Updated: Nvidia GeForce GTX 1070 8GB Pascal Review

Temperature And Noise Results

Temperatures

We start with the direct relationship between power consumption and temperature. Unlike its larger sibling, the Nvidia GeForce GTX 1070 Founders Edition doesn’t have a vapor chamber as part of its cooling solution due to cost-cutting measures. Unfortunately, this fact is reflected in our results. The smaller and more efficient graphics card ends up running at higher temperatures.

Both Founders Edition cards hit their thermal ceilings, capping maximum power consumption. This results in a fairly straight line that doesn't allow us to show each card's limit. We need another set of measurements to illustrate what’s going on.

The following graph shows the ratio of temperature and clock rate. It tells us two things. First, each Founders Edition card demonstrates a clear bend in its curve upon hitting the temperature limit. Second, we see what a good cooling solution that doesn’t hold back its graphics card looks like.

Noise

This section gets more complicated because all of our results were collected in a closed case so far. However, we can’t measure noise like that. So we moved the graphics cards to our audio test station after meticulously recording fan speeds during the preceding benchmarks. We then set each card's fan to the exact same RPM for our audio measurements. The voltage converter noise was below our lower threshold, so we could skip them without really affecting the outcomes.

The results for Nvidia's GeForce GTX 1080 Founders Edition are similar to, but a little different from our launch review. That's a result of higher temperatures in the closed PC case. Once again, we start with the ratio of noise and power consumption.

There’s a huge difference between cards with radial fans and those with several axial fans. Of course, each vendor's fan control curve affects the outcome as well.

MSI is on the conservative side. Its fans only really kick in under high power consumption, and our acoustic readings benefit as a result. Conversely, the two Founders Edition cards always make their presence known, and they can be made out clearly among the other components with high power consumption.

Once frequency is thrown into the mix, we see a similar picture as before. This includes the same bent curve where the Founders Edition cards run into their thermal limit. We wouldn’t call any of the boards in our comparison annoying-sounding, but MSI does a great job with its three contenders and is the clear winner.

So what does all of this data tell us? First, Pascal’s CUDA cores perform exactly like Maxwell’s. They’re just faster and more efficient. Consequently, the Pascal-based cards post better performance numbers at lower power consumption.

Second, overclocking only makes sense in small doses, if at all. The GeForce GTX 1080 is held back by its slow memory, whereas the 1070 turns out to have the higher memory throughput per shader. Still, both Pascal-based graphics cards are definitely worth a look, and the GeForce GTX 1070 may become a worthy successor to the popular GTX 970 once its prices fall to more reasonable levels and availability improves.

The GeForce GTX 1080 might end up sharing the 980’s fate: it’s just too expensive for the modest performance gain you get compared to GTX 1070. It looks more like a placeholder until the next Titan or GeForce GTX 1080 Ti emerges.

We’re looking forward to what else the add-in partners have in store for us. We’re certain that their R&D departments will find a way around Nvidia’s rules. After all, a full 16W at idle’ is ridiculously high in today’s market.

We received feedback and additional information since the original publication of this article. Unfortunately, the press samples and retail examples of MSI's GeForce GTX 1070 Gaming X 8G and GeForce GTX 1080 Gaming X 8G are slightly different. The press boards default to the OC preset, whereas the retail cards default to the gaming preset with a lower base and GPU Boost clock rate setting.

MSI confirmed all of this and pointed toward the included software. At 20 MHz, the base and GPU Boost frequency differences are not large, but pulling these kind of stunts isn’t fair to customers or the competition. For instance, Gigabyte uses the gaming preset and offers any additional overclocking via software.

MORE: Best Graphics Cards

MORE: All Graphics Content

Create a new thread in the Reviews comments forum about this subject
This thread is closed for comments
62 comments
Comment from the forums
    Your comment
  • gromann
    Fury X is going for $399-449 as of yesterday on Newegg.
  • nitrium
    So given the simultaneously lower price and higher performance of the partner boards, only an actual idiot would buy the "Founder's Edition" GTX 1070?
  • George Phillips
    I feel that I should regret getting MSI 1070 FE. MSI's custom designs perform superior then FE cards in every way. Very impressive. Asus and Gigabyte's custom designs must also do better than FE cards.
  • Krushe
    When you're talking about the heat on FE cards. I think the default fan speed is 45-50% at 83c. Make it 80% and the card never reaches 70c even with boost clock up to 1900+. What speeds are the MSI fans running at during your temp measurements?
  • DookieDraws
    Edit: The article has been updated, so I deleted my original comment about the MSI GPU.
  • Tony Casagrande
    "This means that the lowest possible GPU Boost clock rate step gets eliminated from the bottom of the BIOS’ table. So, if you want an additional space at the top, you need to make room for it by getting rid of the very bottom one."

    If it were me, I would have removed a low to middle clock rate instead of the very lowest to get both the low idle power consumption and the OC speed.
  • straatkat
    Still annoyed that they posted this review without including the 970 as a baseline.
  • neblogai
    Regarding the possible audible noise because of power spikes on PEG: it is not really about cheap MB, but about using analog audio out of MB, and not anything digital, right?
    Also, about overclocking: I think reviews of all these new generation nVidia and AMD cards should include average clock that cards operated when doing all game benchmarks. Official boost clock numbers are a bit useless, because AMD cards run games at below boost clocks, and average for nVidia GTX1070 is above boost clocks. Having just official boost clock numbers make it difficult to evaluate overclocking potential and make real gains look much bigger or smaller than expected.
  • Calculatron
    Am I the only one that noticed that the Founder's Edition cards managed to pull over 75 watts from the motherboard PCIe slot and that no one went bonkers over it?
  • TJ Hooker
    Anonymous said:
    Am I the only one that noticed that the Founder's Edition cards managed to pull over 75 watts from the motherboard PCIe slot and that no one went bonkers over it?
    It has narrow peaks that go over 75 W, as have a number of cards in the past. For the most part, what people were concerned about with the RX 480 was that average power was over 75 W. Whether that concern was warranted... well that's another question.
  • avatar_raq
    The last page needs to be updated as well.
    "Missing out on power consumption, operating temperatures and noise due to the constraints of Computex leaves us with an incomplete picture of GeForce GTX 1070, though we can certainly make some assumptions."
    You added those pages.
  • turkey3_scratch
    I don't see why people bought the Founder's Edition. This is just proof how much better aftermarket coolers are.
  • jtd871
    Looking forward to reviews of 3rd party cards from other vendors.
  • Calculatron
    Anonymous said:
    It has narrow peaks that go over 75 W, as have a number of cards in the past. For the most part, what people were concerned about with the RX 480 was that average power was over 75 W. Whether that concern was warranted... well that's another question.



    Yep, and that's why I am poking fun at it.
  • turkey3_scratch
    Yeah the same could be said about anything. An aftermarket GTX 1080 reaches 400W. But that's only a rare thin spike.
  • d_s_c_8
    TJ Hooker said:
    It has narrow peaks that go over 75 W, as have a number of cards in the past. For the most part, what people were concerned about with the RX 480 was that average power was over 75 W. Whether that concern was warranted... well that's another question.


    The GTX 1070 Founder Edition averages 75 watts but realistically has a 65 watt limit.
  • turkey3_scratch
    Anonymous said:
    TJ Hooker said:
    It has narrow peaks that go over 75 W, as have a number of cards in the past. For the most part, what people were concerned about with the RX 480 was that average power was over 75 W. Whether that concern was warranted... well that's another question.


    The GTX 1070 Founder Edition averages 75 watts but realistically has a 65 watt limit.


    That doesn't make any sense.
  • d_s_c_8
    Anonymous said:
    Anonymous said:
    TJ Hooker said:
    It has narrow peaks that go over 75 W, as have a number of cards in the past. For the most part, what people were concerned about with the RX 480 was that average power was over 75 W. Whether that concern was warranted... well that's another question.


    The GTX 1070 Founder Edition averages 75 watts but realistically has a 65 watt limit.


    That doesn't make any sense.


    From page 7 of the review:

    "Taking a closer look at the motherboard slot yields a surprising finding: none of the cards in this round-up use the 3V rail at all. This means that the PCIe slot doesn’t really provide the 75W most enthusiasts assume it does, since the 12V rail only offers about 65W on its own."
  • neblogai
    Anonymous said:


    "Taking a closer look at the motherboard slot yields a surprising finding: none of the cards in this round-up use the 3V rail at all. This means that the PCIe slot doesn’t really provide the 75W most enthusiasts assume it does, since the 12V rail only offers about 65W on its own."


    That is about the PCIe spec at 12V, not the power usage of cards in the test.