Asus 670 DirectCU II TOP - Page 101

Has anyone posted over at Asus ROG forums about these issues?

Exactly.

The simple fact of the matter is that every 670/680 has a Boost Clock, and it has SOME SPECIFIC amount over the Boost Clock that will always be added to that Boost Clock as long as downclocking does not occur ... which when it happens will be due to one or more of the following:
1) Excessive power consumption,
2) Excessive temps,
3) Low GPU Usage load (in which case the card won't run full-bore in order to save power)

This additional clock amount (the delta between 'boost clock' and 'actual clock' that occurs due to nV 'dynamic overclocking' technology) varies from sample to sample but each sample HAS such a value (although I've seen 1 report of a card where that amount appeared to be 0), and it's a reliable quantity on everyone's cards.

On Sean's, that value is 104MHz, on my card, that value is 130MHz (1059 boost clock = 1189 actual operating clock, 1207 boost clock = 1337 actual operating clock, etc, etc, etc). I can do the same types of tests Sean showed screenies of, and show a consistent 130MHz delta just like his card does with 104MHz delta. And so could everyone else who has a Kepler card ... the only thing that varies is: how much is that Delta?

Because there's really no official/accepted term for this Delta Value, and because there's a need to discuss this value amongst ourselves, I coined the term 'Kepler Boost' and have been promoting it's adoption here on the boards.

Sure, it's 'made up' as a name, but it is a real thing, we have need to discuss it, and most importantly, the phenomenon has no 'real name' ... so why not 'Kepler Boost'?

So I should maximize power target to minimize downclocking?

I totally agree, the KB is fixed boost over factory sealed boost clock. I did around 60 heaven runs on different clock settings + memory variations and my +boost on top of 980MHz boost is always constant. I have KB of 137MHz on top, whatever clock i run. It simply stays constant.

Wandering about the same thing. I don't have my card yet and won't do it if I do not know whether or not it will be faulty (should arrive this week). But is anyone with a defective card attempting to squeeze some official response out of Asus?

In theory, yes.

TBH though there have been a few scattered reports of people getting some occasional minor downclocking (meaning seeing clocks dropping slightly below Boost Clock + their Kepler Boost) that have an unknown cause ... even though their card is running well within the power % specified by the slider, temps are in check, and full load is present. So there appears to be, in some isolated cases, an X factor coming into to play that can cause minor downclocking for reasons that cannot quite be explained as of yet, and we've not determined a reliable way to make that stop for those people.

But, again, in theory, raising the power limit to the max should help reduce downclocking, yes.

If not just a temporary hiccup or an internet issue, might've been overheating? Probably not the core, but the VRMs.

Were you playing Diablo 3?

I just tried and maxed out power limit to 117% and voila, no more downclocking even in Heaven, it stays at constant 1280Mhz boost clock throughout entire benchmark. My score increased significantly! Sweet. I love my TOP card even more now