With what seems to be an eternity, AMD and NVIDIA have finally rolled out their newest graphics cards late last year with AMD getting much hype because of their latest offering compared to NVIDIA’s kneejerk reaction upon releasing its own combatant in the high-end GPUÂ space. AMD’s R9 290X is the company’s top-tier GPU offering in the R9 series and is its fastest single GPU to date.
To complement our reference card review, ASUS has sent us their custom-cooled variant of the R9 290X for review today and we are going to pit it against the reference card Â to make this review really relevant, we have the GeForce GTX 780 Ti in tow as well in the comparison. Also, we are proud to announce the inclusion of 4K UltraHD benchmarks in our tests.
The AMD Radeon R9 Series
AMD has gone with a new naming scheme for this generation of graphics card but as we’ve mentioned before in our R9 280X /270X article, the only new cards in this series are the top-end R9 290X and 290 (non-X) and midrange R7 260X. To further complicate the situation, the R9 280X and 270X are both rebrands of existing products making it seems like both AMD and NVIDIA (GTX770/760 are also rebrands) are really lazing off on development efforts in the mid-range desktop component sector.
These new cards featuring the Hawaii Island GPU bring to the table a variety of new technologies such as the Mantle API, TruAudio and DirectX 11.1/11.2. All of these technologies currently have games adopting them but that will surely change in the coming months as more and more developers adopt them in their games.
Let’s talk about specs for a moment. Starting off with clock speeds, the R9 290X will be clocked in at up toÂ 1000Mhz. No more fixed clocks and boosts for you guys. AMD has completely changed how frequencies work in the Hawaii GPUs. More on that later. Moving on, we have memory clocks coming in at 5Ghz on 512-bit bus. The R9 290X will come with 4GB of VRAM which provide more bandwidth at slower clocks. That means overclocking the memory can really give some improvements thanks to that 512bit wide bus.Â CrossfireX has also been changed with the R9 290X not needing bridge cables to be able to run multi-GPU. Power is provided via an 8-pin and 6-pin PCI-e power connector.
The ASUS Radeon R9 DirectCUII
ASUS takes their tried and tested board design and builds the the R9 290X around it. Cooled by a newly-styled DirectCU II cooler, the ASUS R9 290X is built to be superior to the reference design of the 290X. The card ships with a 1050Mhz core clock and is set to Performance mode by default (toggle via BIOS switch.)
- DirectCU IIÂ achieves 20% lower temps with 220% the dissipation area, plus 3X quieter running than reference.
- Exclusive CoolTech FanÂ drives wider airflow to keep critical components cool.
- 1050 MHZ engine clock for better performance and outstanding gaming experience
- DIGI+ VRM with 8-phase Super Alloy PowerÂ delivers precise digital power for superior efficiency, reliability, and performance.
- GPU TweakÂ helps you modify clock speeds, voltages, fan performance and more, all via an intuitive interfac
- GPU Tweak StreamingÂ let you share on-screen action in real time â€“ so others can watch live as games are played.
- 4.7% faster game performance than reference Radeon R9 290X in Metro Last Light
I’m sorry but it is rather disingenuous to rate graphics performance by maximum FPS, instead of Average FPS. I’m getting the feeling that you guys don’t play a lot of games, to not realize something as technically fundamental as that.
Looking at the graphs, I don’t see that the graphics performance was rated by maximum FPS. Are we reading the same article?
Please take a look at the last chart on page #4 for example. It is for “Torchlight II”. The Asus 290X DCU2O OC is positioned at the bottom of the chart based on it’s max FPS (158) being the lowest, even though by it’s average FPS (150.64) it should be placed at the 2nd position. Similarly, in the chart for “Bioshock Infinite” (2560×1440, Ultimate), the 290s are placed at the top even though the 780s have a significantly higher average frame rate. If you look carefully at the charts for other games, it becomes obvious that the performance is being rated based on max FPS, which is technically unsound, because max FPS numbers are based on random performance spikes and hence are never a fair representation of a GPU’s true capability.
Also, it is important to note that high max FPS but low average FPS is indicative of performance inconsistency of the GPU under that particular workload, which means unpleasant stuttering with V-Sync turned on, and significantly profuse screen tearing with V-Sync turned off. Both the scenarios lead to a much poorer user experience, as any regular PC gamer would tell you.
For comparison, please visit a proper hardware review site (e.g. Anandtech.com, Tomshardware.com, Guru3d.com, etc) and look into their benchmark charts, where the Max FPS is often not even mentioned.
We used to only display average FPS for the comparative charts but when readers started demanding they wanted to see both min and max in the graphs.
That said, I can see the flaw in this and we will devise another format for our GPU performance charts.
Cheers. Apart from the aforementioned problem I appreciate the to-the-point article and the format of your site.
Appreciate the inputs also. I look forward to using your advice in our upcoming GPU reviews.