GTX 1070 Boost clock

[Ozbro89]

Hi guys,

I am interested in the GTX 1070. However, I am not sure whether to go for EVGA or MSI.
Now, I already saw a few topics discussing this issue and most of the people tend to go for the cheaper one. In the Netherlands the EVGA GeForce GTX 1070 FTW GAMING ACX 3.0 is 487 euro and the MSI GeForce GTX 1070 Gaming X is 527 euro. Which is a difference of 40 euro's.
However, I stumbled across a Dutch review site who was discussing the turbotechnology of these cards. The question they asked is: "Can the card maintain a steady frequency over time and temperature?"

This lead to the following results (Boost clock on vertical axe and duration on horizontal axe);

I was wondering how much of a difference it would make. Since this would mean that the FPS consistency of the EVGA is worse over time than the MSI since the frequency fluctuates more. This would mean that the MSI is the better choice even with the price difference.

I am wondering what you guys think.

Thanks in advance!

Methodology of the test: They ran a benchmark in F1 2015 seventeen times, which lasted aprox 30 minutes. They kept a log which saved the frequency and temperature every second. This would lead to the graphs shown above. Goal is to see whether a longer gaming session would impact the performance of the GPU.

 

[Kenmitch]

The ability to maintain the max boost clock is really temperature dependent in the end. Good case airflow and ambient temps will be the biggest factor in deciding how long they can maintain the max boost clocks. Not really sure as it's not stated in the methodology about the fan speed being a custom profile or the stock settings of the card. Using a custom fan profile will help eliminate heat soak and allow one to maintain max clocks for a longer period of time. With my GTX 1070 Amp it takes around 70% or so before I even hear the card over my case fans and H100i's fans.

 

[Ozbro89]

Thank you for your reply!

So you are saying that the MSI is better straight out the box. I cannot find anything about fanspeeds. So I assume that they are stockspeed.

These pictures might contain a bit more information (left side is temperature):

So the question still remains. Is the 40 euro extra worth it?

 

[Kenmitch]

FTW and use a custom fan profile is what I'd do.

 

[reb0rn]

Its true thet MSI gaming X will boost more, its not only temp, its more voltage
MSI gaming has 1x 8pin + 6pin power connector
Sad think is my MSI gaming x just blow dies some 2 days ago, I was clocking it at 2Ghz with 85% power limit, it was mining and just dies with a bang

 

[flexy]

"Can the card maintain a steady frequency over time and temperature?"

This is really a "can of worms" subject since Nvidia cards, for a long time, ever since Kepler at least are not even DESIGNED to "maintain a steady frequency". The entire GPU Boost technology is based on dynamic switching of clocks, based on VARIOUS factors like load, temps etc..etc..

The myth here is that people see a fluctuation of say, one or two boost "bins" (which is 12.5Mhz per bin) up or down (like in your graphs) and assume it affects performance. It doesn't. You won't notice even if your card switches between, say, 1750 and 1850 each or so second. This and the entire article you linked seems bunk to me...unless those writers can actually show that the (normal) fluctuation of boost bins leads to "FPS inconsistency". (If this was the case, do you think Nvidia would even have implemented GPU boost?)

The alleged "inconsistency" you show there, to me might simply from the fan RPM/temp, that the EVGA has the card just right at some temperature threshold so it bounces around with boost a bit...but as said..IN REALITY a total non-issue.

(An "issue" would be if the card would actually be throttling, not the normal operating of GPU boost which is seen in the graph)

Edit: If I see anything in the graphs it's that the EVGA has higher temps, which may explain the bouncing and boosting down...but to me its meaningless since the graphs don't show whether the EVGA has room for increasing the FAN speed (with a custom curve) without getting too loud. Otherwise you would just make a custom fan curve, set the fan speed higher and get the same lower temps as the MSI. And then LIKELY the two cards would behave the same, the EVGA might even be better, who knows.