Pages:
Author

Topic: 1060 6 GPU rig, need help (Read 913 times)

full member
Activity: 210
Merit: 100
October 13, 2017, 11:53:03 AM
#26
So hadn't much time to get into it last few days.

I tried mining Zcash getting more or less 290 sols/card without any tweaking and it's stable. I Let it run for a day and not a single problem.
when I tried eth and hashrate was dropping and finaly crashing, i was at PL 70 core -200, -100, +100 (tried a few to see if it changed anything, answer is not from what i saw) and mem+500.
I tried
+100 +800 = crash
-200 +600 = crash
anytime i try to go above +500, it seems it's not stable and eventualy crashes (real pain in the ass, btw)

Since i got no problem whatsoever with PSU when running Zcash, i guess it must be the GPU/drivers that are the issue?
Even with hynix mem i should be able to OC them to get 22Mh on eth i think.

I had the same issue, crashing because of nvlddmkm after mining for a few hours. the nvOC linux helped with that issue for me
newbie
Activity: 24
Merit: 0
October 13, 2017, 09:12:47 AM
#25
So hadn't much time to get into it last few days.

I tried mining Zcash getting more or less 290 sols/card without any tweaking and it's stable. I Let it run for a day and not a single problem.
when I tried eth and hashrate was dropping and finaly crashing, i was at PL 70 core -200, -100, +100 (tried a few to see if it changed anything, answer is not from what i saw) and mem+500.
I tried
+100 +800 = crash
-200 +600 = crash
anytime i try to go above +500, it seems it's not stable and eventualy crashes (real pain in the ass, btw)

Since i got no problem whatsoever with PSU when running Zcash, i guess it must be the GPU/drivers that are the issue?
Even with hynix mem i should be able to OC them to get 22Mh on eth i think.
newbie
Activity: 24
Merit: 0
October 11, 2017, 08:53:14 AM
#24
If you have tested cards one by one without any problem, then try to swap riser and slot, and you still having same problem with same slot different card, your problem could be PSU molex issue, although you have enough power in total, but each single molex has maximum power output only.

I don't think it's the issue here. I got 3 cables powering 6 risers. so 2 risers for 1 cable i did that because i read somewhere not to power more than 2 risers if you don't want it to crash.

The errors i get are drivers ones.

here is the problem:

when i let the rig on and mining, at one point i get this :

http://imageshack.com/a/img924/8324/GOew4p.jpg

and/or this :

http://imageshack.com/a/img922/1346/KgYv9J.png

It's hard to screen because when miner crashes, it's laggy as fuck. It takes me like 3 to 5 minutes just to save the screenshot... next time i'm gonna use my phone and take pictures of my screen like a fucking retard Cheesy, way faster than waiting for nothing.

what is the temperature of the room?

it's about 20°C, nothing too hot. Rig is in an open frame i did myself. Cards are between 60-65°. I'm gonna place 6 80mm fans i got when the rig will be set and good to work. I also plan on adding a 7th card, a R9 290 i got earlier, got a space for it but i wanna be sure PSU can handle it so i didn't placed it yet (and i also need to buy PCIE extender slots and a PSU cable cause it's a 8+6 pins).

here's a picture of my rig:

http://imageshack.com/a/img922/16/nprG5Z.jpg
member
Activity: 97
Merit: 10
October 11, 2017, 05:56:13 AM
#23
what is the temperature of the room?
jr. member
Activity: 183
Merit: 1
October 11, 2017, 03:24:46 AM
#22
I have GB M57SLI-s4 mobo with 2xpcix16 3xpcix1 with an old amd phenom cpu and 5x1063 works perfectly on this old platform
full member
Activity: 224
Merit: 100
IDEX - LIVE Real-time DEX
October 10, 2017, 10:18:41 PM
#21
If you have tested cards one by one without any problem, then try to swap riser and slot, and you still having same problem with same slot different card, your problem could be PSU molex issue, although you have enough power in total, but each single molex has maximum power output only.
newbie
Activity: 24
Merit: 0
October 10, 2017, 09:45:21 PM
#20
And i forgot to mention, did a test on Zcash, getting 1700-1750 sols out of the box without any OC.
newbie
Activity: 24
Merit: 0
October 10, 2017, 09:42:26 PM
#19
4G is enough but with +500 mem oc 18.8 is very small hashrate. try predefined mining OS as smos or ethos. etherminer is very effective for nvidia cards.

Yeah i know, i don't understand because my 1060 6G kfa2 is working fine on windows 10.

I installed the same drivers, same miner, and i'm getting different results. I know that hynix mem will give a smaller hashrate. But for some reason, i can't seem to OC these cards like i do with the card on my computer.

I'll be looking into different drivers i think. to try and get a more stable hashrate.

I sometimes get error drivers and the miner crashes. i'm gonna need to take a look at that aswell.

meaning 4G
you installed a new BIOS?

No, i didn't have to do that. i'm running on Bios 1.2 and it seems to be ok.
member
Activity: 77
Merit: 10
October 10, 2017, 02:37:10 PM
#18
meaning 4G
you installed a new BIOS?

sr. member
Activity: 602
Merit: 255
CryptoTalk.Org - Get Paid for every Post!
October 10, 2017, 01:33:52 PM
#17
4G is enough but with +500 mem oc 18.8 is very small hashrate. try predefined mining OS as smos or ethos. etherminer is very effective for nvidia cards.
newbie
Activity: 24
Merit: 0
October 10, 2017, 01:23:09 PM
#16
yes it's frustrating lol

Now i'm running 6 cards at core -200 mem +500 and i got 18,8 Mh/s per card, it's low but i'm leaving it this way to see if it's holding steady ^^

if i try to get mem higher, afterburner is kinda freezing and i got errors in the miner.

i still have to optimize that, it's weird because my rig has the same driver (382.53) than my main computer where i have a KFA2 1060 6G (samsung memory core-200 mem +850) and i get 24 Mh/s with ethminer 0.9.41 genoil.

My rig is 6x Gigabyte GeForce GTX 1060 WINDFORCE OC 6G (hynix memory  Sad )

On afterburner, if i try to push it to let's say mem 700 or mem 800, it crashes when miner start to work and when i close the miner and go back in afterburner, he's freezing as fuck and the hole computer too.

once i got a memory error but on the computer, meaning 4G is not enough can it be the problem here?



full member
Activity: 210
Merit: 100
October 10, 2017, 12:40:06 PM
#15
ah yes! I was getting a strange error when i was running IGD as the main display.. it would work great for 4 cards but the moment I put 5 in, everything went to shit eventually I used the PEG x16 as the display and everything worked again lol. glad you're online.. shit is frustrating, I spent a week trying to diagnose a previous problem I had with a MSI board.. was convinced it was everything except the board but it was 2 of the PCIe slots that were giving me issues, so went with an ASRock board and everything was better.. was going to sell it but now I just run it with 4 cards lol..
newbie
Activity: 24
Merit: 0
October 10, 2017, 10:34:49 AM
#14
I had all sorts of issues on Win10, switched over to fullzero's nvOC and its working like a charm. super easy to use and supports the nvidia 10 series. If you have an extra 16gb thumbdrive you can image it over and boot it up off the thumbdrive and see if it works without formatting your Win10 off whatever you have installed on

Are you using a prewritten .bat file, if so check the cuda_devices and make sure it has 0 1 2 3 4 5

Also, this might be a stupid question as i doubt it'll let you boot without it after 3 cards but worth checking all possible problems - Did you plug in all the molex power cables to the board terminals?

in the past i had to ensure 4g encoding was enabled and then play with the PCIe settings (trial and error with Gen1 settings, Gen2 settings, etc).

I solved the problem by simply pluggind my screen on the card that was on the x16 slot.
all 6 are mining now.

In bios i'm on Gen 1. doesn't have the option 4G but since it's a pro btc mobo, i read somewhere it runs it already.

Now i have to try some drivers to find the best hashrate and i'll be good to go i think Smiley
full member
Activity: 210
Merit: 100
October 10, 2017, 09:52:23 AM
#13
I had all sorts of issues on Win10, switched over to fullzero's nvOC and its working like a charm. super easy to use and supports the nvidia 10 series. If you have an extra 16gb thumbdrive you can image it over and boot it up off the thumbdrive and see if it works without formatting your Win10 off whatever you have installed on

Are you using a prewritten .bat file, if so check the cuda_devices and make sure it has 0 1 2 3 4 5

Also, this might be a stupid question as i doubt it'll let you boot without it after 3 cards but worth checking all possible problems - Did you plug in all the molex power cables to the board terminals?

in the past i had to ensure 4g encoding was enabled and then play with the PCIe settings (trial and error with Gen1 settings, Gen2 settings, etc).
newbie
Activity: 24
Merit: 0
October 10, 2017, 09:37:22 AM
#12
Update, I got 6 cards showing.

I unplugged all PCI 1 slots and plugged it again, it's now showing right. guess is one of the riser and trouble connecting, (weird because i checked them all one by one and they were showing right, anyway i'm glad i can moove on)

In the meantime, i encounter another issue.

when i want to launch the miner, i got a Error 0xc00007b.

I had first a MSVCP140.dll missing. i looked it up and i downloaded it and placed it where it's supposed to be,
then i got a VCRUNTIME140.dll error. same here, i looked it up and corrected the error by dowloading and placing the library where it's supposed to be.
and now, i got the error 0xc00007b.

As i was typing this, i found the wonder link : https://support.microsoft.com/en-us/help/2977003/the-latest-supported-visual-c-downloads

Installed 2015 and 2017 and it was fixed!

Now, another issue appears :

5 out of 6 cards are actually mining. I 've got only 15.5Mh/s out of the box on 5 cards (hynix memory and i will have to find other drivers i think)

I've made some screenshots :

http://imageshack.com/a/img922/1956/FUuHwj.jpg
http://imageshack.com/a/img923/8353/8qyiKl.jpg
http://imageshack.com/a/img923/3284/7oshp6.jpg

As you can see, miner shows only 5 cards mining. Afterburner doens't show card 1 stats. the card that is not working is on the PCIE x16 slot.
any idea on what to do?
newbie
Activity: 24
Merit: 0
October 10, 2017, 08:17:49 AM
#11
re-check all your PSU cables...every time I have a card that won't work it's related to the cables and connectors

PSU cables are plugged in, again, i tested individualy each card and each of the 6 cards works fine. so that's not the issue.

Try looking for a "Enable above 4G decoding" or a similar option in the BIOS. If you can't enable it/can't find it/still does not work it is probably Windows 10 issue (if you are using it) or driver issues.
Yeah i forgot to mention, Windows 10 ^^

I didn't find enable 4G in bios, all i found was Gen 1 or Gen 2 in some options but i don't know what it means.

You might want to turn the PCIe slots to Gen 1 and test the risers individually. If all of them work then I think it's a software issue.

all risers are working fine since i tested every one when i plugged each card separately from the rest.
In my opinion, it's not an hardware problem, more a software one Wink

i'm gonna try some drivers that citronik gave me on another topic, will see if it works.
member
Activity: 66
Merit: 10
October 10, 2017, 03:57:47 AM
#10
Try looking for a "Enable above 4G decoding" or a similar option in the BIOS. If you can't enable it/can't find it/still does not work it is probably Windows 10 issue (if you are using it) or driver issues.
Yeah i forgot to mention, Windows 10 ^^

I didn't find enable 4G in bios, all i found was Gen 1 or Gen 2 in some options but i don't know what it means.

You might want to turn the PCIe slots to Gen 1 and test the risers individually. If all of them work then I think it's a software issue.
sr. member
Activity: 349
Merit: 250
October 09, 2017, 08:23:47 PM
#9
re-check all your PSU cables...every time I have a card that won't work it's related to the cables and connectors
newbie
Activity: 24
Merit: 0
October 08, 2017, 11:12:16 AM
#8
Try looking for a "Enable above 4G decoding" or a similar option in the BIOS. If you can't enable it/can't find it/still does not work it is probably Windows 10 issue (if you are using it) or driver issues.
Yeah i forgot to mention, Windows 10 ^^

I didn't find enable 4G in bios, all i found was Gen 1 or Gen 2 in some options but i don't know what it means.

Have you identified what PCIE slot is not detected? I can possibly see that one slot is faulty, check also your PSU running 6GPUS might be in power shortage for a 1000w powersupply.
each slot works fine, as i said, i tested them all. PSU is fine, 1000W is more than enough for 6 1060 cards so it must be something else...
sr. member
Activity: 1008
Merit: 297
Grow with community
October 08, 2017, 10:19:31 AM
#7
Try looking for a "Enable above 4G decoding" or a similar option in the BIOS. If you can't enable it/can't find it/still does not work it is probably Windows 10 issue (if you are using it) or driver issues.
Yeah i forgot to mention, Windows 10 ^^

I didn't find enable 4G in bios, all i found was Gen 1 or Gen 2 in some options but i don't know what it means.

Have you identified what PCIE slot is not detected? I can possibly see that one slot is faulty, check also your PSU running 6GPUS might be in power shortage for a 1000w powersupply.
Pages:
Jump to: