AMD Radeon RX 470 vs AMD Radeon RX 590 Comparison

AMD Radeon RX 470 and AMD Radeon RX 590 are both end-of-life GPUs from AMD built on the GCN 4.0 architecture, but they serve very different performance tiers. The RX 470 is the older, lower-clocked card from the Arctic Islands generation, while the RX 590 is a later Polaris 30 refresh with substantially higher clock speeds and more memory. The benchmark data shows a clear hierarchy: the RX 590 wins both recorded head-to-head tests, yet the RX 470 posts a higher average benchmark score across all tested workloads. This apparent contradiction stems from the limited overlap in benchmark suites and the specific strengths of each card. The following sections break down where each GPU wins, what the numbers mean for real-world use, and which card makes sense for different build scenarios.

Where Each One Wins

The RX 590 wins outright in every directly comparable benchmark. In 3DMark Steel Nomad DX12, it scores 1081 versus the RX 470's 842, a decisive 22.1% advantage. In Geekbench Metal, the RX 590 leads with 48407 against 41690, a 13.9% gap. These are the only two tests where both cards were measured under identical conditions, and the RX 590 dominates both. For gamers targeting DX12 titles or Metal-based applications, the RX 590 is the stronger choice by a wide margin.

The RX 470, however, shows unexpected strength in the broader benchmark database. Its average benchmark score across all recorded tests is 28996, which is 17.2% higher than the RX 590's 24744 average. This is driven by the RX 470's strong showings in Geekbench OpenCL (33568) and Geekbench Vulkan (39884), tests that the RX 590 does not have recorded scores for. These compute-oriented workloads favor the RX 470's configuration, suggesting that for OpenCL or Vulkan compute tasks, the older card may actually outperform its successor.

In terms of percentile ranking against all GPUs, the RX 470 sits at the 74th percentile, while the RX 590 sits at the 70th. This means the RX 470 outperforms a larger fraction of the GPU population in the database's average score metric, despite losing the direct head-to-head matchups. The RX 590's higher absolute performance in gaming workloads does not translate to a better overall percentile because the database includes a mix of compute and graphics tests, and the RX 470's compute scores pull its average up significantly.

FAQ

Q: Which card is faster in DX12 gaming?

A: The RX 590 is significantly faster. In 3DMark Steel Nomad DX12, it scores 1081 versus the RX 470's 842, a 22.1% advantage. This is the largest margin between the two cards in any recorded benchmark.

Q: Does the RX 470 have any performance advantage over the RX 590?

A: Yes, in average benchmark score. The RX 470 averages 28996 across all recorded tests, while the RX 590 averages 24744. This is due to the RX 470's strong Geekbench OpenCL score of 33568 and Vulkan score of 39884, tests that the RX 590 lacks recorded results for.

Q: How do their percentile rankings compare?

A: The RX 470 ranks at the 74th percentile against all GPUs, while the RX 590 ranks at the 70th. Despite losing both head-to-head tests, the RX 470's compute-heavy benchmark results give it a better overall standing in the database.

Q: What is the memory capacity difference?

A: The RX 590 comes with 8 GB of GDDR5 memory, while the RX 470 has 4 GB. Both use a 256-bit bus, but the RX 590's memory runs at 2000 MHz (8 Gbps effective) for 256.0 GB/s bandwidth, versus 1650 MHz (6.6 Gbps effective) for 211.2 GB/s on the RX 470.

Q: Which card has higher clock speeds?

A: The RX 590 has substantially higher clocks. Its base clock is 1469 MHz with a boost of 1545 MHz, compared to the RX 470's 926 MHz base and 1206 MHz boost. This clock advantage is a primary driver of the RX 590's gaming performance lead.

Q: Are both cards the same physical size?

A: Nearly identical in length. The RX 470 measures 240 mm (9.4 inches), while the RX 590 measures 241 mm (9.5 inches). The RX 470 is dual-slot with a 35 mm width, while the RX 590's height and width are not recorded in the database.

Head-to-Head Benchmarks

The 3DMark Steel Nomad DX12 test provides the clearest performance separation. The RX 590 scores 1081, which is 239 points higher than the RX 470's 842. This translates to a 22.1% delta, the largest gap between the two cards in any shared benchmark. This test is heavily graphics-bound, and the RX 590's higher shader count and clock speed drive a substantial rasterization advantage. For any DX12 workload, the RX 590 is the definitive winner.

The Geekbench Metal benchmark shows a narrower but still decisive margin. The RX 590 scores 48407 versus the RX 470's 41690, a difference of 6717 points or 13.9%. Metal is Apple's graphics API, and this result indicates the RX 590 maintains its lead even in API-specific scenarios. The smaller percentage gap compared to 3DMark suggests the RX 590's advantages are slightly less pronounced in Metal, but it still wins comfortably.

The database records no head-to-head results for Geekbench OpenCL or Geekbench Vulkan, meaning there is no direct comparison for those workloads. However, the RX 470's standalone scores in these tests (33568 OpenCL, 39884 Vulkan) are high enough to push its average benchmark score to 28996, which exceeds the RX 590's average of 24744. This implies that if such head-to-head tests were run, the RX 470 could potentially win or at least compete closely in compute-heavy scenarios, though this remains unverified.

The wins tally is 2 for the RX 590 and 0 for the RX 470 in direct comparisons. Every measured matchup favors the newer card, and the deltas are consistent, ranging from 13.9% to 22.1%. The RX 470's only consolation is its better average score and percentile ranking, but those figures are influenced by non-overlapping benchmarks.

Specification Differences

The RX 590 uses the Polaris 30 chip on a 12 nm process, while the RX 470 uses the Ellesmere chip on a 14 nm process. Both are manufactured by GlobalFoundries and contain 5,700 million transistors on a 232 mm² die, giving both a transistor density of 24.6M per mm². The die size and transistor count are identical, so the performance gap comes entirely from clock speeds and configuration changes.

Clock speeds differ dramatically. The RX 470 runs at 926 MHz base and 1206 MHz boost, while the RX 590 runs at 1469 MHz base and 1545 MHz boost. This is a 543 MHz difference at base clock and 339 MHz at boost. Memory clocks also differ: the RX 470 uses 1650 MHz (6.6 Gbps effective), while the RX 590 uses 2000 MHz (8 Gbps effective), resulting in bandwidth of 211.2 GB/s versus 256.0 GB/s.

Memory capacity is the most user-visible difference: 4 GB on the RX 470 versus 8 GB on the RX 590. Both use GDDR5 with a 256-bit bus. The RX 590 also has more execution resources: 2304 shading units and 144 texture mapping units, versus 2048 shading units and 128 TMUs on the RX 470. Both have 32 ROPs.

Pixel and texture rates scale accordingly. The RX 590 achieves 49.44 GPixel/s and 222.5 GTexel/s, while the RX 470 achieves 38.59 GPixel/s and 154.4 GTexel/s. FP32 compute is 7.119 TFLOPS on the RX 590 versus 4.940 TFLOPS on the RX 470, with both offering 1:1 FP16. Power requirements differ as well: the RX 470 has a 120 W TDP with a single 6-pin connector and a 300 W suggested PSU, while the RX 590 has a 175 W TDP with a single 8-pin connector and a 450 W suggested PSU.

Physical dimensions are close: the RX 470 is 240 mm long, 95 mm tall, and 35 mm wide, while the RX 590 is 241 mm long with unrecorded height and width. Both are dual-slot and use PCIe 3.0 x16. Display outputs are identical: 1x HDMI 2.0b and 3x DisplayPort 1.4a. Both support DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3.

Architecture Differences

Both cards share the GCN 4.0 architecture, so their fundamental design is identical. The RX 590 is not a new architecture but a refined version of the same Polaris design, built on a smaller 12 nm process versus the RX 470's 14 nm. This process shrink allows the RX 590 to run at much higher clocks while keeping power manageable, though its TDP still rises from 120 W to 175 W.

The chip difference matters: the RX 470 uses Ellesmere, while the RX 590 uses Polaris 30. Despite the name change, both chips are 232 mm² with 5,700 million transistors, confirming that the RX 590 is essentially a higher-clocked, more fully enabled version of the same silicon. The shading unit count increases from 2048 to 2304, and TMUs from 128 to 144, indicating the RX 590 unlocks additional compute units that were likely disabled on the RX 470.

Memory architecture is identical in bus width (256-bit) but the RX 590's faster memory clock and double capacity (8 GB versus 4 GB) give it a meaningful bandwidth and capacity advantage. This matters for modern games that exceed 4 GB VRAM usage, where the RX 470 would hit a hard limit.

Both cards use the same API support and display outputs, and neither has ray tracing or tensor cores. The RX 590's generation is listed as Polaris (RX 500), while the RX 470's is Arctic Islands (RX 400), but both are GCN 4.0 at heart. The RX 590's successor is listed as Vega, while the RX 470's predecessor is Pirate Islands and successor is Polaris. The RX 590 simply extracts more from the same architecture through process improvements and higher clocks.

The Verdict

The RX 590 is the clear winner for gaming performance. It beats the RX 470 by 22.1% in 3DMark Steel Nomad DX12 and by 13.9% in Geekbench Metal, with both margins being substantial enough to feel in real-world frame rates. Its 8 GB VRAM and higher memory bandwidth (256.0 GB/s versus 211.2 GB/s) also future-proof it better for texture-heavy titles. If the choice is between these two for a gaming rig, the RX 590 is the only sensible pick.

The RX 470, however, is not without merit in the database. Its average benchmark score of 28996 exceeds the RX 590's 24744, and it ranks at the 74th percentile versus the RX 590's 70th. This is driven by its strong Geekbench OpenCL and Vulkan results, which suggest the RX 470 may be competitive or even superior in compute workloads that leverage those APIs. For users who prioritize OpenCL compute, video encoding, or Vulkan-based non-gaming tasks, the RX 470's lower power draw (120 W versus 175 W) and simpler power connector (6-pin versus 8-pin) make it an easier fit for smaller power supplies.

The RX 590's nearest rivals in the database include the NVIDIA RTX A5000 Mobile (within 0.1% of its average score) and the Intel Arc A350M (0.4% behind), while the RX 470 sits near the AMD Radeon RX 6800M and Intel Arc A370M. These comparisons reinforce that the RX 470's average score is inflated by compute tests, while the RX 590's lower average reflects a lack of those same compute results.

Ultimately, the RX 590 wins on performance where it matters most: gaming. The RX 470 wins only on aggregate compute scores. For a builder with a 450 W PSU and an 8-pin connector, the RX 590 delivers a meaningful performance uplift across every shared benchmark. For a builder with a 300 W PSU or a focus on OpenCL/Vulkan workloads, the RX 470 remains a viable, lower-power alternative. There is no scenario where the RX 470 outperforms the RX 590 in a direct graphics test, so the verdict is straightforward: choose the RX 590 for gaming, consider the RX 470 only for specific compute use cases.