The Radeon RX 7600 features a Navi 33 XL GPU equipped with 13.3 billion transistors engraved in 6 nm. It uses the RDNA 3 cores of the RX 7900 XT/XTX, but from a conceptual point of view, we are very far from the complexity of the Navi 31/32 which use a stack of tiles (chiplets) engraved in 5 nm. The Navi 33 XL is therefore a traditional chip, called monolithic, less expensive to produce, because of a greater engraving finesse than that of the RX 7900.
To sum up, the GPU of the RX 7600 is an RX 7900 XTX GPU, whose characteristics would have been divided by three. This gives 32 Compute Units, 32 units dedicated to raytracing and 64 units dedicated to AI. Finally, 2048 classic calculation units complete this GPU (6144 for the RX 7900 XTX).
The Radeon RX 6600 GPU has 28 Compute Units, 28 units dedicated to raytracing and 1792 classical calculation units, while being devoid of artificial intelligence units. Its in-game frequency is given at 2250 MHz, which is an additional 206 MHz compared to the RX 6600.**
The GPU also includes 32 MB of Infinity Cache – which speeds up processing – accompanied by 8 GB of GDDR6 at 18 Gb / s on a 128-bit bus. On this point, apart from the memory speed, the characteristics are identical to those of the RX 6600.
The thermal envelope of the Radeon RX 7600 is on the other hand revised upwards, despite a lower engraving fineness. At 165W, that’s 33W more than the RX 6600.
