Introduction

XFX Radeon RX 7600 XT QICK 309 is a premium custom-design graphics card based on the new mid-range GPU from AMD. The new RX 7600 XT is designed by AMD to compete with NVIDIA’s GeForce RTX 4060, while attempting to upstage it with its larger 16 GB of memory that it wants content creators and generative AI enthusiasts to take advantage of. The XFX QICK 309 tops this up with a premium-looking board design that looks like it’s from a segment above; and is designed for low fan noise. Although designed for 1080p, these cards should be capable of 1440p with medium-high settings, or by taking advantage of FSR, Fluid Motion Frames (which works on any DirectX 11 or DirectX 12 game), the newer FSR 3 Frame Generation, or the all-encompassing HyperRX performance enhancement you can find in the Radeon Software application.

AMD’s purpose behind the RX 7600 XT appears to be to chase down the performance lead the RTX 4060 established over the RX 7600. Since it had maxed out the 6 nm Navi 33 silicon to create the RX 7600, we were wondering how AMD would go about creating the RX 7600 XT. The larger Navi 32 chiplet GPU is much more complex, and might be hard to sell around the $300-mark; and so AMD took a different approach—to push the Navi 33 to its limits. The first thing AMD did was increase the memory size to 16 GB. This is still operating at 18 GB, and yielding 288 GB/s. Secondly, it increased the GPU frequency to 2.47 GHz game clock, compared to 2.25 GHz of the RX 7600. Thirdly, it increased the power limits to 190 W, up from 165 W on the RX 7600 (which meant that the card now needs two power connectors). As a final premium touch, you’re assured to get DisplayPort 2.1 with the RX 7600 XT—partners could opt to have DisplayPort 1.4a on the RX 7600.

The new Radeon RX 7600 XT is based on the company’s latest RDNA3 architecture, which introduces several architectural improvements. The new RDNA3 Compute Unit supports dual-issue instruction rate, support for new math formats, and a 17% increase in IPC over RDNA2. The new AI accelerators prepare matrix math workloads for crunching by the stream processors, speeding up this process. The second generation AMD Ray accelerator uses several optimizations to increase ray intersection performance by 50% over the previous generation. There are other improvements, such as a decoupled GPU front-end, which runs at a higher clock speed than the Shader Engines. As we mentioned earlier, the RX 7600 XT maxes out all components on the Navi 33 silicon, which means all 32 CU, worth 2,048 stream processors, 64 AI accelerators, 32 Ray accelerators, 128 TMUs, and 64 ROPs. The 128-bit memory bus drives 16 GB of 18 Gbps GDDR6 memory.

The QICK 309, called the Speedster Quicksilver 309 in long format, combines the RX 7600 XT with a heavy aluminium fin-stack heatsink. The card is strictly two slots thick, but with a rather heavy heatsink. XFX is combining this cooling solution with a factory overclock of 2.53 GHz game clock, compared to 2.47 GHz game clock reference, while leaving the memory speed untouched. XFX is pricing the RX 7600 XT Speedster QICK 309 at $350, a $20 premium over the AMD MSRP.