Sapphire Radeon RX Vega 64 Nitro+ Review

Cooling & Noise

Cooling Solution & Backplate

There’s a direct relationship between power consumption and waste heat, and it's the thermal solution's job to cope with the latter. Sapphire's Radeon RX Vega 64 Nitro+ has a huge cooler, as we've seen, with a massive fin-laden sink sitting on top of a big vapor chamber. That vapor chamber’s waste heat is dissipated across the cooler's body through no fewer than six heat pipes that are installed above and lie down on top of it.

Sapphire applies a black coating to the backplate's inside surface to help it absorb radiated heat more effectively. A more active approach is taken for some of the components on the PCB's back side: the phase doublers, the memory voltage converter’s gate driver, and the PWM controller are connected to the backplate using thermal pads.

Cooling System Overview
Type
Air cooler
GPU Cooling
Vapor chamber
Cooling Fins
Aluminum, vertical orientation
Tight spacing; some bent
Heat Pipes
3x 8mm + 3x 6mm
Copper composite material, nickel-plated
VRM Cooling
GPU and memory VRM via cooling frame
RAM Cooling
Memory cooling of HBM2 module via vapor chamber
Fans
2x 9.7cm fan (10cm opening), nine blades
1x 8.7cm fan (9 cm opening), 11 blades
Semi-passive control
Backplate
Aluminum
Cools phase doublers, PWM controller, and VRM gate driver

Apart from the vapor chamber responsible for the GPU and HBM2, there are three central 8mm heat pipes and two outer 6mm pipes, all of which are made of nickel-plated composite material. They transport the GPU’s thermal energy to different areas of the cooler’s body. There’s also a sixth 6mm heat pipe, which is shorter and bent, above the voltage converters.

The cooler’s performance is truly awesome. As a side note, we can also see just how many screws Sapphire uses around the vapor chamber. Four of them fasten the vapor chamber to the package, and another four are used in the cooling frame area around the package. This way, pressure on the package’s sensitive interposer is distributed evenly.

Fan Curves & Noise

The fan curves illustrate Sapphire's custom implementation. A semi-passive mode is enabled through a programmable controller, which also results in WattMan displaying rotational speed, even after the controller turns off the fans. This little trick might create some confusion when looking at the sensor readings, but an external tachometer helps clarify. The shape of the curve is interesting as well, assuring us that the GPU doesn't exceed its 70°C target temperature.

Switching to the stress test doesn’t really change anything. Overall, the card’s three fans are sufficient.

Sapphire's Radeon RX Vega 64 Nitro+ does have some reserves where fan speed is concerned. This is to say that the card could operate more quietly, but there would be a price to pay for this in the form of a lower GPU frequency. The default configuration represents a well-balanced compromise.

Fan RPM & Noise Measurements
Fan RPM Open Benchtable Maximum
1493 RPM
Fan RPM Open Benchtable Average
1485 RPM
Fan RPM Closed Case Maximum1611 RPM
Fan RPM Closed Case Average1585 RPM
Noise (Air) Maximum
39.8 dB(A)
Noise (Air) Average
39.1 dB(A)
Noise (Air) Idle0 dB(A)
Noise Characteristics /
Subjective Impressions
Almost no low-frequency bearing noises
Some motor noises below 1 Hz
Moderate air and turbulence noises
Some coil whine

This snapshot illustrates the entire frequency range of our laboratory measurements, adding some data to our subjective observations.

A 39.1 dB(A) measurement is great for a high-performance graphics card like this one. We were particularly impressed by its behavior in the closed case, with our hotbox reaching up to 50°C during our overclocking efforts.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content


Create a new thread in the Reviews comments forum about this subject
This thread is closed for comments
31 comments
Comment from the forums
    Your comment
  • docswag
    What mosfets is sapphire using for VDDC
  • AgentLozen
    Even though Vega 64 is rough around the edges, I appreciate that you gave the Sapphire Nitro an Editor's Choice award for it's technical prowess.

    There is so much hardware out there with cut corners that it's nice to see something work so well.
  • FormatC
    Quote:
    What mosfets is sapphire using for VDDC
    Plase take a look at page Two. You will get the full info, pictures included. ;)

    I wrote on page Two:
    Quote:
    ...This so-called doubling is achieved through a total of seven IR3598s located on the back of the board. Voltage conversion for the 14 circuits is handled by one IRF6811 (on the high side) and one IRF6894 (on the low-side) for each circuit. The latter also include the necessary Schottky diode.
  • docswag
    Anonymous said:
    Quote:
    What mosfets is sapphire using for VDDC
    Plase take a look at page Two. You will get the full info, pictures included. ;)

    I wrote on page Two:
    Quote:
    ...This so-called doubling is achieved through a total of seven IR3598s located on the back of the board. Voltage conversion for the 14 circuits is handled by one IRF6811 (on the high side) and one IRF6894 (on the low-side) for each circuit. The latter also include the necessary Schottky diode.


    My bad, I must have skimmed over that part. Thanks!
  • Cryio
    So once the drivers will actually apply to 3rd party OEMs as well given the reference with slower clocks is sometimes faster, will basically make Vega64 universally and sometimes significantly faster than the 1080 and guaranteed substantially faster in DX12/Vulkan games.

    So ... with some underclocking and undervolting, this should perform on the level of the 1080 while being more future proof. Got it.

    Wonderful work AMD! Too bad miners skyrocketed the price.
  • Sakkura
    Is this ever going to be sold to people, though?

    And will there be a Vega 56 version?
  • Wisecracker
    Thanks for the bench work. Interesting boost in minimums in some titles considering struggles with 'optimizations'

    What's the verdict on the previously reported 'elevation difference' between the memory stacks and processor?
  • FormatC
    This is randomly the molded version. That means, no difference. :)

    Powercolor got unmolded and the result is well-known. But I also know from others, that Sapphire is using also both packages. The 3rd package version is Vega56 only.
  • Cryio
    While also being more efficient or as efficient as the 1080 I might add.
  • FormatC
    It depends at your preferences, what you need in your rig. Only the price is currently a big con. Too bad.
  • mahone7
    a gtx 1080 equivalent 16 months later
    WELP
  • Martell1977
    Maybe I missed it, but did you OC the HBM2? I read a lot about how Vega tends to get a bigger boost from overclocking the HBM2 than from OC the core. Would like to see a comparison between this card at stock and with a nice HBM2 OC.
  • FormatC
    We wrote OC und UV reviews, this card is not different :)

    @mahone7:
    I saw the first Vega over one year ago in Sonoma. And I got my first PVT samples of a custom card in summer. AMD lost too much time for nothing. :(
  • techy1966
    Anonymous said:
    Sapphire’s engineers took their sweet time creating a graphics card that heats up the third-party Vega 64 market by staying cooler than the competition. We run its Radeon RX Vega 64 Nitro+ through our test suite.

    Sapphire Radeon RX Vega 64 Nitro+ Review : Read more


    Very nice review of Sapphires new card as always they just do it right without to many issues. I currently have Sapphires Tri-X 390x 8GB card and it performs like a champ and runs way cooler than other 390 series cards I have seen in reviews for them. If I was going to get a new video card right now and it was gonna be a new AMD card it would be one made by Sapphire for sure as I just really like their cooler design and how effective it is on hardware that is not known to play nice with most other coolers out their.
  • FormatC
    The Sapphire RX Vega Nitro+ is a good example to show, how to do cooling right.
  • TJ Hooker
    Anonymous said:
    While also being more efficient or as efficient as the 1080 I might add.

    How do you figure that? Sapphire Vega 64 looks to be maybe 15% faster on average than a 1080 FE, while using ~90% more power. That means the Sapphire Vega 64 performance per watt is only ~60% of a 1080's perf/watt. Do you really think you can increase the performance per watt of a Vega 64 by 50% through underclocking/volting?

    Edit: I forgot to take into account the fact that the sapphire card seems to be performing at less than it's full potential due to drivers. So it's perf/watt is probably going to end up a bit better at stock settings than I estimated here. Still doubt you can get it to match or exceed a 1080s perf/watt through tweaking.
  • Rock_n_Rolla
    Hoowaah! Finally.., a heaivly OC'd Vega 64 \ii/ kudos to Sapphire, actually they're one of ATi's long time partners way back when the company was still on its own (not yet bought by AMD), till now they still have exclusivity (aside from other long time ATi partners) on Radeon's top end GPU models to fiddle and tweak with.

    IMO, based on the benchies it really is freakin fast tho i dont have any idea what the percentage increase it will yield when AMD release a vendor specific driver for those heavily OC'd models like this one but based on the specs and benchies for sure it will perform much faster.
  • silverblue
    I was wondering if it was possible that AMD would release a Vega Nano, as downclocking Vega should put it back within a decent efficiency window (like with the Fury Nano), however this time around the competition from the 1070 and its mini ITX derivatives is too strong. Moving Vega to the next manufacturing node cannot come soon enough, but we can't blame the current node for everything.
  • rush21hit
    Good review. Good luck finding one of these, tho.

    Even the reference are elusive, if not at sky high price. Understandably, all thanks to its monstrous compute output. It's an AMD's tradition; an overly engineered gpu with dissapointing actual performance because of driver incompetence.
    Well, to AMD's defense, Gameworks titles run like crap on Nvidia's too. So there's that.
  • photonboy
    VEGA-64 at a cheaper price, lower power consumption (coming in 2018), better drivers, and more GAMES that properly utilize it would make a really great product.

    *However, NVidia will be coming out with a Volta product that adds in better ASync Compute etc so competition is going to heat up between quite similar GPU architectures in 2018.

    (Gamers Nexus did a review of the Titan V100 that has a Volta GPU in it. It also has tensor cores that don't get utilized for gaming but in his quick gaming tests it appears it has good ASync Compute as it managed to get up to 2x the FPS as a GTX1080 at a lower frequency... I won't bore you with the details but normally things don't scale CUDA vs FPS etc so it's a good SUGGESTION at the very least that gaming Volta for DX12/Vulkan especially is on track... probably other features in there too, and frankly I think AMD and NVidia GPU's are going to get incredibly CLOSE in architecture especially considering that the PS4, XBOX ONE and likely PS5 etc will be based on a GCN architecture so I think we'll just fine-tune that)