Summary
There are a number of interesting observations that we noticed as we were spending time with the Siege of Shanghai on both the GTX 770 and the R9 280X. We noticed a very high level of CPU usage on our overclocked review rig, as well as a rather significant memory footprint usage after we had been playing for a while. We also found some differences between the AMD and NVIDIA graphics cards that were a bit unexpected. While raw framerate was clearly tipped to one side, the actual feel and fluidity and smoothness of the game was tipped to the other. The tables have turned in this game from what might be expected from past experiences.
CPU, System Memory & GPU Memory Usage
One of the first things that we noticed when we initially starting playing the beta was the high level of CPU usage that was occurring on the author’s personal gaming system, this is an Core i7-2600K system with 16GB of memory and two Radeon HD 7970 cards in CrossFire driving three 1920x1200 monitors. As we were getting acquainted with the basics of game play. Remembering back to Battlefield 3, we were accustomed to seeing 40-50% CPU usage during game play, however, during the Battlefield 4 Beta, we often observed CPU usage in excess of 80-90% on my personal system, at the onset of his game testing.
Moving on to our official review system with the the GeForce 770 GTX, during game play, we observed an average load across all CPU cores in the 90-95% range during each of the testing scenarios. However, with the R9 280X, we were observing CPU usage around 80-85%. Initially we began testing with just 8GB of system memory in the review system. After a significant amount of gameplay, we were noticing that 8GB of memory may not provide enough space for the game. We were experiencing memory being swapped out to the hard drive in virtual memory, meaning we were exceeding 8GB of RAM and this was affecting our smoothness and performance.
We upgraded our test platform to have 16GB of system memory, which is the level that we performed all of our graphed testing at here today. Subjectively, there did feel like there was a difference in the overall gameplay experience by utilizing a larger amount of memory, especially with the GTX 770. More testing into memory utilization needs to be done. The game seems to consume more memory the longer you play. In our testing scenario, we got a maximum of 6.5GB of system RAM utilized just doing our short run-throughs on the previous page. However, it is after several hours of gaming, that the RAM will be pegged through the roof, and in the case of 8GB of system RAM, it just wasn't enough for long sessions of gameplay.
One of the biggest questions that we have been getting is whether 2GB of GPU VRAM memory would be sufficient to play Battlefield 4 at high levels of performance. During our gameplay experience with each card, we observed the actual VRAM usage did not exceed 2GB for each of the tests that we performed. At the 2560x1600 resolution, we typically observed anywhere from 1800MB to 2015MB used at any given time. We find this to be somewhat surprising, as it seemed like it would be likely to exceed the 2GB of VRAM on the 3GB R9 280X. Confirming the 3GB R9 280X, we did not observe video memory usage to exceed 2GB, but it was near the max on the GTX 770.
However, in reality, some of the slowdowns that we experienced on the GTX 770 (that we did not feel on the R9 280X) felt like the issue was related to an insufficient amount of video memory. More testing will be needed to flesh this out more, as the Beta also does not have the full graphics settings included in it. The full-version game may be more graphically intense, have more art assets, and have better image quality putting a greater demand on VRAM and performance.
Overall Performance and Playability
When you look at the FPS results that we have assembled during the course of this article, with just looking at the data, it appears that the GTX 770 blows the R9 280X out of the water from a performance perspective across the board. However, when you sit down and spend some time playing with each card, it becomes rather apparent that the frame rate shown on the graph is a significant misrepresentation of how each card plays during the game.
Although it checked in with a much lower frame rate, the R9 280X provided a much better gaming experience than the GTX 770.
Eyefinity Observations
While we did not do an in-depth analysis of Eyefinity and Crossfire performance, we did log some time on a dual Radeon HD 7970 system running three 24" screens at a total resolution of 5760x1200. It was comfortably playable using Ultra settings and 2X MSAA at that resolution. Once Battlefield 4 launches into retail channels, we will perform a more in-depth analysis of Crossfire, SLI, Eyefinity, and Surround gaming configurations.
The Bottom Line
After logging hours of play time in Battlefield 4 Beta using both the Radeon R9 280X and GeForce GTX 770, the AMD Radeon R9 280X appears to deliver a far superior gaming experience compared to the NVIDIA GeForce 770 GTX no matter what the framerate graphs show. Since the game is so new, there might be a significant amount of performance that could be unlocked with driver updates from both NVIDIA and AMD that could quickly change the observations that we have today. The full-version game could also change this game in big ways, as all the graphics options come into play.
I really do get around 65
