AMD Shares More Details On The Radeon VII

Yesterday was an exciting day for AMD aficionados as the California chipmaker announced its Radeon VII and demoed its third-generation Ryzen processors. AMD aims to target gaming enthusiasts and content creators with its new gaming graphics card.

Credit: AMDCredit: AMD

Many enthusiasts were praying for a Navi announcement. However, they received a die-shrunk version of Vega 10 instead. The Radeon VII employs AMD's existing Vega 20 graphics processor. In case you haven't been following AMD's discrete graphics card evolution, Vega 20 made its debut with the chipmaker's Radeon Instinct MI60 and MI50 data center graphics cards a couple of months ago. Vega 10 takes after the Graphics Core Next (GCN) 5.0 microarchitecture while Vega 20 is based on the updated GCN 5.1 revision.

Meet The AMD Radeon VII

Credit: AMDCredit: AMD

Vega 10 was manufactured by GlobalFoundries under the company's 14nm FinFET node. It measured 487mm² and contained up to 12,500 million transistors. For Vega 20, AMD turned to TSMC and its 7nm FinFET manufacturing process. Although Vega 20 only measured 331mm², the die housed up to 13,230 million transistors. Thanks to this die-shrunk, AMD was able to deliver up to 1.8 times the gaming performance per area and double the memory capacity and memory bandwidth on the Radeon VII in comparison to the Radeon RX Vega 64.

Credit: AMDCredit: AMD

The AMD Radeon VII comes equipped with 60 compute units, and since the graphics card adheres to the GCN standard, this tallies up to 3,840 stream processors, 240 TMUs (texture mapping units), and 64 ROPs (render output units). AMD states that the Radeon VII will feature a 1,450MHz base clock and a boost clock that escalates up to 1,800MHz. The Radeon VII is good for up to 13.8 TFLOPs of single-precision floating-point performance.

On the memory side, it sports 16GB of HBM2 (second-generation High-Bandwidth Memory) operating across a 4,096-bit memory bus, which is capable of delivering a phenomenal memory bandwidth of up to 1 TB/s. AMD didn't disclose the Radeon VII's memory speed. However, if we do the math, the memory would need to be clocked at 2,000MHz to achieve a throughput of 1024 GB/s.

Credit: AMDCredit: AMD

The VII's reference design flaunts a silver shroud, three cooling fans, and an illuminated Radeon logo on the side. On top of that the graphics card is outfitted with two 8-pin PCIe power connectors. Although AMD didn't reveal the TDP (thermal design power), the Radeon VII's triple-fan cooler suggests it's somewhere in the range of around 295W. The reference model exhibited at AMD's CES 2019 keynote comes with three DisplayPort outputs and a single HDMI port.

Radeon VII vs Radeon RX Vega 64 vs GeForce RTX 2080

In order to prove the Radeon VII's prowess, AMD pitched the graphics card against the Radeon RX Vega 64 and the Nvidia GeForce RTX 2080 with the first being the flagship that it's replacing, and the latter being the rival it's targeting. AMD's results looked promising and revealed that the red chipmaker is closing the performance gap with Nvidia. 

In the DaVinci Resolve 15 test, the Radeon VII is reportedly up to 27 percent faster than the RX Vega 64, and beats the GeForce RTX 2080 by 2 percent. While editing 8K footage in Adobe Premiere, the Radeon VII performs up to 29 percent and 8 percent faster than the Radeon RX Vega 64 and GeForce RTX 2080, respectively. However, the card only really begins to flex its muscles when it gets to LuxMark. The VII runs up to 62 percent faster than Vega 64 and 65 percent faster than the RTX 2080. Lastly, the Radeon VII is 27 percent faster than the Radeon RX Vega 64 in Blender.

For the gaming tests, AMD benchmarked its new flagship at three different resolutions: 2560x1440, 3440x1440 super wide, and 3840x2160. The Radeon VII beats the RTX 2080 by 11.6 percent and the RX Vega 64 by 28.9 percent in Battlefield V at 2560 x 1440. In Forza Horizon 4 at 3440 x 1440, the Radeon VII is 14.8 percent faster than the Radeon RX Vega 64 but falls behind the GeForce RTX 2080 by 1 percent. When playing Strange Brigade at 3840 x 2160, it delivers up to 42.6 percent and 19.2 percent more performance than the RX Vega 64 and RTX 2080, respectively.

AMD also compared the Radeon VII in a couple of eSports titles. The card delivered over 300 FPS at 1080p, 200 FPS at 1440p, and 100 FPS at 4K in Tom Clancy's Rainbow Six Siege. When it comes to Fornite, it's around 25 percent faster than the RX Vega 64. AMD puts a strong emphasis on streaming. The performance cost for gaming and streaming simultaneously, Fornite at 1080p, 60 FPS with Open Broadcaster Software (OBS) on a Radeon VII is just a mere 3 percent.

Availability and Pricing

Consumers can purchase the Radeon VII for $699 directly from AMD beginning on February 7. AMD will later offer the graphics card through third-party retailers from March 7. For a limited time, it'll also come with the 'Raise the Game Fully Loaded' bundle, which includes free copies of Resident Evil 2, Devil May Cry 5, and Tom Clancy's, The Division 2.

    Your comment
  • JamesSneed
    "The AMD Radeon VII comes equipped with 60 compute units, and since the graphics card adheres to the GCN standard, this tallies up to 3,840 stream processors, 240 TMUs (texture mapping units), and 64 ROPs (render output units)"

    I thought it had 128 ROP's?
  • jeffreydanielbyers
    It says 128ROPs now so a mistake must have been corrected.
  • Brian_R170