Always stay cool: Sapphire RX Vega64 Nitro+ in review

Sapphire took its time, which certainly did not harm the whole project. While some competitors have already scattered PVT samples, i.e. the small series, among the media before the start of the actual mass production, S... Board layout Sapphire differs significantly from AMD's reference design when designing the board. This even makes sense in many places, as the cooling has been optimized and is in the foreground. Due to the elimination of the external spokes... Important preliminary remark In some games we were able to register the phenomenon that the nominally faster Sapphire RX Vega64 Nitro+ fell behind the reference card of the RX Vega. However, that this is not a measurement error or even ... Results in WQHD (2560 x 1440 Pixel) Breathe in sapphires, because the much cooler card now (as expected) not only has the reference card fully under control, but also scores above all with the frame times and thus also the minimum frame rates. You... Results in WQHD (2560 x 1440 Pixel) Although Destiny 2 is actually a DirectX-11-based game, which should be more the GeForce cards, AMD's Vega cards can stand up to their GeForce competitors. The Sapphire RX V... Results in WQHD (2560 x 1440 Pixel) Now the picture returns, because the Sapphire card is slightly slowed down (obviously by the driver), which makes it just behind the reference card. But you don't have the GeForce cards... Results in WQHD (2560 x 1440 Pixel) Doom and Volcano are not anything that could shake the RX Vega64. Here, the Sapphire RX Vega64 Nitro+ is again, as expected, in front of the nominally weaker reference card, although not quite as clear as... Results in WQHD (2560 x 1440 Pixel) Even if the RX Vega64 has to be sorted just behind Nvidia's GeForce GTX 1080, the hierarchy within the AMD guild remains consistent. The Sapphire card is in front of the reference model, if only breathless... Results in WQHD (2560 x 1440 Pixel) All three interesting cards are in fact equal, with the Sapphire card surprisingly pulling the shortest in this trio. Driver optimizations shouldn't really come into play here ... Results in WQHD (2560 x 1440 Pixel) If one had to name a prime example of the advantage of a board partner card, then this game, at least in WQHD, would probably be a really good candidate. Despite DirectX11, you have the GeForce GTX 10...results in WQHD (2560 x 1440 pixels) All relevant cards are very close together, with the Sapphire RX Vega64 Nitro+ leading the upper-middle-class trio as expected. However, the lead is very close. Results in U... Power consumption at different loads The power consumption in the gaming loop is at the measured approx. 327 watts well above the 300 watt mark and thus also above what was measured in the reference model. For this there is the more constant leis... Overclocking Due to the good cooler, the limits of this map are well above what the reference map can do and even comes close in places to the water-cooled conversions. But more than what we have achieved is hardly anything more ... Cooling system and backplate Of course, the generated waste heat is directly related to the recorded power, for which the cooling solution is responsible for optimum dissipation. This cooling system, which is therefore almost monumental, is designed by a rie... Summary Since we have not been able to definitively clarify the question of the drivers and some of the performance disadvantages of the board partner cards compared to the reference at this point, we leave this point out of un...

Overclocking

Due to the good cooler, the limits of this map are well above what the reference map can do and even comes close in places to the water-cooled conversions. More than what we have achieved, however, it is hardly possible to make sense if you do not want to damage your ears permanently.

In our articles "Overclocked and boiled: AMD Radeon RX Vega64 water-cooled" and "AMD Radeon RX Vega64 without temperature limit with interesting findings" has already been explained extensively. In any case, Sapphire's implementation does not change these basic statements.

Temperatures and clock rates

Why we limit ourselves to the output of the values reported as GPU temperature is because the telemetry of our test sample also uses this value. What it is all about with the sometimes significantly higher hotspot temperature, you can read here again: ""AMD Radeon RX Vega64 without temperature limit with interesting findings". We were able to log up to 13 degrees higher values on the map, but they would not have been of concern at any time.

Even with the maximum overclocking, the temperature specified in the sensor loop for the HBM2 memory modules was always below 80°C, so that one can also give a warning here, which also speaks very clearly for Sapphire's cooling solution.

We now tabulate the achieved start and end values for temperatures and GPU clock (boost):

	Initial	Final value
Open Benchtable
GPU Temperatures	39 °C	70 °C
GPU clock	1525 MHz	1507 MHz
Ambient temperature	22 °C	22 °C
Closed Case
GPU Temperatures	41 °C	70 °C
GPU clock	1529 MHz	1507 MHz
Air temperature in the housing	24°C	43°C
OC (Closed Benchtable, Witcher 3 4K)
GPU temperatures (approx. 2944 rpm)	29 °C	73 °C
GPU clock	1696 MHz	1643 MHz
Air temperature in the housing	24°C	50°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.

Gaming and overclocking

We see that pretty much all areas stay cool with the gaming loop, because the airflow is sufficiently dimensioned. However, the GPU tries to keep the 70°C at a good time, so that the area below the package remains sufficiently cool. The voltage converters are all in the green area and the doublers are cooled by the backplate in the mounted state anyway.

This also applies to the use in a closed housing, where we average on the board only by approx. three degrees higher temperatures.

If we overclock at maximum, adjust the fans so that the GPU is also not shot into orbit and close the housing, this draught is still sufficient at almost 390 watts of power consumption to at least approx. 73°C for the GPU. However, the circuit board below the package remains well below 80°C, despite measurable heat migration from the direction of the voltage converters, so that the decision to leave the MLCC in this way is understandable. The voltage converters do not pose any risk at a high of 77°C.

Stress

The stress test puts a bit of a on the power consumption compared to normal gaming, but everything stays almost as cool as with the gaming loop.

Even in the closed housing, this time it is only up to two degrees more at the neuralgic points, nothing more. You can really live comfortably with it.

Warm-up and cooling phase

Finally, we look at these two processes, in which the images show well where the most heat is generated or where the heat is generated. where the cooling starts most directly.