AMD Radeon RX 590 vs AMD Radeon RX 6600 XT Comparison

The AMD Radeon RX 590 and AMD Radeon RX 6600 XT are separated by nearly three years and a complete architectural overhaul, yet the benchmark database places them extremely close in average scores. The RX 590 averages 24,744 points, while the RX 6600 XT averages 24,442 points—a mere 1.2% difference in favor of the older card. However, this aggregate parity masks a decisive split in individual workloads: the RX 6600 XT wins both head-to-head tests by wide margins, suggesting the RX 590’s average is bolstered by its performance in a broader legacy benchmark set. The data shows a generational leap in modern compute efficiency, with the newer card dominating the two tests where they are directly compared.

Where Each One Wins

The head-to-head data is unambiguous: the RX 6600 XT wins every direct comparison. In the 3DMark Steel Nomad DX12 test, the RX 6600 XT scores 1,804 against the RX 590’s 1,081, a delta of -40.1% from the perspective of the older card—meaning the RX 590 trails by roughly 40%. This is a modern DirectX 12 workload, and the RDNA 2 architecture’s advantage is stark. Similarly, in Geekbench Metal, the RX 6600 XT posts 88,720 points versus 48,407 for the RX 590, a -45.4% delta. The RX 590’s Metal score is barely half of the newer card’s, indicating that Apple’s compute API heavily favors the newer hardware’s feature set and throughput.

The RX 590’s wins are not in these head-to-head tests but in its aggregate profile. Its average benchmark score of 24,744 places it slightly above the RX 6600 XT’s 24,442, and its nearest rivals include the NVIDIA RTX A5000 Mobile (24,763, delta -0.1%) and the Intel Arc A350M (24,647, delta 0.4%). The RX 6600 XT, by contrast, sits just below both of those same rivals, with deltas of -0.8% to the Arc A350M and +0.7% to the GTX 1630. In practice, this means the RX 590 holds a slight edge in legacy or mixed-workload benchmarks that are not part of the direct comparison set, while the RX 6600 XT is the clear winner in the two modern, compute-heavy tests that are.

For use-case selection, the pattern is straightforward: the RX 6600 XT is the choice for DirectX 12 gaming (as evidenced by Steel Nomad) and Metal-based applications (as evidenced by Geekbench). The RX 590, with its older GCN architecture, still competes in overall average score but offers no modern head-to-head victory. Users running older DirectX 11 or OpenGL titles may find the RX 590’s aggregate standing respectable, but the data shows no scenario where it beats the RX 6600 XT in a shared, current benchmark.

Architecture Differences

The two cards represent AMD’s two major pre-RDNA designs. The RX 590 uses the Polaris 30 chip on GCN 4.0 architecture, fabricated on GlobalFoundries’ 12 nm process. The RX 6600 XT uses the Navi 23 chip on RDNA 2.0, built on TSMC’s 7 nm node. The process shrink is dramatic: the RX 590 packs 5,700 million transistors into a 232 mm² die, yielding a density of 24.6M transistors per mm². The RX 6600 XT crams 11,060 million transistors into a slightly larger 237 mm² die, achieving 46.7M per mm²—nearly double the density. This explains how the newer card delivers more performance at a lower TDP.

Architecturally, the differences are foundational. The RX 590 has 2,304 shading units, 144 TMUs, and 32 ROPs. The RX 6600 XT has fewer shading units (2,048) and TMUs (128), but doubles the ROPs to 64. The RX 6600 XT also introduces 32 dedicated ray tracing cores, a feature entirely absent from the RX 590. Clock speeds reflect the process advantage: the RX 590 runs at a 1,469 MHz base and 1,545 MHz boost, while the RX 6600 XT runs at 1,968 MHz base, 2,359 MHz game clock, and 2,589 MHz boost. The result is a massive throughput difference: the RX 6600 XT delivers 10.60 TFLOPS FP32 versus 7.119 TFLOPS for the RX 590. The FP16 gap is even larger—21.21 TFLOPS (2:1 ratio) versus 7.119 TFLOPS (1:1).

Memory systems also diverge. Both have 8 GB, but the RX 590 uses GDDR5 on a 256-bit bus at 8 Gbps effective, while the RX 6600 XT uses GDDR6 on a 128-bit bus at 16 Gbps effective. Interestingly, both achieve exactly 256.0 GB/s of bandwidth, meaning the narrower bus is compensated by faster memory. The RX 6600 XT also supports PCIe 4.0 x8, while the RX 590 is limited to PCIe 3.0 x16. API support differs as well: the RX 590 supports DirectX 12 (12_0), while the RX 6600 XT supports DirectX 12 Ultimate (12_2) and Vulkan 1.4 versus 1.3 for the older card.

FAQ

Q: Which card has a higher average benchmark score?

A: The RX 590 has a slightly higher average, at 24,744 points, compared to the RX 6600 XT’s 24,442 points. However, the RX 6600 XT wins both direct head-to-head tests.

Q: How much faster is the RX 6600 XT in the 3DMark Steel Nomad DX12 test?

A: The RX 6600 XT scores 1,804 versus the RX 590’s 1,081, which is a -40.1% delta for the RX 590—meaning the older card is about 40% slower in this test.

Q: Do both cards have the same memory bandwidth?

A: Yes, both achieve 256.0 GB/s. The RX 590 uses GDDR5 on a 256-bit bus at 8 Gbps effective, while the RX 6600 XT uses GDDR6 on a 128-bit bus at 16 Gbps effective.

Q: Which card has ray tracing support?

A: Only the RX 6600 XT, which includes 32 ray tracing cores. The RX 590 has no RT cores.

Q: What is the process node difference?

A: The RX 590 is built on GlobalFoundries’ 12 nm node, while the RX 6600 XT uses TSMC’s 7 nm node. This allows the RX 6600 XT to fit nearly double the transistor density (46.7M vs 24.6M per mm²).

Q: What is the launch MSRP of each card?

A: The RX 590 had a launch MSRP of 279 USD, while the RX 6600 XT launched at 379 USD.

Specification Differences

The most significant specification gaps are in process technology, architecture, and clock speeds. The RX 590 uses GCN 4.0 on 12 nm, while the RX 6600 XT uses RDNA 2.0 on 7 nm. Transistor count nearly doubles from 5,700 million to 11,060 million, despite a similar die size (232 mm² vs 237 mm²). Clock speeds are substantially higher on the newer card: base 1,469 MHz vs 1,968 MHz, boost 1,545 MHz vs 2,589 MHz. The RX 6600 XT also has a defined game clock of 2,359 MHz, which the RX 590 lacks.

Compute resources differ in configuration. The RX 590 has more shading units (2,304 vs 2,048) and TMUs (144 vs 128), but the RX 6600 XT has double the ROPs (64 vs 32) and adds 32 RT cores. Pixel rate jumps from 49.44 GPixel/s to 165.7 GPixel/s, and texture rate from 222.5 GTexel/s to 331.4 GTexel/s. FP32 performance rises from 7.119 TFLOPS to 10.60 TFLOPS, and FP16 from 7.119 TFLOPS (1:1) to 21.21 TFLOPS (2:1).

Memory type, bus width, and effective speed differ, though bandwidth is identical at 256.0 GB/s. The RX 6600 XT uses GDDR6 on a 128-bit bus at 16 Gbps, versus GDDR5 on a 256-bit bus at 8 Gbps for the RX 590. Bus interface changes from PCIe 3.0 x16 to PCIe 4.0 x8. TDP is lower on the newer card: 160 W vs 175 W. Display outputs are similar, but the RX 6600 XT upgrades to HDMI 2.1 from HDMI 2.0b. DirectX support advances from 12 (12_0) to 12 Ultimate (12_2), and Vulkan from 1.3 to 1.4. Physical dimensions differ: the RX 590 is 241 mm long, while the RX 6600 XT is 190 mm long, 110 mm tall, and 40 mm wide. Release dates are November 14, 2018 for the RX 590 and July 29, 2021 for the RX 6600 XT.

Head-to-Head Benchmarks

The two available direct comparisons tell a consistent story of generational dominance. In the 3DMark Steel Nomad DX12 test, the RX 6600 XT scores 1,804, which is 723 points higher than the RX 590’s 1,081. The delta of -40.1% for the RX 590 indicates that the older card delivers only about 60% of the newer card’s performance in this workload. This is a demanding, modern DirectX 12 test, and the RX 6600 XT’s RDNA 2 architecture with its higher clocks (2,589 MHz boost vs 1,545 MHz) and superior pixel rate (165.7 GPixel/s vs 49.44 GPixel/s) provides a decisive advantage. The RX 590’s extra shading units (2,304 vs 2,048) cannot compensate for the massive clock and efficiency gap.

The Geekbench Metal test shows an even larger margin. The RX 6600 XT scores 88,720, while the RX 590 scores 48,407—a -45.4% delta. This means the RX 590 achieves roughly 55% of the newer card’s Metal performance. Metal is a compute-oriented API, and the RX 6600 XT’s advantages in FP32 (10.60 TFLOPS vs 7.119 TFLOPS) and FP16 (21.21 TFLOPS vs 7.119 TFLOPS) are directly exposed here. The 2:1 FP16 ratio of the RDNA 2 card gives it a particular edge in mixed-precision workloads.

The RX 6600 XT’s other benchmark scores—Geekbench OpenCL at 80,503, Vulkan at 72,417, and Passmark G3D at 16,461—are not directly compared to the RX 590, but they contextualize its dominance. The RX 590’s only benchmark beyond the head-to-head tests is its Geekbench Metal score (48,407), which is the same value used in the comparison. Across the two shared tests, the RX 6600 XT wins by an average margin of roughly 43%, cementing its status as the superior modern performer. The RX 590’s aggregate average score of 24,744, which edges out the RX 6600 XT’s 24,442, is thus a quirk of the benchmark mix rather than evidence of real-world superiority in current applications.