AMD Fury X scale performance better that nVidia Titan X

Todays it is kinda rare, that someone try CrossFire (AMD cards) versus SLI (nVidia cards) using up to four cards. The reasons are pretty simple - usualy reviewer is happy to get one piece. Having 4 pieces of Fury X and Titan X cards is not very common.

But in the pursuit of fastest graphic power (World records on HWbot are waiting!), people sometimes do even more crazy things that "just" using four of the fastest graphic cards:

You are not dreaming 🙂

The results could be demonstrated best on 3DMark Fire Strike test - while using single card (bottom), the Fury X lags a bit to the Titan X (3%, actually). When using 2 cards, then suddently Fury X is neck to neck with Titan X. And the tide turn on 3 crds setup - there Fury X gained good 6% speed up over Titan X and when using the dream setup with four fastest graphic cards todays, then Fury X actually beat Titan X by 9%.

So while this is not enought to make Fury X claim the crown for fastest graphic card today, it is sure proof that on multi GPU setups are AMD Fury X faster that the best nVidia can offer. If we did not take overclocking into the picture, that's also true.

But maybe new drivers (and/or DX12) that let applications use the great resources of Fury X make the card shine…? We see. All the results are there and despite the site is in language most peoples did not understand, the pics and graphs speaks for themselves:

Hooray for Korea testers!

Thats because the top 3 NV cards are starved for air on that board…and are thermal throttling...whereas the Fury's are water cooled.

One problem...outside of Epeen benchmarking on an open bench....what case has room for FOUR single bay rads?! /whistle

A more fair / honest test would have been WC vs WC units.  But that would not get the page clicks. :S

Good point! However there is not stock Titan X with water. eVGA is selling one:
…but that is not stock card like the Fury X. So when 3rd company come with better cooling solution (for example just replacing the rad wih one for dual 120mm fans), the pefrormance again increase (as well, as overclockability)... so it is hard to judge that.

But you are right - watercooling could play a huge role when there are 4 cards stacked closely on each other. Where the next 3 Titan X should take their air?


So such test should be made with the temperature (and throttling) monitoring of each card included. If they throttle, then no wonder Fury X wins. Good point  😉

PS. I bet, that in John's favorite box case: is actually possible to fit all the four Fury X radiators that way, they get clearn air from outside. However... it will be a monster!  ;D

  • 3
  • 846