GTX 1060 hashrate tweaking problem

Powningcabbage

Hello! I have a rig of 6 MSI GTX 1060 GAMING X+ 6G. Windows 10 and NVIDIA latest updates. Using Claymore miner and Ethermine as pool.

I can't get over the 20 MH/s per card. They where 16-17 MH/s at stock, I have overclocked them with MSI afterburner; 900+ mem and 100+ core. I have searched a lot on the internet about this card and the overclock capability's. Many people get 22-25 MH/s with a GTX 1060 with the same overclock, but I'm stuck at 20 MH/s What am I doing wrong?

- MSI Z170A GAMING M5
- Intel Pentium G4400
- 6x USB powered risers
- Crucial DDR4 1x4GB 2133 CL15
- Corsair Force LS V2 Series SSD 60GB
- 6x MSI GeForce GTX 1060 GAMING X 6G
- be quiet! Dark Power Pro 11 1200W

All cards have the Samsung memory

Thanks!

mrmine

I see you have a 60gb ssd , is your page file 16gb or more, claymore specifies your page file needs to be that large for multiple cards, also that over clock seems kind of high on your memory, those cards already have an effective frequency of 9000mhz, now in afterburner for me it shows my memory clock as a half value, so if you were at 4500 added 900 that puts you at 5400 or 10800, that seems like a lot for the memory on those cards, i know of a 1070 that was oc'd up to 650(4000+650) but ussually left around 500-550 that burned out its v-ram, now that had micron memory, but still

mrmine

I am currently running a 1060 +150 core +200 memory and getting 19.6, and my memory is the 8000mhz stock, while yours with the gaming x plus is 9000mhz stock

KingWizard

Changing ssd won't help as mentioned before.
Have exactly the same setup, MSI Z270 motherboard and 5x GTX 1060 Gaming X+.
Tried anything that I could and finished out switching to mine zcash.

Element

I have a bunch of the MSI GTX 1060 Gaming X+ cards, and they are all crap. I'm getting way better from stocks settings on older EVGA cards. I'm really pissed, I think they're all going to have to get returned... I wish there was a way to force certain CUDA memory usage parameters in the miners to try SOMETHING. These 9GBps cards from MSI, so far, are 15% less pwoerful than 8GBps cards. WTF?

DjithBit

Im getting the geforce gtx 1060 armor 6gb OC V1 .. they are 8GBps .. Picking them finaly up next week!

Rig will be:
Win10 Pro 64bit
ASRock Z97 anniversary
8Gb DDR3
Pentium G3260
SSD samsung 120Gb
Corsair PSU 650W
2x Geforce GTX 1060 6Gb Armor OC V1

More GPU will added later and also PSU then.. but first getting this setup.

MrN1ce9uy

mrmine said:

I see you have a 60gb ssd , is your page file 16gb or more, claymore specifies your page file needs to be that large for multiple cards, also that over clock seems kind of high on your memory, those cards already have an effective frequency of 9000mhz, now in afterburner for me it shows my memory clock as a half value, so if you were at 4500 added 900 that puts you at 5400 or 10800, that seems like a lot for the memory on those cards, i know of a 1070 that was oc'd up to 650(4000+650) but ussually left around 500-550 that burned out its v-ram, now that had micron memory, but still

What burned out the vram on the 1070? +650 or +500-550?

xMines

what powerconsumption do you guys have?

ecs87

mrmine said:

I see you have a 60gb ssd , is your page file 16gb or more, claymore specifies your page file needs to be that large for multiple cards, also that over clock seems kind of high on your memory, those cards already have an effective frequency of 9000mhz, now in afterburner for me it shows my memory clock as a half value, so if you were at 4500 added 900 that puts you at 5400 or 10800, that seems like a lot for the memory on those cards, i know of a 1070 that was oc'd up to 650(4000+650) but ussually left around 500-550 that burned out its v-ram, now that had micron memory, but still

I assume the damaged card was generating lots of incorrect shares? I thought I heard a theory about massive amounts of memory errors can lead to premature card failure.

MrN1ce9uy

ecs87 said:

mrmine said:

I see you have a 60gb ssd , is your page file 16gb or more, claymore specifies your page file needs to be that large for multiple cards, also that over clock seems kind of high on your memory, those cards already have an effective frequency of 9000mhz, now in afterburner for me it shows my memory clock as a half value, so if you were at 4500 added 900 that puts you at 5400 or 10800, that seems like a lot for the memory on those cards, i know of a 1070 that was oc'd up to 650(4000+650) but ussually left around 500-550 that burned out its v-ram, now that had micron memory, but still

I assume the damaged card was generating lots of incorrect shares? I thought I heard a theory about massive amounts of memory errors can lead to premature card failure.

But how do you know if an Nvidia card is having memory errors? It doesn't show in HWInfo. I'm asking becaue I have a 1070 and it will run at over +600 but I keep it at +500 for longevity.

zuzux

i have amd 480 and i add 1060 to same system ,cuda 8 i install 1st and last driver ,and on claymore amd+nvidia i get same 28+dec results but 1060 add only 3mh and 50or so dec
what i didi wrong ?

ecs87

MrN1ce9uy said:

ecs87 said:

mrmine said:

I see you have a 60gb ssd , is your page file 16gb or more, claymore specifies your page file needs to be that large for multiple cards, also that over clock seems kind of high on your memory, those cards already have an effective frequency of 9000mhz, now in afterburner for me it shows my memory clock as a half value, so if you were at 4500 added 900 that puts you at 5400 or 10800, that seems like a lot for the memory on those cards, i know of a 1070 that was oc'd up to 650(4000+650) but ussually left around 500-550 that burned out its v-ram, now that had micron memory, but still

I assume the damaged card was generating lots of incorrect shares? I thought I heard a theory about massive amounts of memory errors can lead to premature card failure.

But how do you know if an Nvidia card is having memory errors? It doesn't show in HWInfo. I'm asking becaue I have a 1070 and it will run at over +600 but I keep it at +500 for longevity.

I'd measure it through incorrect shares. If you get 0 (I try to measure this over 48-72 hours) then most likely your memory error rate is tolerable.

DjithBit

Set oc at +850 on my gtx 1060 .. they do 23Mh/s .. not running hot and no rejected shares.. samsung memory they have.

hknoener

I've noticed that I don't have any rejected/stale shares when OC, but my average is going down nevertheless. I'm sitting at 70%/Core +175/Mem +375. It is stable up to +450, but the average will go lower for some reason. This is the best result strangely.

MLN84

Keep in mind that the difficulty is rising quickly. My average hashrate and shares are also decreasing slowly over the past days. It doesn't neccessarily mean that your hardware has problems.

alnasim

mrmine said:

I see you have a 60gb ssd , is your page file 16gb or more, claymore specifies your page file needs to be that large for multiple cards, also that over clock seems kind of high on your memory, those cards already have an effective frequency of 9000mhz, now in afterburner for me it shows my memory clock as a half value, so if you were at 4500 added 900 that puts you at 5400 or 10800, that seems like a lot for the memory on those cards, i know of a 1070 that was oc'd up to 650(4000+650) but ussually left around 500-550 that burned out its v-ram, now that had micron memory, but still

Ethereum Free Instantly to your ETH wallet ID Register the following weblink to credit https://keepstake.io/join?kid=SVYTD also you will receive 0.1 ETH free for every additional referrer. Do not miss this golden opportunity guys.