MSI GTX 1070TI Ethereum -> Nanopool problems

[ChrisR1]

Hi,

I'm mining since July 2017 with some GTX 1070 cars. All worked perfect.

Now I have for testing here a MSI GTX 1070 TI GAMING  X card.

I testet it on
ethminer (latest version)
Claymore 10.2
with latest nvidia drivers.
https://imgur.com/a/QLL97

see image here: https://imgur.com/a/QLL97

before the black reported hasrate-line I mined with ethminer (often some high peaks but terrible low zero points) nanopool reportet 8MH/s.
After the first high peak hashrate dropped to zero than later the second peak, but why such immense huge difference???
Ethminer and Claymore allway reported absolutely constant ...32MH/s.
Later I tested with claymore. But also with claymore very low fall points. 

I'm wondering what to do or is the claymore and ethiner -since yet- not optimized for gtx1070TI.

Ethminer reportet always hashrate between 32..33MH
with claymore always 32.8MH absolutely constant.

Why do the nanopool such a unstable hashrate-graph?

Afterburner
PWR 60percent
mem + 800 MHz
Core - 100 MHz
Temps are very low very good airflow about 42°C.
only one card @Z170 carbon mainboard.

Are there special tweaks for ethminer (working size cuda ??) or @claymore for optimizing gtx1070TI?

Do anybody can send me PM for contact ethminer-developer?

[Dannyh2]

I would start with lower your overclocks...

[ChrisR1]

I do not thinks  that is caused by an overlcocking?

[Dannyh2]

It does.
If your cards get memory errors and fail to deliver valid shares you will end up with lower share rates and thus lower rewards.

[ChrisR1]

Ok.

I downloaded hwinfo (lastest beta version). Problem is there is no row for MEMORY ERROR. I think hwinfo do not supply or cannot read gtx1070TI errors?

If it is true what you said, why do claymore show 100% accepted shares no share is wrong?
Or do nanopool internally check the share and mark it as not valid?

[gt_addict]

You cant run a 1070ti on 60% Power limit. It will struggle. Your clocks are high but feasible. Try increasing your power to 65-70% and see how you get on.

[worldofcrypto]

Why don't you try troubleshooting step by step first before jumping to conclusions nanopool doesn't work?

Try
1. Less aggressive clock speeds
2. Less aggressive power limint in AB

Do it bit by bit to test for stability. It took me 3 days minimum  to get system stable

[ChrisR1]

Hi guys,

sorry.

first:
- I think I have much experience in overclocking GTX1070.
I have all my rigs with PWR 50% running Zotac GTX1070 AMP XTREME (expensive card but best cooling and best untervolting).

So now after Blackfriday with new MSI GTX1070TI GAMING X.

- very flucutating peaks and zero points.
- wondering same at CLAYMORE and ethminer.
- all share were accepted. no memory errors!!!

@around 20 (8pm.) and tuned parameter ethminer --farm recheck 3800 (see imagelink magic is is table graph now!!! wow!)

https://imgur.com/a/T1P2W
https://imgur.com/a/T1P2W
(before I did tests without overclocking with claymore 10.2 and ethminer no chance. everytime the same shit!!)

parameter @8pm.:

PWR 60percent
GPU CORE -100MHz
MEM + 600MHz
ethminer --farm recheck 3800

3800 seemed like a perfect solution for GTX1070TI (the TI has more shadercores ?? maybe there is a juge difference than the 1070, but hashrates seemed like the same as 1070).
I think if the farm recheck rate (default I think 2000 without parameter @ethminer) is very bad for 1070TI.

.... now I do some overclocking...

by the way

I have here also for test a Zotac 1070TI AMP Xtreme.
Card is huge but performance (overclocking is very poor). MSI 1070TI Gaming X is the best on!! (very good undervolting PWR60 -> around 100W xtreme good overclocking stable +800MHz mem). Hashrate about 33MH/s.

[ChrisR1]

Update sorry false alarm 

Rig with 8 GTX 1070TI is stable, that is normal that nanopool has high and very low peaks in the history.
Also mit GTX1070 rigs have it in the graph-history.

...all ok now... (there was no fault!)