AMD Radeon RX 470 vs AMD Radeon RX 570 Comparison

The AMD Radeon RX 470 and RX 570 are remarkably close siblings, both built on the same 14 nm process with identical 5,700 million transistors and a 232 mm² die. The data shows a split decision: the RX 570 takes the newer DirectX 12 and Vulkan workloads, while the older RX 470 counters in OpenCL and Metal. The average benchmark scores are nearly inseparable—the RX 470 sits at 28,996 versus the RX 570’s 28,766—placing both in the 74th percentile of all GPUs. The real story is in the architecture and clock speed adjustments that shift performance in specific scenarios, not in a generational leap.

Head-to-Head Benchmarks

The most decisive win for the RX 570 comes in the Vulkan API, where it scores 42,752 against the RX 470’s 39,884. That is a 6.7% advantage, the largest delta in either direction across the four tests. Vulkan’s low-overhead nature tends to reward higher raw throughput, and the RX 570’s boost clock of 1244 MHz (versus 1206 MHz) likely drives this gap. The RX 570 also wins the 3DMark Steel Nomad DirectX 12 test, posting 880 points versus 842—a 4.3% lead. This is the modern gaming workload, and the newer card’s 5.095 TFLOPS FP32 compute (compared to 4.940 TFLOPS) translates directly into a measurable frame-rate advantage.

However, the RX 470 fights back in compute-heavy tests. In Geekbench Metal, it scores 41,690, which is 5.9% higher than the RX 570’s 39,349. The OpenCL result is closer but still favors the RX 470: 33,568 versus 32,084, a 4.6% edge. These are synthetic compute benchmarks that stress shader throughput and memory bandwidth differently than gaming APIs. The RX 470’s lower base clock of 926 MHz (versus 1168 MHz) suggests it may have more aggressive boost behavior under sustained compute loads, or that its memory timing characteristics—though at a lower 6.6 Gbps effective versus 7 Gbps—interact better with these specific workloads. The final tally is 2 wins each, with the RX 570 leading in gaming-oriented tests and the RX 470 leading in general-purpose compute.

Looking at the broader competitive landscape, both cards sit in a tight cluster. The RX 470’s nearest rival is the AMD Radeon RX 6800M at a 0.4% difference, while the Intel Arc A370M is 0.6% behind. The RX 570’s closest competitor is the AMD Radeon RX 6800M at -0.4%, with the Radeon R9 M295X trailing by 0.7%. Notably, the two cards are each other’s rivals: the RX 470 is 0.8% ahead of the RX 570 in average score, while the RX 570 is 0.8% behind the RX 470. This mutual proximity confirms that neither card has a commanding overall lead—the choice depends entirely on which benchmarks matter most to the user.

Where Each One Wins

The RX 570 is the pick for DirectX 12 and Vulkan gaming. Its 4.3% lead in 3DMark Steel Nomad and 6.7% lead in Geekbench Vulkan are the two most relevant results for modern game engines. The higher boost clock (1244 MHz) and memory bandwidth (224.0 GB/s versus 211.2 GB/s) give it a real edge in draw-call-heavy scenes and compute shaders that Vulkan exposes. If the workload is contemporary AAA titles or Vulkan-based emulators, the RX 570’s benchmark profile is clearly superior.

The RX 470 wins in OpenCL and Metal. The 4.6% OpenCL advantage and 5.9% Metal advantage suggest it handles general-purpose compute tasks—like rendering, physics simulations, or machine learning inference—with better efficiency. The Metal result is particularly notable for macOS users or anyone running Metal-based creative applications. The RX 470’s lower power draw (120 W versus 150 W) also makes it a more attractive option for compact or older systems where thermal headroom is limited. Its suggested PSU of 300 W (versus 450 W for the RX 570) means it can slot into pre-built machines without a power supply upgrade.

For esports titles or older DirectX 11 games, the data is less clear—neither card has a benchmark that directly measures this. But the 3DMark Steel Nomad result implies the RX 570’s architectural tweaks benefit newer rendering paths more than legacy ones. The RX 470’s higher average benchmark score (28,996 versus 28,766) suggests it may have better sustained performance in mixed workloads, even if its peak gaming performance is lower.

The Verdict

Choose the RX 570 if you prioritize gaming. The 3DMark Steel Nomad win (4.3%) and Vulkan win (6.7%) are the benchmarks that most closely predict real-world gaming performance in modern APIs. The 224.0 GB/s memory bandwidth and 5.095 TFLOPS compute are the highest in this comparison, and the 1244 MHz boost clock is the fastest. This is a straightforward recommendation for anyone building a budget gaming rig that will run DirectX 12 and Vulkan titles.

Choose the RX 470 if you need compute performance or have power constraints. Its wins in Metal (5.9%) and OpenCL (4.6%) are significant for productivity tasks. The 120 W TDP and 300 W suggested PSU make it the safer choice for upgrading an older system without swapping the power supply. It is also the slightly faster card in average benchmark score (28,996 versus 28,766), meaning it handles a broader mix of workloads more consistently.

The honest answer is that the difference is marginal. Both cards are end-of-life, both sit in the 74th percentile, and both have identical core counts (2048 shading units, 128 TMUs, 32 ROPs). The RX 570 is a tuned RX 470 with higher clocks and memory speed—nothing more. If the price difference is negligible, the RX 570’s gaming wins make it the default choice. If the RX 470 is significantly cheaper or the system has a weaker PSU, the compute wins and lower power draw justify the older card.

FAQ

Q: Which card is faster in Vulkan games?

A: The RX 570 wins Geekbench Vulkan with a score of 42,752 versus 39,884, a 6.7% advantage. This is the largest performance gap between the two cards in any benchmark.

Q: Does the RX 470 beat the RX 570 in any tests?

A: Yes, it wins Geekbench Metal (41,690 versus 39,349, a 5.9% lead) and Geekbench OpenCL (33,568 versus 32,084, a 4.6% lead). Both cards win two tests each.

Q: What is the average benchmark score difference?

A: The RX 470 has an average score of 28,996, while the RX 570 scores 28,766. The RX 470 is 0.8% ahead, making the overall performance virtually identical.

Q: Are the memory specifications different?

A: Yes, the RX 570 has faster memory at 7 Gbps effective (1750 MHz) with 224.0 GB/s bandwidth, while the RX 470 runs at 6.6 Gbps effective (1650 MHz) with 211.2 GB/s. Both use 4 GB of GDDR5 on a 256-bit bus.

Q: Which card requires a less powerful power supply?

A: The RX 470 has a suggested PSU of 300 W and a 120 W TDP, while the RX 570 needs a 450 W PSU and draws 150 W. Both use a single 6-pin power connector.

Q: Do the cards have identical core configurations?

A: Yes, both have 2048 shading units, 128 texture mapping units, and 32 ROPs. The RX 570’s advantage comes from higher clocks: 1168 MHz base and 1244 MHz boost versus 926 MHz base and 1206 MHz boost for the RX 470.

Architecture Differences

Both cards are built on the GCN 4.0 architecture, fabricated on GlobalFoundries’ 14 nm process. The silicon is identical in transistor count (5,700 million) and die size (232 mm²), yielding the same 24.6 million transistors per square millimeter. The RX 470 uses the Ellesmere chip, while the RX 570 uses Polaris 20—but these are the same physical design with different binning and clock profiles. The RX 570’s base clock is 1168 MHz versus 926 MHz for the RX 470, a 26% higher floor, while the boost clocks are closer (1244 MHz versus 1206 MHz). Memory is the other major difference: the RX 570 runs at 1750 MHz (7 Gbps effective) versus 1650 MHz (6.6 Gbps effective), pushing bandwidth from 211.2 GB/s to 224.0 GB/s.

The generations differ: the RX 470 belongs to the Arctic Islands (RX 400) family, while the RX 570 is part of Polaris (RX 500). The RX 570’s predecessor is Arctic Islands, and its successor is Vega, whereas the RX 470’s predecessor is Pirate Islands and successor is Polaris. Both cards support DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. Neither has ray tracing or tensor cores. The FP16 performance is 1:1 with FP32 on both, meaning no dedicated half-precision acceleration.

Specification Differences

| Specification | RX 470 | RX 570 |

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

| Base Clock | 926 MHz | 1168 MHz |

| Boost Clock | 1206 MHz | 1244 MHz |

| Memory Clock | 1650 MHz (6.6 Gbps) | 1750 MHz (7 Gbps) |

| Memory Bandwidth | 211.2 GB/s | 224.0 GB/s |

| Pixel Rate | 38.59 GPixel/s | 39.81 GPixel/s |

| Texture Rate | 154.4 GTexel/s | 159.2 GTexel/s |

| FP32 Compute | 4.940 TFLOPS | 5.095 TFLOPS |

| TDP | 120 W | 150 W |

| Suggested PSU | 300 W | 450 W |

| Display Outputs | 1x HDMI 2.0b, 3x DisplayPort 1.4a | 1x DVI, 1x HDMI 2.0b, 3x DisplayPort 1.4a |

| Length | 240 mm (9.4 inches) | 241 mm (9.5 inches) |

| Height | 95 mm (3.7 inches) | Not specified |

| Width | 35 mm (1.4 inches) | Not specified |

| Release Date | 2016-08-03 | 2017-04-17 |

| Launch MSRP | 179 USD | 169 USD |

The RX 570 adds a DVI output that the RX 470 lacks. Physical dimensions are nearly identical, with the RX 570 being 1 mm longer. The RX 570’s higher TDP (150 W versus 120 W) necessitates the larger suggested PSU (450 W versus 300 W). All other core specifications—shading units, TMUs, ROPs, memory size, bus width, and API support—are identical.