AMD Radeon RX 590 vs AMD Radeon RX 6700S Comparison

Where Each One Wins

The benchmark data splits cleanly between these two AMD parts, and the Radeon RX 6700S takes every recorded head-to-head victory. Out of the two shared tests, the RX 6700S wins both, while the RX 590 records zero wins. This does not mean the older card is without merit, but it does indicate that in the specific workloads captured by the database, the newer mobile-oriented GPU holds a consistent edge.

The RX 590 finds its home in legacy API scenarios and raw rasterization throughput for its generation. Its 3DMark Steel Nomad DX12 score of 1081 places it at the 70th percentile among all GPUs in the database. That percentile suggests it still outperforms a substantial majority of recorded graphics cards, even if it lags the RX 6700S in direct comparison. The RX 590's average benchmark score of 24744 sits slightly above its nearest rival, the NVIDIA RTX A5000 Mobile, which averages 24763, a delta of just 0.1%. It also edges out the Intel Arc A350M by 0.4% and the AMD Radeon RX 6600 XT by 1.2%. Where the RX 590 wins is in consistency across older DirectX titles, a trait inherited from its GCN architecture, though the database only provides two shared tests for direct comparison.

The Radeon RX 6700S, by contrast, wins in modern API performance and compute-heavy workloads. Its 3DMark Steel Nomad DX12 score of 1451 is 25.5% higher than the RX 590's 1081. In Geekbench Metal, the RX 6700S posts 72864 against 48407, a 33.6% advantage. Its average benchmark score of 22154 places it at the 67th percentile, slightly below the RX 590's 70th percentile, but this is due to a broader test suite that includes lower scores in legacy DirectX 9, 10, and 11 tests. The RX 6700S wins where it matters for modern gaming: DirectX 12 Ultimate, Vulkan, and compute. Its Passmark G3D score of 15066 and GPU compute score of 5990 indicate strong general compute capability, while its Geekbench Vulkan score of 76150 and OpenCL score of 70965 reinforce that pattern. The RX 590 has no recorded OpenCL or Vulkan scores in the database, so its standing in those APIs is unmeasured.

Architecture Differences

The two GPUs come from different architectural generations and manufacturing processes, which explains their divergent benchmark behavior. The RX 590 uses the Polaris 30 chip, built on GCN 4.0 architecture, fabricated by GlobalFoundries on a 12 nm process. It contains 5,700 million transistors on a 232 mm² die, yielding a transistor density of 24.6 million per square millimeter. The RX 6700S uses the Navi 23 chip, built on RDNA 2.0 architecture, fabricated by TSMC on a 7 nm process. It packs 11,060 million transistors into a 237 mm² die, nearly doubling the transistor density to 46.7 million per square millimeter. The newer node and denser design give the RX 6700S a fundamental efficiency advantage.

The compute configurations differ substantially. The RX 590 fields 2304 shading units, 144 texture mapping units, and 32 raster output pipelines. The RX 6700S has fewer shading units at 1792, fewer TMUs at 112, but doubles the ROPs to 64. This ROP advantage explains the RX 6700S's much higher pixel rate of 128.0 GPixel/s versus 49.44 GPixel/s on the RX 590. The RX 6700S also includes 28 ray tracing cores, a feature entirely absent from the RX 590, which predates hardware ray tracing support. The RX 6700S supports DirectX 12 Ultimate (12_2), while the RX 590 tops out at DirectX 12 (12_0), a meaningful gap for modern titles that leverage mesh shaders or ray tracing.

Memory subsystems also differ. Both cards have 8 GB of VRAM, but the RX 590 uses GDDR5 on a 256-bit bus, delivering 256.0 GB/s of bandwidth. The RX 6700S uses GDDR6 on a 128-bit bus, delivering 224.0 GB/s. Despite the narrower bus, the RX 6700S's higher memory clock of 1750 MHz (14 Gbps effective) keeps bandwidth within 12.5% of the RX 590. The RX 590's memory runs at 2000 MHz (8 Gbps effective). Clock speeds favor the RX 6700S as well: its base clock is 1700 MHz and boost clock is 2000 MHz, with a game clock of 1890 MHz. The RX 590's base is 1469 MHz and boost is 1545 MHz. The RX 6700S also has higher FP32 throughput at 7.168 TFLOPS versus 7.119 TFLOPS, and dramatically higher FP16 at 14.34 TFLOPS (2:1 ratio) versus 7.119 TFLOPS (1:1 ratio) on the RX 590.

Power and form factor tell a stark story. The RX 590 has a TDP of 175 W, requires a dual-slot cooler, uses a single 8-pin power connector, and needs a 450 W power supply. The RX 6700S has a TDP of just 80 W, is an IGP (integrated GPU) form factor, requires no power connectors, and has no suggested PSU rating. The RX 6700S also uses PCIe 4.0 x8, while the RX 590 uses PCIe 3.0 x16. The RX 6700S supports Vulkan 1.4, a newer API version than the RX 590's Vulkan 1.3.

Head-to-Head Benchmarks

Two shared tests appear in the database, and the RX 6700S wins both by comfortable margins. In 3DMark Steel Nomad DX12, the RX 6700S scores 1451 against 1081 for the RX 590, a delta of 25.5% in favor of the newer card. This is a substantial gap in a DirectX 12 workload, and it reflects the architectural advantages of RDNA 2.0 over GCN 4.0 in modern rendering paths. The RX 6700S's 28 ray tracing cores and DirectX 12 Ultimate support likely contribute to this lead, though the database does not separate those effects.

In Geekbench Metal, the RX 6700S scores 72864 versus 48407 for the RX 590, a 33.6% advantage. This is the larger margin of the two tests, and it indicates that Apple's Metal API scales particularly well with the RDNA 2.0 architecture. The RX 6700S's higher FP16 throughput (14.34 TFLOPS versus 7.119 TFLOPS) may play a role here, as Metal compute workloads often leverage half-precision math. The RX 590's FP16 performance is identical to its FP32, reflecting the 1:1 ratio of GCN, while the RX 6700S doubles its FP16 rate via the 2:1 ratio of RDNA 2.0.

The RX 6700S also has a broader benchmark footprint in the database, with additional scores for Geekbench OpenCL (70965), Geekbench Vulkan (76150), and five Passmark tests. Its Passmark G3D score of 15066 and GPU compute score of 5990 are strong. The RX 590 only has the two shared tests recorded, so its performance in OpenCL, Vulkan, and Passmark is unknown. The RX 6700S's lowest recorded scores are in legacy DirectX tests: Passmark DirectX 9 at 195, DirectX 10 at 87, DirectX 11 at 134, and DirectX 12 at 58. These low numbers suggest that RDNA 2.0 does not prioritize older API paths, while the RX 590's GCN architecture, designed for that era, likely handles them better. However, without recorded RX 590 scores in those tests, this remains inference from the data available.

FAQ

Q: Which GPU wins in DirectX 12 performance?

A: The AMD Radeon RX 6700S wins. In 3DMark Steel Nomad DX12, it scores 1451 versus 1081 for the RX 590, a 25.5% advantage.

Q: What is the difference in average benchmark scores?

A: The RX 590 has a higher average benchmark score of 24744, while the RX 6700S averages 22154. The RX 590 sits at the 70th percentile, the RX 6700S at the 67th.

Q: Does the RX 6700S support ray tracing?

A: Yes. The RX 6700S includes 28 ray tracing cores and supports DirectX 12 Ultimate (12_2). The RX 590 has no ray tracing cores and supports DirectX 12 (12_0).

Q: How do the memory bandwidths compare?

A: The RX 590 has 256.0 GB/s from GDDR5 on a 256-bit bus. The RX 6700S has 224.0 GB/s from GDDR6 on a 128-bit bus. The RX 6700S has a 12.5% bandwidth deficit despite the faster memory type.

Q: Which card has lower power consumption?

A: The RX 6700S has a TDP of 80 W and requires no power connectors. The RX 590 has a TDP of 175 W and requires a single 8-pin power connector and a 450 W power supply.

Q: What are the nearest rivals for each card?

A: The RX 590's nearest rival is the NVIDIA RTX A5000 Mobile, with a delta of 0.1%. The RX 6700S's nearest rival is the AMD Radeon RX 5700, with a delta of 0.1%.

The Verdict

The data points to a clear choice for modern workloads: the AMD Radeon RX 6700S. It wins both shared benchmarks by 25.5% and 33.6%, offers ray tracing support, consumes less than half the power (80 W versus 175 W), and supports newer API standards including DirectX 12 Ultimate and Vulkan 1.4. For anyone building a system around DirectX 12 games, Vulkan titles, or compute-heavy applications, the RX 6700S is the superior option based on recorded measurements.

The RX 590 retains relevance for legacy scenarios and for users who prioritize raw rasterization in older API titles. Its higher average benchmark score of 24744 and 70th percentile ranking suggest it remains a capable card in the broader database context. Its 256-bit memory bus provides higher bandwidth (256.0 GB/s) than the RX 6700S, which may benefit certain memory-intensive workloads that do not appear in the shared test suite. The RX 590 also has a higher shading unit count (2304 versus 1792), which can help in compute tasks that scale with ALU count rather than architectural efficiency.

The choice depends on the workload. The RX 6700S wins every recorded head-to-head test, but the RX 590's higher average score and percentile indicate that it holds its own against a wider field of GPUs. The RX 590's nearest rival, the NVIDIA RTX A5000 Mobile, is nearly identical in average score (24763 versus 24744, a 0.1% delta). The RX 6700S's nearest rival, the AMD Radeon RX 5700, also sits within 0.1%. Both cards are competitive within their respective tiers, but the RX 6700S's modern architecture and efficiency make it the better pick for current and future software.

Specification Differences

| Specification | AMD Radeon RX 590 | AMD Radeon RX 6700S |

|---|---|---|

| Architecture | GCN 4.0 | RDNA 2.0 |

| Process Node | 12 nm (GlobalFoundries) | 7 nm (TSMC) |

| Transistors | 5,700 million | 11,060 million |

| Die Size | 232 mm² | 237 mm² |

| Transistor Density | 24.6M / mm² | 46.7M / mm² |

| Base Clock | 1469 MHz | 1700 MHz |

| Boost Clock | 1545 MHz | 2000 MHz |

| Game Clock | None recorded | 1890 MHz |

| Memory Clock | 2000 MHz (8 Gbps effective) | 1750 MHz (14 Gbps effective) |

| Memory Type | GDDR5 | GDDR6 |

| Memory Bus Width | 256 bit | 128 bit |

| Memory Bandwidth | 256.0 GB/s | 224.0 GB/s |

| Shading Units | 2304 | 1792 |

| TMUs | 144 | 112 |

| ROPs | 32 | 64 |

| Ray Tracing Cores | None | 28 |

| Pixel Rate | 49.44 GPixel/s | 128.0 GPixel/s |

| Texture Rate | 222.5 GTexel/s | 224.0 GTexel/s |

| FP32 | 7.119 TFLOPS | 7.168 TFLOPS |

| FP16 | 7.119 TFLOPS (1:1) | 14.34 TFLOPS (2:1) |

| TDP | 175 W | 80 W |

| Slot Width | Dual-slot | IGP |

| Power Connectors | 1x 8-pin | None |

| Suggested PSU | 450 W | None recorded |

| Bus Interface | PCIe 3.0 x16 | PCIe 4.0 x8 |

| DirectX | 12 (12_0) | 12 Ultimate (12_2) |

| Vulkan | 1.3 | 1.4 |

| Display Outputs | 1x HDMI 2.0b, 3x DisplayPort 1.4a | Portable Device Dependent |

| Release Date | 2018-11-14 | 2022-01-03 |

| Production Status | End-of-life | End-of-life |

| Launch MSRP | 279 USD | None recorded |