EVGA GTX 1080 FTW Hybrid Review vs. MSI Sea Hawk X

By Published August 30, 2016 at 6:23 pm

Additional Info

  • Awards: Editor's Choice
  • Original MSRP: 730
  • Manufacturer: EVGA

 

EVGA GTX 1080 FTW Hybrid Thermal Benchmark vs. GN Hybrid, Sea Hawk

Our thermal tests are presented using two different primary values: Delta T over ambient and the GPU diode temperature. For the peak average thermal tests, we're showing a Delta T value to subtract out potentially significant fluctuations in ambient temperature. For the endurance tests, we use the GPU Diode temperature to better demonstrate potential thermal throttle points or hitches in frequency performance, as correlated to GPU diode temperature.

Two K-type thermocouples are deployed on our thermocouple reader: One at the fan intake, about 2mm away from the radiator fan, and one that's ~8” above the test bench to collect room ambient. We use a spreadsheet of our own design to perform necessary calculations, resulting in the below charts.

evga-1080-ftw-hybrid-temp-eq

From our automated thermal tests with the stock cards, we're seeing a performance output that lands the EVGA 1080 FTW Hybrid at 29.61C delta T over ambient – which isn't only not that good for liquid, but also far from what we expected. The Corsair Hydro GFX / MSI Sea Hawk X – which theoretically uses a worse cooler, but does use a better fan and run a lower voltage – is at 18.5C delta T. That's 11C cooler than EVGA's card.

This is what prompted all the tests we ran shortly after, including a swap of the fans, increasing the Corsair vCore to match EVGA's, and even requesting a second Hybrid to validate that we didn't have a leaky card. Finally, after we decided to film a tear-down video, EVGA's engineering started to make sense. First, vs. time charts.

EVGA GTX 1080 FTW Hybrid Thermals over Time

This is the same data as used to generate the above charts, but plotted against a Y-axis time to demonstrate “soak” period for the cooling solutions:

evga-1080-ftw-hybrid-temp-time

EVGA & Sea Hawk X Controlled Thermal Tests

evga-1080-ftw-hybrid-temp-controls

These tests demonstrate some controls we put in place to determine who makes the objectively superior fan, objectively superior liquid cooler (though both are made by Asetek, they have modifications), and how voltage impacts temperatures. Note also that we ran one stock EVGA FTW Hybrid test with the VRAM plate removed, so we could strictly cool the GPU and look at a more direct comparison between the models. This is what revealed to us just how effective that VRAM cooling is, as it's driving up apparent thermals of the GPU Diode by ~7C (dropped to ~23C after removing the VRAM cooling).

The reason for this is straight-forward: The CLC only has so much dissipation potential, and is now being made to share its cooling abilities between both the GPU (direct contact + internal microfins) and the 8x GDDR5X Micron VRAM modules. GDDR5X runs lower voltage than GDDR5 and doesn't get as hot, but still generates heat. Because the VRAM is sinked by a copper plate, and because that plate communicates through TIM to the coldplate, the CLC's liquid warms up and reaches a higher “equilibrium” value than found when strictly cooling the GPU.

From these results, ignoring the FTW and using only the Sea Hawk as a baseline, it's clear that EVGA is using Asetek's superior coldplate for GPU cooling; the copper protrusion moves the bottom line for thermal “saturation” of the solution prior to its stabilization at the resting load temperature. Still, that difference is only within the ~1C range against Corsair's H55 CLC that has flattened the coldplate for this generation of GPU cooling. Corsair is using a tuned SP120L fan which is far-and-away superior to what EVGA has mounted to its card.

Note that EVGA connects its fan to the card, then controls that fan via PWM for GPU diode-based speed modulation. As the GPU diode heats up, the fan will ramp-up its speed. Standard stuff. Corsair, however, is talking straight to the motherboard and without pulse-width modulation, so the speed sits at a firm 100% for our tests (though a user could reduce the speed, maintain high thermal performance, and reduce noise). This does account for some of the idle reduction in the EVGA Hybrid + Corsair radiator fan combination, as EVGA's Hybrid runs a slower radiator fan when idle.

The “best” solution, strictly looking at thermals and ignoring all other considerations, would be the EVGA Hybrid CLC + Corsair SP120 fan, if you were to build your own chart-topping card. Lowering voltage also helps to some degree, as we're seeing a gain from ~18.5C to 19.82C load (and a doubling in idle – forcing a high idle voltage causes that) when changing the Sea Hawk X to more closely resemble the EVGA FTW card's resting voltage. EVGA sits closer to the ~0.982v range during operation – Boost 3.0 fluctuations bouncing that around – while Corsair and MSI sit closer to the ~0.82-0.89v range. That's significant enough to cause ~1~2C change in load temperature.

From a competitive perspective, EVGA would be wise to invest in a better fan. They've already got the right idea with pulse-width modulation as issued by the GPU – no need to run high RPM at idle – but the company is losing out on a huge amount of cooling potential by electing to use an inferior fan. The stock fan struggles to dissipate heat as efficiently as Corsair's, particularly when taking into account the VRAM coldplate. That's a lot of extra heat that ultimately warms the liquid temperature and moves the bottom line temperature output up by ~11C.

The unit performs acceptably out of box, but if you're an enthusiast with a few bucks to spare, it may be worth the project to invest in a better fan from Corsair. The change won't necessarily impact your every day use experience, but it'd be fun for a benchmarking project.

EVGA GTX 1080 FTW Hybrid Endurance

Here's a look at our 2+ hour burn-in charts, which help look at real-world gaming use outside of benchmark scenarios.

evga-1080-ftw-hybrid-endurance

Before moving on to FPS, here's a quick look at our endurance over time chart. Note that maxing out the power limit will flatten the clock-rate more aggressively, which is what we want, though Boost 3.0 behavior does still tank the frequency when it thinks it can get away with an ounce of power savings. The dives aren't nearly as bad in this case as when we were thermally limited, like with the Titan X.

MSI GTX 1080 Sea Hawk X Endurance

Just as a reminder, here's the chart we published in our Corsair Hydro GFX (“Sea Hawk X”) review:

hydro-gfx-temp-endurance

“This chart shows frequency versus temperature and time, plotted over a two-hour burn-in period. The clock-rate is fairly tame for the first 90 minutes, with the average frequency range at at about +/-80MHz. Later, we start seeing hard drops to 215MHz – and that's with an advanced cooling solution on the Hydro GFX.

We've seen this behavior on a couple of Pascal GPUs. During the latter half of the endurance test, the clock-rate will occasionally, for a one-second period, drop to 215MHz from the near-2GHz mark a second prior. We've reached out to ask nVidia about this anomaly, as it is occurring somewhat regularly on Pascal cards. In this instance, it's not a thermal throttle – likely power or something similar. NVidia has informed us that this is part of normal Boost functionality, but we are waiting for further feedback as to why it happens on some cards and not others. So far, the answer appears to be that not all cards or GPUs are equal, and that we need a 'huge' sample size for further analysis.”

EVGA GTX 1080 FTW Hybrid Power Draw

evga-ftw-hybrid-power

Keep in mind that this is total system power draw, not per-device power draw. 


Last modified on August 30, 2016 at 6:23 pm
Steve Burke

Steve started GamersNexus back when it was just a cool name, and now it's grown into an expansive website with an overwhelming amount of features. He recalls his first difficult decision with GN's direction: "I didn't know whether or not I wanted 'Gamers' to have a possessive apostrophe -- I mean, grammatically it should, but I didn't like it in the name. It was ugly. I also had people who were typing apostrophes into the address bar - sigh. It made sense to just leave it as 'Gamers.'"

First world problems, Steve. First world problems.

We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.

Advertisement:

  VigLink badge