Forza Motorsport 7 Video Card Performance Update

Forza Motorsport 7 gaming performance has changed, video cards stack up differently when compared. We take Forza Motorsport 7 and apply new NVIDIA GeForce 387.92 and AMD Crimson ReLive 17.10.1 drivers to find out how these compare, what performance differences there are, and if AMD Radeon RX Vega is still king in this game.

Introduction

The new game Forza Motorsport 7 was released on October 3rd and has been a very hot topic since its release. It all started when reviews came in that AMD’s Radeon Vega GPUs were doing well in the game, performing well above the competition's performance for a change. It was a total reversal of what we’ve seen from AMD Radeon RX Vega GPUs in previous game testing.

Article Image

In our recent AMD Radeon RX Vega 64 4K Performance Review we included Forza Motorsport 7 for this very reason. We had to test it ourselves and get to the bottom of how it performs in real-world gameplay, not the built-in "canned" benchmark. We set about creating a custom circuit track with heavy weather to stress performance.

Using the NVIDIA "Game Ready" GeForce 385.69 driver for Forza Motorsport 7 and the latest (at the time) AMD Crimson ReLive Edition driver 17.9.3, we evaluated performance at 4K. Our results showed that indeed, AMD Radeon RX Vega 64 was faster than the GeForce GTX 1080 by 7%. The AMD Radeon RX Vega 64 did not however come close to GeForce GTX 1080 Ti performance at 4K.


NVIDIA Fires Back

Nothing lasts forever, and in the case of performance, it didn’t take long for AMD to be dethroned by NVIDIA in Forza Motorsport 7 performance. On October 9th NVIDIA released yet another "Game Ready" driver, version GeForce 387.92, for you guessed it, a performance uplift in Forza Motorsport 7. In an email sent to press it states:

"Forza Motorsport 7 Sees Performance Improvements of up to 25%. Our driver team’s job does not end once a game launches. They work tirelessly to continue improving performance even after a game launches. The fruits of their labor can be seen with the performance improvements in Forza Motorsport 7, which exhibits performance improvements between 15 and 25 percent, depending on the GPU and configuration you are testing."

Article Image Article Image

As you can see in the screenshots above, the original driver 385.69 was definitely a "Game Ready" driver for Forza Motorsport 7. In the second screenshot you can also see how 387.92 is also another "Game Ready" driver, oddly.

It is not common that we see multiple "Game Ready" drivers for the same game. Normally the "Game Ready" driver is a one-time deal that is expected on game launch day, or before. Performance additions would simply be delivered as subsequent driver releases, noting the performance uplift, rather than calling it another "Game Ready" driver. At any rate, it is easy to link the fact that AMD was ahead in performance for a change, and NVIDIA wasn’t going to let that stand for long.

Re-Evaluating Forza Motorsport 7 Performance

In today’s follow-up performance evaluation of Forza Motorsport 7 we are going to utilize NVIDIA’s new driver and see what the real-world performance improvement actually is. Then we are going to compare the GeForce GTX 1080 Ti, GTX 1080, and AMD Radeon RX Vega 64 once again using all new drivers to see how performance stacks up. We are also going to do this at 4K and 1440p this time. In addition, we are going to test GeForce GTX 1070 and Vega 56 against each other with the new drivers as well. This will be a more comprehensive look at performance and how video cards stack up in this new game with latest drivers.

Not to be left out, there is also a newer AMD driver. AMD Crimson ReLive Edition 17.10.1 is out, and we are going to use this driver today and re-test performance as well. There is nothing in the driver notes that indicates any performance changes for Forza Motorsport 7 for this new driver, but we are going to retest with it because it is the latest driver.

Article Image Article Image

A new Forza Motorsport 7 patch has been released as well. You can note the version change in the two above screenshots. Note that the new patch notes do not indicate any performance changes. However, there are stability fixes with it, and we are of course using the newer version of the game in today’s testing.

Article Image Article Image Article Image

We are also making a slight change to the game settings for all future tests, compared to our previous testing. In the game graphics settings instead of leaving the scaling set to "Dynamic" we are setting it to 100%. This makes sure the resolution scaling in the game matches the viewport resolution.

We should note that to ensure proper resolution in UWP DX12 Windows 10 games like this we also set the desktop to the same resolution to make sure the viewport resolution matches perfectly. In some games we’ve experienced the viewport resolution running independently of the in-game rendered resolution. Otherwise, all settings are the same as previous which means the highest possible in-game settings as shown above.

We are completely re-testing performance on all cards, with all drivers, even the previous 385.69 driver, all data is fresh. Therefore, do not compare the performance in this article today with our previous article, the test is setup a little different and won’t compare exactly. This test today is a more mature and set up to graphically challenge the video cards. All tests were done in this review are new, and on game version 1.107.5368.2.


PresentMon

A note about PresentMon. Due to this game not supporting "Exclusive" full screen support, we cannot use Mirillis Action! Therefore, we must use PresentMon to capture UWP DX12 gaming in Windows 10. PresentMon captures data differently than other known framerate capture applications. PresentMon captures faster than each framerate in one second, it captures frametime in fractions of a second. We then convert those frametimes to framerate.

However, not everything you see is actually the framerate felt or delivered in-game. There are fluctuations that go wildly high, or low, these are not felt during gaming. We remove the maximum and minimum framerate table because it is not representative of what is experienced. However, the average framerate is still accurate, and is what we experience in terms of average framerate.

Therefore, what you should focus on is the average framerate, and the consistency of the framerate. In the graph what you are looking at is frametime converted to framerate. Hopefully this should clear up some confusion about PresentMon. There is currently no way to force PresentMon to sample every one second, it would solve a lot of problems, but the program is currently limited in that way. Basically, it’s the best we got right now at representing performance on a graph for UWP Windows 10 DX12 games.

Article Image