Need for Speed SHIFT Gameplay Performance and IQ

We have an exclusive first look at performance in Need for Speed SHIFT using a new patch due out at the end of this month, as well as a new AMD driver which improves performance in this game. We test gameplay on nine video cards, low end to high end, and examine gameplay performance and image quality.


Apples to Apples Performance

We saw some spectacular performance from all nine of the video cards we used in this evaluation. Nonetheless, what we have done up until now is show you the subjective differences between these video cards; that is, how these video cards perform within the context of their competition. In this section, we will explore performance when all things are equal. The following graphs represent tests that we felt were important to do, but did not necessarily fit within the structure of our highest playable settings testing.

Radeon HD 5870 vs. GeForce GTX 285

Article Image

Here, we see the Radeon HD 5870 outperforming the GeForce GTX 285 by 92% at 2560x1600 with 8X MSAA enabled. The GTX 285 is almost playable at this setting, but there is just too much input latency and jitteriness.

Radeon HD 5870 vs. Radeon HD 4870 X2

Article Image

Here, we see the Radeon HD 5870 and HD 4870 X2 almost as equals, with the HD 4870 X2 taking a small lead. Though the difference in framerates is small, the difference in gameplay was huge. Using the HD 4870 X2 at 2560x1600 with 24X CFAA, the game was not smooth at all. It was choppy and there was a huge amount of input latency. It took longer for the car or brake to turn after we pressed the button telling it so, and so we crashed a lot more. Aside from being frustrating, the strain of looking at such a choppy picture was too much to bear, and it gave us a headache during testing. Using the HD 5870, however, it was very smooth and fluid, and the gameplay experience was fantastic.

Radeon HD 5770 vs. Radeon HD 5750

Article Image

When the settings are the same, Radeon HD 5770 is about 15% faster than the Radeon HD 5750. Of course, the price difference between the two is currently about 20%, which doesn’t amount to too much of a premium for higher performance.

GeForce GTX 295 vs. GeForce GTX 275

Article Image

For multi-GPU scaling, we checked out the GeForce GTX 295 and the GTX 275. We found that performance in NFS Shift certainly does scale with 2 GPUs, but of course the scaling is not ideal. The GeForce GTX 295 is about 63% faster than the GTX 275. In our experience, this amount of scaling is average, with some games seeing a greater increase, and some seeing less. Of course, with the huge price difference (more than 2X) that exists between the cost of a GTX 275 and the cost of a GTX 295, we would expect to see more. 

Patch 2 Performance Scaling

After we received Patch 2 and the Hemlock driver from AMD, our biggest question was that of performance scaling. We wanted to see if it was purely the patch that improved performance or a combination of new AMD driver and patch.

Naturally, we set out to answer these questions.

Patch 1 vs. 2

First, we'll look at performance between patches 1 and 2.

Radeon HD 5870 - Catalyst 8.663.1 Beta Hemlock Driver

Article Image

Between Patch 1 and Patch 2, we found a small decrease in performance when using Patch 2 vs. Patch 1 with the Hemlock driver. The difference was small, at about 7%, but it was consistent. And yes, this graph and analysis are correct. Keep reading to find out where our Radeon performance boost came from.

GeForce GTX 285 - ForceWare 191.07 WHQL

Article Image

While the graph looks as though there is a performance increase, the net result is only a 2.2 fps difference. That is certainly not enough of a "boost" to say that performance is improved overall for NVIDIA-based video cards in the NFS Shift Patch 2.

Radeon HD 5870 - Catalyst 8.663.1 Beta vs. 9.10 WHQL

Since we didn’t see the performance increased we hoped to see with patch 2, but we were seeing some performance gains throughout testing on the Radeon HD 5000 series video cards, we could only assume that the Hemlock driver was causing it.

Article Image

And we were right. Between Catalyst 9.10 WHQL and Catalyst 8.663.1, which is the beta driver of Catalyst 9.12, we saw an average framerate increase of 10.9 FPS, which was an increase of about 21%.