With triple fan and overlength: MSI GeForce GTX 1080 Ti Gaming X Trio in review

Take the board of the , enlarge the cooler so that three fans fit on it, adjust the board accordingly and cool the majority of the voltage converters directly for the first time - the Gaming X Trio is ready, which has the space between the norma... MSI relies on a rather tidy multi-layer board with a somewhat unconventional design, which was probably also developed from a thermal point of view. The board of revision 6 corresponds approximately to revision 2, as it ... Benchmarks in 2560 x 1440 pixels We have deliberately dispensed with Full HD (1920 x 1080p) as the card runs into the CPU limit even in the highest settings. In WQHD (2560 x 1440 pixels), on the other hand, the actual working environment and the kar... Benchmarks in 3840 x 2160 pixels The card is also significantly faster in this high resolution than a GeForce GTX 1080 Founders Edition or GeForce GTX 1080 Founders Edition. TitanX (Pascal) in Nvidia reference design. Overall, many titles are quite good in UH... Power consumption at different loads The power consumption in the gaming loop is pretty much exactly on the point that MSI set as the power target with 285 watts in the BIOS. In the Torture loop, the power consumption even drops below this value. Auc... Overclocking Manual overclocking with air cooling is average, if you can withstand the fan, which then works at a slightly over 2500 rpm at maximum speed. On the one hand, of course, as always, we have the GPU lottery, on the other hand, we have... Cooling system and backplate The special feature is the use of a now more shortened "sandwich" system, which uses a kind of cooling and stabilization frame between the top of the board and the actual heat sink. That's what you're doing now ... Summary A lot helps a lot, so much is already certain. With the GeForce GTX 1080 Ti Gaming X Trio, MSI not only improves the old Gaming X, but for the first time relies on a new cooling system, THE VR MOSFETs, coils and also the capacitors...

Overclocking

Manual overclocking with air cooling is average, if you can withstand the fan, which then works at a slightly over 2500 rpm at maximum speed. On the one hand, of course, as always, we have the GPU lottery, on the other hand it is difficult with air to reach the 2.1 GHz limit even in closed enclosures and, if so, to keep it permanently. The memory was also moderately overclocked, but it is also getting quite hot, so you should think about this carefully.

Temperatures and clock rates

First, we compare the achieved start and end values for temperatures and GPU clock (boost) in tabular form:

	Initial	Final value
Open Benchtable
GPU Temperatures	39 °C	66 °C
GPU clock	1974 MHz	1911-1923 MHz
Ambient temperature	22 °C	22 °C
Closed Case
GPU Temperatures	40 °C	67 °C
GPU clock	1974 MHz	1898-1911 MHz
Air temperature in the housing	22°C	40°C
OC (Open Benchtable)
GPU temperatures (approx. 2530 rpm)	28 °C	62 °C
GPU clock	2068 MHz	2055 MHz
Ambient temperature	22°C	22°C

Overview Graphs: Temperatures vs. Clock

For better illustrations now again the respective courses considering our timeline of a total of 15 minutes each for the warm-up time.

Thermal analysis of the back of the board

Finally, we consider the thermal analysis of the respective load states. In order to better understand the results and the cooler, let's first compare the warm-up and cooling phases.

Warming up and cooling

As a hotspot, we first select the GPU, some memory modules, and the two voltage converter groups for the GPU (left, vertical) and memory (top, half-left). It will then be interesting to observe the heat development and migration.

When the fully heated board cools down, we see that most of it is dissipated via the GPU and its package.

Operating temperatures after 30 minutes

We see that the VRM range of the memory and the memory modules positioned close to it become very hot, with the waste heat moving very strongly over time towards the GPU due to the lack of a directly cooling VRM heat sink, thus reducing the memory modules on the way with heats up. We have already been able to show this fact to the responsible MSI engineer here in the laboratory and we are curious to see whether our proposals will be incorporated into this revision.

We can clearly see that the values measured below the GPU package are higher than those of the GPU diode inside. Here, the GPU indirectly also cools the board. Nothing dramatic, but a clear indication.

The stress test shifts the temperature load more in the direction of memory and memory VRM, which is easy to see.

In the stress test, we then measure after approx. 30 minutes on the hottest memory module in the closed housing approx. 98 degrees, 3 degrees more than the manufacturer specifies in the work area specifications. This is not really dramatic, but it is somewhat borderline.