So Far, The Radeon 7 Beats The RTX 2080 (But Only In One Game)

AMD finally lifted the lid on their 7nm consumer devices yesterday. And as part of the big reveal, AMD showed some graphs highlighting the Radeon 7 keeping pace with the RTX 2080, and even outstripping it in one particular game.

But when you dig into the finer details, there's some key info that's missing.

Now, a strong caveat with all of this. As we learned from Intel's i9 launch, any figures from a manufacturer pre-release are to be taken with enough salt to kill a mountain of slugs. Independent third parties haven't been able to run their own tests, and to my knowledge review models of the Radeon 7 haven't gone out to press yet.

Intel's New CPU Launch Sparks Backlash Over Paid Benchmarks

The launch of a new CPU, particularly a gaming-centric one, is supposed to be a flashy event. And when Intel unveiled their 9th generation desktop CPUs, including what they termed as "the best gaming processor ever", that was undoubtedly the plan.

Read more

If you go through AMD's marketing material for the new Radeon 7, or just listen back to yesterday's livestreamed keynote, you might have seen this graph:

Image: Anandtech

It basically shows the Radeon 7 matching, or beating, the RTX 2080 in a bunch of different games. That's entirely normal so far: every company does this before the launch of a product. They need to give you a reason to buy it, after all.

But what's interesting about these slides is what information was included where.

Here's a bit about the improved performance for the Radeon 7 versus Vega 64.

Extreme gaming performance: The AMD Radeon VII graphics card delivers exceptional performance in DirectX® 12- and Vulkan®-based games, including up to 35 percent higher performance in Battlefield™ V10, and up to 42 percent higher performance in Strange Brigade®11, compared to the AMD Radeon RX Vega 64 graphics card. It also delivers up to 25 percent higher performance in the widely popular esports title, Fortnite®.12

The footnotes are worth noting, because that's always where AMD, Intel and Nvidia mention more about the tests that were run. The footnotes are where the system specs are outlined - in most of these cases, tests were run with the Ryzen 7 2700X or the Intel i7 7700K.

In those notes, most of the comparisons are between AMD cards. But there were a few comparisons between the RTX 2080 and the Radeon 7, particularly in Strange Brigade where AMD quietly claimed that their 7nm GPU beat out the RTX 2080 by 14fps at 4K with Ultra settings:

Testing done by AMD performance labs 1/3/19 on Intel i7 7700K,16GB DDR4 3000MHz, Radeon VII, Radeon RX Vega 64, AMD Driver 18.50, RTX 2080, Nvidia driver 417.22 and Windows 10. Using Strange Brigade, Vulkan, Ultra settings 4K: Radeon VII scored 87 fps. RTX 2080 scored 73 fps. Radeon RX Vega 64 scored 61 fps. PC manufacturers may vary configurations yielding different results. All scores are an average of 3 runs with the same settings.

The same release, interestingly, has no information about the testing environment for the RTX 2080 in Battlefield, however:

Testing done by AMD performance labs 1/3/19 on Intel i7 7700K,16GB DDR4 3000MHz, Radeon VII, Radeon RX Vega 64, AMD Driver 18.50, and Windows 10. Using Battlefield V, DX11, Ultra settings 4K: Radeon VII scored 68.1 fps. Radeon RX Vega 64 scored 50.5 fps. PC manufacturers may vary configurations yielding different results. All scores are an average of 3 runs with the same settings. Performance may vary based on use of latest drivers.

Also, Far Cry 5 is mentioned once. It's included in a suite of games and results that were used to test the Vega 64 and the Radeon 7, but the Nvidia card isn't included in that testing note, so we know nothing about the conditions that resulted in AMD getting that one frame advantage over the RTX 2080.

Obviously someone at AMD was running tests against the RTX 2080, and some of those tests were run in the same machine that ran some of the tests for the Radeon 7. But what's weird about this is the absence of information, information that manufacturers normally provide to avoid accusations of biased testing environments or unfairness.

Having games like Far Cry up on stage and then next to nothing about them in the official marketing material raises a basic question: why it was included it all? Someone ran a test between the RTX 2080 and the Radeon 7 in a bunch of games, but the only data AMD has provided that anyone will be able to corroborate relates exclusively to Strange Brigade.

The fact that most of these tests were run on January 3 - days before CES began - is pretty telling. Did AMD not get the card in time for testing? An initial theory was that Strange Brigade's inbuilt benchmark was highlighted because it's faster than putting together scripts and testing games that don't have an automated, in-built benchmark.

But Far Cry 5 - and every modern Ubisoft game - has an inbuilt benchmark of their own, so someone has the data there somewhere. There's some missing info here, and it's a little strange. And it's another reason to take all figures provided by manufacturers with a metric ton of salt.

The Radeon 7 officially launches on February 7 internationally, although Aussie pricing and availability is still unknown.


Comments

    Strange brigade! \o/

    The fact that most of these tests were run on January 3 - days before CES began - is pretty telling

    Yes it tells us that they were using the latest AMD drivers. If the tests were done a month previous they'd likely be different a different version and a different score. I'd imagine they used Strange Brigade because it's a highlight - a case where there is a big gain over Nvidia. *Not* because it's easier to bench. That's marketing 101. What looks better "we beat them by 0.6% in game A" or "we beat them by 15% in game B"?

    Additionally it looks like an AMD affiliate type game: https://www.amd.com/en/gaming/strange-brigade

    The more interesting thing to me is they used Nvidia 417.22 drivers when the 417.35 ones were released in mid December. That indicates one of two things, either they started the testing back when the 417.22 drivers were the latest and they didn't want to switch since that would involve retesting. Or the 417.35 drivers improved Nvidia performance so the results weren't as favourable.

    "Did AMD not get the card in time for testing"..

    For anyone that does benchmarking or follows this stuff, the answer is simple. Drivers! Reviewers can often get the card days or weeks before the launch driver is available.

    Most likely this card was run using the most relevant and latest test build - that isn't finished for further testing or public use yet. Nvidia only got drivers to reviewers like 2-3 days out from the 2080/ti launch!

    Nothing fishy or strange is going on.. This was a preview of what's coming and announcement. But as always, wait for benches.

    But while your at it, why aren't we pointing out Nvidias own benchmark tests which are inaccurate and misleading. No straight ray tracing benchmarks provided at all for 2080/ti, and the "latest" benches at the 2060 launch show you their internal results using both Ray tracing and DLSS on which resulted in better performance than with either off. Nobody can test bith Raytracing with DLSS as no games have launched with both, and other games shown don't have either enabled for reviewers.

Join the discussion!

Trending Stories Right Now