i think. U have to manage to find the fine tunning. I am not sure about this but i bet 1080ti can grab better performance on -i 26, i think u assign intensity per gpu like -d 0,1,2,3,4,5,6 -i 25 -d 7,8,9,10,11 -i 26 Anyway i get a bad experience with krnlx, before ccminer update, i was thinking about going back to palginmod with -20% hashrate ( not the new one ) I try to fix my variation GPU volatility, i cant run watchdog without lot of reboot by GPU sometime go to 0% for 1 second, everything is fine but watchdog dont like this When u try to reboot miner via quake, i dont know why but on ccminer build, script cant kill "irq" is state in zombie mode. So when u launch miner again, GPU cant assign correct value cause if this "irq" zombie, just do "sudo pkill -e irq" and u can launch it
|
|
|
|
fullzero (OP)
Legendary
Offline
Activity: 1260
Merit: 1009
|
|
August 20, 2017, 04:39:23 PM |
|
Hi,
can you explain to me how to set Power limit for GPU-z ? also when i try to OC on bash file no effect, i must do it in powermizer to work.
On another note I have Asrock H110 BTC+ PRO with 13*P106-100 6GB GPU-z, they all work fine, and i see ppl asking about CPU, G3900 i not enough minimum you should tell them is i3 7100 3.9ghz 2-Cor/4-Thread i had G3900 but system froze a lot as CPU usage was 70+% now with the i3 everything is good, CPU usage avg. 20-30%
The problem is with the P106-100 not being properly recognized in nvOC. Until they are; you will not be able to use powerlimit or OC with them. Hi, i have successfully reduced power-limit (via 1bash) as was instructed by the program its just trial and error, as there is not a tutorial for nvOC. ( set maximus =YES and Individual Power set =YES then set preferred values and restart system.) I confirm that it works.for OC powermiser works but tomorrow we will test if it will work with 1bash file. we are running 1*gtx1060 for vga(display) and 12*p106-100 all cards are running and working. Until you get hold of P106-100 to make it possible to run 13*P106-100 ppl must use one normal gaming GPU. i can confirm that one normal GPU with many p106 works well out of the box. This is good to know. Most likely P106-100 issues members have had are resulting from using integrated graphics rather than the P106-100s. I still need to test accessing a pure P106-100 rig remotely (which has never had a monitor attached to its integrated graphics) to verify this. It is significantly easier to use / manage a rig in remote anyway. The one part I expect most members might have trouble with, while accessing a fully headless rig; is finding the new rigs IP. This is easily solved using a network scanner; of which there are now many GUI, user friendly, free applications. The Fing app is probably the easiest to use. https://www.fing.io/
|
|
|
|
fullzero (OP)
Legendary
Offline
Activity: 1260
Merit: 1009
|
|
August 20, 2017, 04:41:41 PM |
|
How to add Claymore 9.8 Im keep getting error with screen attach?
download 9.8, extract to ~/eth/9_8, replace 9_7 with 9_8 in 1bash. It is easier to extract only the executable over (replacing) the existing one in the 9_7 directory to avoid needing to change anything in 1bash.
|
|
|
|
fullzero (OP)
Legendary
Offline
Activity: 1260
Merit: 1009
|
|
August 20, 2017, 04:44:10 PM |
|
Thanks for posting this info - I banged my head against a wall for quite a while trying to figure this error out before giving up on it. This worked like a charm for me.
Though I think I'm still getting really low hashrates for skunk - whattomine thinks I should be getting 450mh/s with 5 1080 Tis and 7 1070s but I was only getting around 300mh/s. With the new krnlx miner I'm still only getting about 370Mh/s, which still seems pretty low. My miner also restarts with the watchdog reporting under utilization. My 1080 Tis are only running at about 60% and 1070s at 80-85% capacity. Can't figure out what's causing that.
EDIT: I added the '-i 25' flag to the command line and boosted core clock on the 1070s to +125 now I'm up at 480mh/s =)
Hum just stay away from krnlx release ... Low report share is a "know" problem ... I lost one week of mining on this stupid release ... Just dont USE krlnx this dev drop a brocken build and vanish ... First and last one i use something from him ... So a basic; krlnx mod boost your hashrate but you loose 30% expected share ... Use new ccminer release from tpru u get same hashrate and REAL SHARE. For 1070 u can push -i 25, Coreclock + 100/140 depend on card, MemoryClock -1500. I dont use 1080, but i bet u can push them to -i 26, to grab a solid 50mh/s+ per 1080ti https://github.com/tpruvot/ccminer/releases link to tpruvot version. Yes indeed... How do you lose 30% of expected shares? You mean rejected shares? Not rejected shares, is a weird thing and i dont analyze deeper on this. I get 70% of expected share with this hashrate report, so remove 30% of your hashrate when u use krnlx Where are you getting your expected shares tally from? I think that varies based on difficulty of the shares calculated which is determined by the stratum server, not your machine, so i don't know how you'd reliably determine expected shares in order to reason that you're losing 30%. Right now with 460-480Mh/s I'm getting about 22 coins an hour, which seems to be in line with the predictions of whattomine.com's mining calculator profit wise, ~ 525-550 coins a day @ $.07 per coin = ~$37-38.00 /day. I haven't mined at length with TPruvot ccminer 2.2 with the current configuration on my system, so I don't know if it would result in higher profit, but krnlx seems to be on point for profitability estimates now that I've got it properly configured and my clock settings are dialed in. FWIW, I'm not sure why you're downclocking your memory so much. I overclock mine +100 on 1070s and +800 on 1080 Tis. I see no benefit from underclocking memory on Nvidia cards really since power usage comes from core clock speed primarily, and it could result in unstable mining to underclock memory too much. EDIT: I had the under use problem with Watchdog as well, but I took care of that by raising my clock speed on my GPUs from +100 to +125 and setting intensity to -i 25, now I don't have watchdog restarts at all. I get same result on suprnova or pool.mn, for 2 days on each after i install krnlx, i lost 30% income and lost 30% on wathtomine prediction. I give a try to push my MemoryClock to +100 and i lost hashrate instant, prolly depend from the card. Maybe try to lower your MemoryClock u gona gain more hashrate. I use 8 x ASUS 1070 8OG, i have stable for more them 24h running and get 242+mh/s After reading this: I don't want to include the krnlx fork. Instead I will update the tpruvot to the newest version for v0019. Is this the optimal client right now for skunk? I ran the 2.2 version of TPruvot ccminer w/ skunk and got horrible hashrates. I will recompile and test it again, but for now, krnlx is the fastest for me. I'm not sure why there is a perceived drop of 30% in his setup, but I'm not seeing that at all. I'm mining exactly at the expected coin rate per day based on both whattomine.com and crypto-economy.com calculators (450mh/s nets me around 550 coins a day +/- 25). I will note that when I tried changing clock settings a little higher it really affected hashrates, dropping them over time considerably. There is definitely a 'sweet spot' clock rate for skunk, and deviation appears to create greatly reduced production. Also mining at -i 26 seems to have reduced my hashrates, so -i 25 is the proper setting for 1070 / 1080 Ti at least. Lower settings or leaving -i blank greatly reduce hashrates, as is to be expected. I'll update with the results of a recompile with tpruvot's most recent ccminer release on 0018. Thanks for keeping an eye on this. EDIT: the latest tpruvot ccminer release runs at just about the same hashrate as krnlx within 1-2mh/s (463mh/s vs. 462mh/s) and in a 2 hour test it mined 26-27 coins an hour on sigt.pool.mn which is a small number more than I was mining before, but it's a shjort test length so I'll update again after 24 hours to get a better average. I was able to run tpruvot with skunk from command line in guake, but it crashes for some reason when running from the script almost immediately after it starts mining (before watchdog kicks in). Not sure why that's happening, the command line is the same in both as I have it configured. In the end, I don't see a dramatic difference between the two miners' performance except that I know krnlx runs from the script properly, whereas TPccminer 2.2 does not work in the script yet. I'll keep playing around with it to see if some code in the script I added has an error. It's strange because the miner actually starts mining properly, but then errors out, rather than just not running at all which is normally how that goes. Are you using the newest watchdog; the one from the purple download link on the OP?
|
|
|
|
fullzero (OP)
Legendary
Offline
Activity: 1260
Merit: 1009
|
|
August 20, 2017, 04:57:01 PM |
|
Hi fullzero, Thank for your awesome distro. It help me alot!!!
If u dont mind, i think here are some improvements for your 1 bash file:
1. Use "eval" instead of very very long repeatedly redundancy source code, For example:
for j in $TI do for (( i=0; i<GPUS; i++ )) ; do eval \${NVD} -a [gpu:\${i}]/GPUGraphicsClockOffset[\${j}]=\${__CORE_OVERCLOCK_$i} eval \${NVD} -a [gpu:\${i}]/GPUMemoryTransferRateOffset[\${j}]=\${MEMORY_OVERCLOCK_$i} done done
or
if [[ $MANUAL_FAN == "YES" && $Maxximus007_AUTO_TEMPERATURE_CONTROL == "NO" ]] then for (( i=0; i<GPUS; i++ )) ; do eval \${NVD} -a [gpu:\${i}]/GPUFanControlState=1 eval \${NVD} -a [fan:\${i}]/GPUTargetFanSpeed=\${FAN_SPEED_$i} done fi
2. Move repeatedly source code to a function For example: function SLOW_USB_KEY_DELAY(){ if [ $SLOW_USB_KEY_MODE == "YES" ] then sleep 6 fi }
3. Use directly $CLAYMORE_VERSION to create claymore path instead of using if...fi For example: $CLAYMORE_VERSION="9_8" HCD="/home/m1/eth/$CLAYMORE_VERSION/ethdcrminer64"
Sorry for my bad English but hope that it will help u Cheers,
Thank you for input. I am, among other titles; a CS: so I am familiar with optimizing code. I will fully OO what is now called 1bash after I have implemented member requests. Using OO code that is functionally identical to the overly procedural code; is not a priority. It is also harder for a member who is not familiar with linux, let alone bash; to understand OO code. nvOC is not only a mining OS; it is also a method of exposing thousands of members to linux and teaching them some very basic bash scripting. In v0019 the implementation is separate from the user configuration / settings; so it will start to make more sense to optimize the code: as I expect most members will no longer interact with the implementation at this point.
|
|
|
|
|