Unlock Ultimate Gaming Experience: AMD’s RDNA 4 GPUs Revolutionize Encoding and Ray Tracing Performance!

The big picture: AMD has shared the specs, pricing, and performance of the Radeon RX 9070 and RX 9070 XT graphics cards. We’re eager for full reviews soon. In the meantime, let’s dive into some important details about improvements in the new Radeon’s video encoding quality, a crucial feature often overlooked.

Historically, AMD’s GPU encoders faced criticism for their video recording quality during game streaming. This often pushed users towards Nvidia for better encoding options.

With their new RDNA 4 architecture, AMD claims to have made big strides in encoding quality. They highlighted impressive examples of 1080p H.264 and HEVC at 6 megabits per second, which is a standard setup for many users. The visual quality has reportedly seen a noteworthy boost.

We’ll have to see if these improvements hold in different scenarios. Historically, AMD’s focus on encoding quality leaned more toward new formats like AV1. Now, with RDNA 4, they seem to emphasize real-world performance gains, suggesting a greater confidence in their encoders.

AMD reports a 25% increase in H.264 low-latency encoding quality and an 11% boost for HEVC. There’s better AV1 encoding with support for B-frames and a 30% performance leap for 720p encoding. These figures likely align with VMAF scoring metrics.

Aside from encoding, there are other exciting upgrades. The ray tracing core now has two intersection engines. This change significantly boosts the speed of ray-box and ray-triangle interactions. A new ray transform block also helps ease the workload on shaders.

The BVH (Bounding Volume Hierarchy) has also doubled in width, and many improvements have been made to the ray tracing setup. This is why RDNA 4 seems to have better ray tracing performance compared to rasterization improvements.

Compute, Memory, and Display Enhancements

The compute engine comes with various optimizations and supports a PCIe 5.0 x16 interface. It uses GDDR6 with a 256-bit memory bus. AMD claims improved memory compression, and the GPUs now offer 16GB of VRAM, suitable for most current games.

However, the display engine has some mixed reviews. Although it supports DisplayPort 2.1, its capabilities haven’t changed, with a max bandwidth of UHBR 13.5, less than Nvidia’s newer standards for 4K 240Hz displays.

The cards also support HDMI 2.1b. AMD says they’ve reduced power use during idle times for multi-monitor setups, and video frame scheduling can now be managed by the GPU.

The Navi 48 die measures 357mm² and is built on TSMC’s 4nm process, containing 53.9 billion transistors. This makes it 5% smaller than Nvidia’s RTX 5080 die but with 18% more transistors, showing a more compact design. However, Nvidia still has the advantage in die size and efficiency. The RTX 5080 is expected to be around 15% faster in rasterization and over 50% faster in ray tracing than what AMD claims for the RX 9070 XT, all while being less power-hungry.

A few other notes: AMD is not making reference models for the RX 9070 XT or RX 9070. Instead, designs will come only from partners like ASUS, Gigabyte, and PowerColor.

These cards should be available starting March 6th, as they’ve been ready since early January.

Also, AMD is launching AFMF 2.1, a new version of its frame generation technology. This will be available for the Radeon RX 6000 series and newer. It aims for better image quality and includes Radeon Image Sharpening that can enhance any game or app at the driver level. There are a few AI features too, but they’re not revolutionary.