Reviews Video Cards

RX Vega 64 Liquid “Unleashed” 28-game Overclocking Showdown vs. the GTX 1080 FE

Intro

This overclocking showdown is part of our RX Vega 64 “Unleashed” series benchmarking the Liquid Cooled Edition (LC) vs. the GTX 1080 Founders Edition (FE).  Today, we have optimized our overclocks with all performance options set to their highest limits to get the most performance from each card.

At stock, we found that the $699 liquid cooled edition of the RX Vega 64 trades blows in performance with the $549 stock-clocked reference GTX 1080 FE using 27 PC games but it failed to impress with 10 virtual reality (VR) games.  This time, we will add Prey as our 28th PC game, and overclock the GTX 1080 FE and the RX Vega 64 liquid cooled edition each as far as they will go to see where they stand in relation to each other fully overclocked.

Overclocking the Gigabyte RX Vega 64 Liquid Cooled Edition

The $699 Gigabyte Liquid Cooled edition of the RX Vega 64 is no different than any other AMD Vega 64 liquid cooled edition, and it is probably impossible to find at this price outside of buying an AMD hardware bundle.  Prey and Wolfenstein II: The New Colossus are usually bundled with this card if you can find one.  The liquid cooled RX Vega 64 is clocked up over the reference version, and we saw our overall stock-clocked Boost average around 1678MHz. We found that as long as the ambient temperatures remain cool and the power and temperature limits are maximized, it will not throttle even with the stock fan profile.  

After much testing, we settled on a  2.0% overclock to the core for an overall average 1716MHz boost, and with memory clocks overclocked from the reference 945MHz to 1080MHz with the radiator fan allowed to ramp to 100%.  Temperatures never reached 70 C and even at 100%, the fan is not annoying, and at stock profile it is rather quiet.We found that higher Vega 64 memory clocks gained significant performance in games.  However, it appears that the Gigabyte RX Vega 64 liquid cooled edition is already running very close to its maximum core clocks, and we needed WattMan’s maximum 1250mV for complete stability in all of our tested games with only a 2% maximum overclock.  

Unfortunately, the power draw goes through the roof when the RX Vega 64 LC is overclocked versus the overclocked GTX 1080 FE.  Below we see the total system draw using the overclocked GTX 1080 FE running maxed-out Heaven 3.0 at 2560×1440 versus the overclocked RX Vega 64 LC on the right.  You can run overclocked GTX 1080 SLI with about the same total system power draw as with a single overclocked RX Vega 64!  

GTX 1080 FE OC (left) vs. RX Vega 64 LC OC (right) total system peak draw.

We tried undervolting but we did not have good results nor could we maintain any meaningful overclock with our sample of the liquid cooled RX Vega 64.  We also noted that higher ambient temperatures caused instability and we needed to run our benchmarks in a room that was cooler than 78 F, unlike with the GTX 1080 where we tested with indoor temperatures peaking around 80 F during a recent Southern California heatwave.

Overclocking the Founders Edition of the GTX 1080

We devoted a separate evaluation to the GTX 1080 Founders Edition overclocking which you can read here.  It currently is priced at $549 and it generally comes bundled with Destiny 2.  Because of an intense heatwave which occurred during testing, we settled on a final stable overclock of +175 MHz to the core which settled in around or above 1935MHz with GPU Boost, and we added 375MHz to achieve a 5375MHz final stable memory clock

We did not need to adjust the fan profile, but left it on automatic. The GTX 1080 memory overclock greatly contributed to the increased performance.  The fan never became obtrusive as we were able to leave it at stock and the GPU remained relatively cool in the mid-70s C.

Testing Platform

Our testing platform is Windows 10 Home 64-bit, using an Intel Core i7-6700K at 4.00GHz which turbos to 4.6GHz for all cores as set in the ASRock Z170 motherboard’s BIOS, and 16GB of HyperX DDR4 at 3333MHz. The settings and hardware are identical except for the two cards being tested.

We also feature our newest 2017 games, Prey and DiRT 4, and we also include eight DX12 games, plus we also benchmark DOOM on the Vulkan API.  We also test VR with Futuremark’s Orange and Blue Room benchmarks and will compare the performance of 28 modern games at 1920×1080, 2560×1440, 3440×1440, and at 3840×2160 resolutions with maximum settings.

Before we run benchmarks, let’s check out the test configuration.

  • Mike Jackson

    So many things wrong with this benchmark comparison review 1) MSAA is currently has a massive performance hit bug on Vega 64, Try benchmarking Crysis 3, Dirt 4, Ashes using anything but MSAA and Vega would easily kill that OC 1080 2) Disabling Aysnc Compute knocking off a good 20% performance boost for Vega in Doom Vulkan 3 was stupid 3) Only show GPU score in 3Dmark testing and not show total score which represents the entire system CPU as well 4) Not buying your Battlefield 1 numbers everyone else showing Vega clearly killing the 1080 in that game 5) Don’t even waste your time trying to overclock Vega Liquid when its already at its max clocks (1677-1750), just going to blow up those watts for nothing, only overclock the memory which give a good 3% to 5% boost in performance without any increase to power, other than that considering this site is littered with Geforce ads all over the place, these results clearly do not surprise me.

    • apoppin

      1) BTR has always used the same benchmarks at the same settings and we do not change settings with Crysis 3, DiRT 4, and AoTS to benefit one card that has issues with MSAA. 2) We did not disable Async Compute in DOOM Vulkan. 3) We always show the total score in 3DMark testing. 4)BF1 is tested in DX11 since OCAT began having issues with DX12. 5) This is an overclocking evaluation which means we had to settle for a 2% OC on the core; Power Draw is pretty high even without overclocking the core. 6) We accept ads from major tech companies. This current ad campaign ends in a few hours and will be replaced with Google ads.

    • xchickenx

      @disqus_vNNVXqw2Nv:disqus derp fanboy

    • Alvin Shiu

      AMDrones as always with a load of excuses.

  • Maxim Egorushkin

    Why did you turn off AMD Optimized tessellation, which is on by default?

    • apoppin

      AMD’s Optimized Tessellation is set to “Use application settings” as it is the only way to compare apples-to-apples with Nvidia settings.

      You got 300 more points in FireStrike extreme with a faster CPU.

      • Maxim Egorushkin

        I do not think an average user disables AMD optimized tesselation. You tested rather unusual settings for Vega.

        • apoppin

          We don’t really know what the “average user” does. However, to compare identical tessellation settings between AMD and Nvidia cards, it is necessary to let the application decide. As to the other settings, Vega has been “unleashed” by setting the power and temperatures limits to maximum and allowing the fan to spin up to 100% to prevent any throttling under overclocked settings. BTR has always tested games at these settings.

          The older AMD cards evidently had issues with ‘extreme’ tessellation used by some devs back in the days of the original Heaven benchmark which is why AMD created this setting. Each successive generation of Radeon graphics since Cypress (5800 series) has improved in regard to tessellation, and I am pretty sure that Vega has improved in this regard over Fiji and Polaris.