AMD Radeon Graphics Cards Trump NVIDIA Alternatives in VRMark Cyan Room

[Raevenlord]

Benchmarking company Futuremark has recently introduced a new benchmark to its VRMark suite, the Cyan Room, which brings the latest in rendering technologies to the VR world. Futuremark expects this test to leverage the latest hardware and software developments in DX12 to better utilize today's GPUs still somewhat untapped power. In something of a plot twist, AMD's Radeon architectures (in the form of Polaris 20-based RX 580 and Vega-based RX Vega 56 and RX Vega 64) trump NVIDIA's equivalent offerings in pure performance numbers.

Testing was performed by pairing a Ryzen 7 1800X CPU with a selection of graphics cards from both AMD and NVIDIA, supported by 16GB of DDR4-2933 system memory, and Windows 10 x64. In a post on Radeon gaming, Scott Wasson said that "The Cyan Room (...) highlights AMD's continued performance leadership on this (VR) front," adding that "the Radeon GPUs we tested have clear leads over their direct competition. What's more, all the Radeon GPUs are meeting the key requirement for today's VR headsets by delivering at least 90 frames per second in this test."





Which is true, hands-down: it's just a matter of looking at those AMD internal test numbers, where the RX Vega 56, which usually trades blows with the GTX 1070 Ti, is seen far ahead of NVIDIA's GTX 1080 graphics cards. An equally impressive showing is delivered by AMD's RX 580, which leads its NVIDIA competitor, the GTX 1060 6 GB, by a wide as margin as the Vega-based cards. This is impressive, to say the least, and a far-cry from what we're used to seeing from the AMD camp this generation, at least when it comes to pure frame rate competitiveness.



View at TechPowerUp Main Site

 

[P4-630]

So where are the 1070 and 1080Ti?

 

[arbiter]

at least when it comes to pure frame rate competitiveness.

This isn't old days where who gets the highest fps win's. It comes down to which provides smooth more consistent experience. Doesn't matter if reported fps is higher as their could be under lying issues of frame times that negate it.

 

[Kaapstad]

Cyan Room has very consistent fps when it runs and Vega 64 is very quick on it.

 

[bug]

What plot twist? DX12 has always run better on AMD hardware. And do notice the 81fps in there, the requirement for VR is a smooth 90.

 

[Kaapstad]

What plot twist? DX12 has always run better on AMD hardware. And do notice the 81fps in there, the requirement for VR is a smooth 90.

It runs ok on my crappy old NVidia card.

https://www.3dmark.com/vrpcr/4256

 

[the54thvoid]

Missing 1080ti of course. On a price comparison fair enough but on BOM costs (perhaps someone can help?) Vega 64 is surely more expensive to produce than a 1080ti. It is sold at a very low profit margin, compared to the 1080ti. On a hardware scale, Vega 64 should be where it is, comfortably. The fact it doesn't seem to compete at all with GP102 (let alone the full core GP102) in DX12 or Vulkan shows the hardware is still not a match for Nvidia.

Cost is the leveller though so it's good to see the 64 card soundly winning against the 1080 but still....

For ref: Vulkan

 

[silentbogo]

So where are the 1070 and 1080Ti?

Gone for better marketing, I guess....

HERE

 

[the54thvoid]

Didn't realise at first this was AMD PR.

Oh well, no news here.

 

[medi01]

RX 580 vs GTX 1060 6Gb gap is impressive.

Vega 64 is surely more expensive to produce than a 1080ti.

1080Ti is more expensive to produce, as nVidia asks more for its chips, but this is somehow very relevant to consumers, right?

 

[lexluthermiester]

So, once again, Radeon and Geforce are trading blows and Vega competes.

 

[bug]

So, once again, Radeon and Geforce are trading blows and Vega competes.

Tbh, AMD has always looked good in benchmarks. But lately they've started to also look good in games, which is cool. If they manage to close the power draw gap as well, we'll be in user's/buyer's heaven

 

[the54thvoid]

RX 580 vs GTX 1060 6Gb gap is impressive.



1080Ti is more expensive to produce, as nVidia asks more for its chips, but this is somehow very relevant to consumers, right?

I'm suggesting the 1080 ti is cheaper to produce than the highly complex Vega 64 with its HBM2. The cost to the consumer is due to performance competitiveness. Vega 64's soul mate rival is not the 1080, it's the 1080ti (if not Titan Xp). It's only cheaper because it's worse. And that's why AMD didn't put the 1080 ti on the chart. On their PR chart.

 

[cucker tarlson]

Titan Xp owner (on ocn.net) posted his score with 213fps, so 52% over the 1080 from amd chart ?
Just a guess here, but if this is same story as latest dx12 games nvidia might slack off with a slower CPU and will gain the lost ground once paired with 7700K or faster.

 

[silentbogo]

1080Ti is more expensive to produce, as nVidia asks more for its chips, but this is somehow very relevant to consumers, right?

One point of relevance is the current market price. If we take US price as the most stable at this point, then the cheapest 1080Ti is only $50 more expensive than the cheapest available Vega64.
I think in this case $50 is definitely worth the 20-25% performance bump (even in non-conventional VR benchmarks).
Also, Cyan Room is relatively new and was purposefully selected to showcase this "advantage".
VRMark Orange Room and more demanding Blue Room tell the whole different story (check HWBot or 3DMark rankings).

 

[lexluthermiester]

No offense but I've reported your post for being utterly irrelevant and off topic.

You're genuinely better off starting your own thread on Fanboy culture but for that there is an exceptionally simple answer that no Ph.D is required to dissect.

Internet fanboys exist because people are insecure and cowardly.

Also -we have very many non-english forum members that have a great grasp on the language but writing styles such as yours push the boundaries a bit too far. Plain english is better for such things.

Agreed, and joined you.

 

[bug]

Agreed, and joined you.

Do I smell some deleted posts?

 

[lexluthermiester]

Do I smell some deleted posts?

You do.

 

[TheDeeGee]

That's pretty cool for those 1 or 2 Games out there that use DX12.

Meanwhile Nvidia works better for the other 99%.

 

[bug]

That's pretty cool for those 1 or 2 Games out there that use DX12.

Meanwhile Nvidia works better for the other 99%.

Not 99%, but it certainly fares way better then the hand picked DX12 or VR demos would have you believe.
That said, if you're an enthusiast, DX12 and VR may be exactly your thing, so I have no problem with hardware being shown/benchmarked from different angles. As long as we understand the point of view.

 

[Fierce Guppy]

My GTX 1080 Ti is factory overclocked, but I've set it to stock speeds for this test. Boost clock = 1582 , Mem clock 11G. VRMark scores it 8548.

btw, anyone with a VR headset ought to try out the Cyan room VR experience. Nice...

 

[Kanan]

Then again, it's more than 50$ difference between cheapest Vega 64 and 1080 Ti, because it's stupid to buy the cheapest 1080 Ti. I'm not gonna invest over 600 on a GPU and then cheap out on it. So better count it to be 80-100$ more expensive. Vega 64, even cheapest, on the other hand is the same quality as all the other ones, and the PCB is well known and of highest quality, cooler/fan aside.

 

[arbiter]

Then again, it's more than 50$ difference between cheapest Vega 64 and 1080 Ti, because it's stupid to buy the cheapest 1080 Ti. I'm not gonna invest over 600 on a GPU and then cheap out on it. So better count it to be 80-100$ more expensive. Vega 64, even cheapest, on the other hand is the same quality as all the other ones, and the PCB is well known and of highest quality, cooler/fan aside.

blower card is about 30$ more, and when i looked a 3rd party cooler like ACX, gigabyte WF style and MSI one were 50$ more then a vega 64 with a water rad so 80-100$ is bit exaggerated. I bet most VR games would way better then on vega64. These are synthetic benchmarks and those vs real word games don't always look that good on amd cards. Least in case of most history.

 

[Kanan]

blower card is about 30$ more, and when i looked a 3rd party cooler like ACX, gigabyte WF style and MSI one were 50$ more then a vega 64 with a water rad so 80-100$ is bit exaggerated. I bet most VR games would way better then on vega64. These are synthetic benchmarks and those vs real word games don't always look that good on amd cards. Least in case of most history.

Well prices always differ so it's probably a waste of time to debate that. That said, I'm not defending the current price situation of Vega 64 vs 1080 Ti - I would most definitely go with 1080 Ti then. Vega 64 is just definitely more cheap if you're going with a Sync monitor - Gsync adds about 200 bucks extra to that 50 bucks of 1080 Ti vs Vega 64, so in that situation I'd go with Vega 64.