AMD Radeon R9 290X Hawaii within the take a look at – In opposition to the TITANs

A little new architecture

AMD’s new flagship GPU is finally here and we have all of the benchmark results to tell you where it stands !!

Note: We have also tested the new AMD Radeon R9 290X in CrossFire and at 4K resolutions; Check out the full frame rating history here !!

AMD brought media, analysts, and customers to Hawaii last month to discuss a new graphics chip due for release this year. As you might have guessed from the location, the code name for this GPU was actually Hawaii. It was aimed at the upper end of the discrete graphics market to rival NVIDIA’s GTX 780 and GTX TITAN.

Earlier this month we tested the AMD Radeon R9 280X, R9 270X, and the R7 260X. None of them were based on this new GPU. Instead, these cards were all about rebranding and repositioning existing hardware in the market (albeit at reduced prices). These lower prices made the R9 280X one of our current favorite GPUs because it offers NVIDIA unmatched performance per price point.

But today it’s a little different, today we’re talking about a much more expensive product that has to live up to some pretty lofty goals and ambitions set by the AMD PR and Marketing machine. With an MSRP of $ 549, the new AMD Radeon R9 290X becomes the flagship of the Radeon brand. The question is: where is this ship going?

The AMD Hawaii architecture

To get right up front, the Hawaii design is very similar to that of the Tahiti GPU on the Radeon HD 7970 and R9 280X cards. Based on the same GCN (Graphics Core Next) architecture, AMD has assured us as a long-term vision of improving Hawaii in some key areas while maintaining the same core.

Hawaii is based on shader engines, of which the R9 290X has four. Each of them contains 11 CU (arithmetic units), each containing 4 SIMD arrays. If we do the quick calculation, we get the total number of stream processors of 2,816 on the R9 290X.


The four shader engines enable the R9 290X to process almost twice as many primitives as the earlier R9 280X (HD 7970). There are four tessellation engines in total, so tessellation performance should get a modest boost compared to previous products and be more in line with what NVIDIA has in their latest cards.


AMD’s GCN architecture has proven to be very robust and flexible. Each CU consists of 4 x 16 wide vector units, a single scalar unit, 4 texture filter units, the associated texture fetch load / store units and many registers, data enables and cache. While the base unit for the R9 290X remains the same as the parts released last year, they are still very powerful computing units that represent a massive advance over the previous VLIW 4/5 architectures of earlier AMD cards.


Double the primitives, double the geometry throughput, and double the tessellation. AMD spiced up the Hawaiian front-end so it could better compete with NVIDIA’s high-end GTX 780 and GTX Titan. The performance of tessellation and geometry has always been lower than what NVIDIA offers, but this should cover most of the basics when it comes to real-world gameplay with tessellation.


This chip offers up to 64 pixels per clock. It also offers up to 128 Z / Stencil operations. Each render backend is connected to a memory controller and offers an enormous fill rate per cycle. With a frequency between 800 and 1000 MHz, a single card should be able to color many pixels on multiple screens without any problems.


Speaking of memory controllers: The R9 290X has eight of them. This enables a memory bus width of 512 bits. AMD delivers this card with 4 GB of memory at 5 GHz effectively, which gives about 320 GB / s bandwidth. It outperforms the 320-bit memory controller of the HD 7970 / R9 280X, the GTX 780 and the GTX Titan.


Hawaii is a bigger chip than what AMD is used to. It’s quite a bit smaller than the GK110 at 530mm square, but it actually seems like a small step over that other part in terms of performance across different workloads. In this table we see that we have a doubling of the primitives, 1.3 times the processing power, higher tex and pixel fill rates, higher memory bandwidth (which could possibly get higher with faster memory) and a die area well below these Relationships between the Tahiti chip and lies Hawaii. AMD has reorganized the GCN architecture to maximize all of these properties without exploding chip size and transistor count. This is a very powerful chip, and it’s not much bigger than the previous high-end from AMD.

Comments are closed.