Two GTX 1070s mining on Nicehash. One is 22% faster. How?

[AtomicDog602]

EDIT: Solution found.  See my post about 7 messages down

OK this just bugs me and I need to ask.  I have 2 rigs that mine.  Most of the cards are gtx 1060s, but each rig has gotten a new gtx 1070.  An Asus 1070 something gaming card, the other A Zotac 1070 Amp edition.  Both are factory overclocked and boast almost identical specs.

Why would the Asus average 400 sols/s Equihash, and the Zotec only 325 Sols/s?

I mean that is a huge difference for a card that is running at the same same GPU clock, same mem clock.  The only difference is the zotec is connected using a PCI riser.  But I have never noticed them cause performance issues before.  Drivers up to date, both on gen 6 intel i series processors, systems are spec'd out similarly.

Thanks for your thoughts, because I am perplexed...

[joaocha]

Luck it is, you should check speed at miner, and try mining on flypool , it gives to you 24 average mining speed

[JaredKaragen]

OK this just bugs me and I need to ask.  I have 2 rigs that mine.  Most of the cards are gtx 1060s, but each rig has gotten a new gtx 1070.  An Asus 1070 something gaming card, the other A Zotac 1070 Amp edition.  Both are factory overclocked and boast almost identical specs.

Why would the Asus average 400 sols/s Equihash, and the Zotec only 325 Sols/s?

I mean that is a huge difference for a card that is running at the same same GPU clock, same mem clock.  The only difference is the zotec is connected using a PCI riser.  But I have never noticed them cause performance issues before.  Drivers up to date, both on gen 6 intel i series processors, systems are spec'd out similarly.

Thanks for your thoughts, because I am perplexed...

ASUS VS Zotac... I see a difference in quality already (IMHO).

Swap cards between systems and re-test....  see if they change places.

*edit* Temps could also play a key role in this as well;  one case does better at exhausting heat, etc.

[philipma1957]

how did you clock the gpus?

msi afterburner?

[AtomicDog602]

how did you clock the gpus?

msi afterburner?

If all else fails yea I will swap the cards and see.  I did use MSI afterburner to exactly match specifications, but left to its own the Zotac was almost identical in temp/clock/load already according to gpu-z and afterburner. (zotac was 1 degree cooler BUT only 67% TDP vs 92% for Asus)

I also swapped risers just in case.  No luck there either.  I have just enough room to actually install the zotac on the board, I might try that also.

But even if Zotac is a lesser brand (which honestly it feels really well made), all things being equal it should not be such a dog.  BTW I am using the same miner mining the same pool on both systems BTW, and the 1060s are running at the same speed on both.

I feel like I am missing something somehow, but... I dunno.

[JaredKaragen]

how did you clock the gpus?

msi afterburner?

If all else fails yea I will swap the cards and see.  I did use MSI afterburner to exactly match specifications, but left to its own the Zotac was almost identical in temp/clock/load already according to gpu-z and afterburner. (zotac was 1 degree cooler BUT only 67% TDP vs 92% for Asus)

I also swapped risers just in case.  No luck there either.  I have just enough room to actually install the zotac on the board, I might try that also.

But even if Zotac is a lesser brand (which honestly it feels really well made), all things being equal it should not be such a dog.  BTW I am using the same miner mining the same pool on both systems BTW, and the 1060s are running at the same speed on both.

I feel like I am missing something somehow, but... I dunno.

I believe it's GPUz that has the "ASIC quality" test?   That could also play a part... but 20% a large percentage.

Ill note on a miner that was too hot running mixed FE and aftermarket coolers;  the FE cards got lower hashrates... once cooling as adjusted they lifted back up again...

so its seeming like heat is your issue.... you gotta ventilate them pretty well.

[AtomicDog602]

69 degrees is the temp.  I also checked for throttling, but it isn't.  On the surface everything looks good and similar except for the stark contrast in TDP (67% vs 92%)

***UPDATE*** So I fiddled with all sorts of things. No change.  But I recall reading once that Nicehash suggests creating a very large Virtual memory page file.  Not sure why, and I never had an issue with 3 GTX1060s and 2 GTX 1050s.  But when I created a huge 24GB page file suddenly my hash rate jumped up to 390 sols/s.  (BTW ONLY the algorithms equihash and DaggerHashimodo were impacted all others were expected for a 1070)

So Why Nicehash needed so much extra virt mem for a 1070 is unknown to me.  Why low virt mem only impacted DaggerHashimodo and equihash is even more perplexing. I am only thankful I remembered something about that and tried it.

Thank you to everybody that replied with suggestions!