AMD Radeon RX 560 vs NVIDIA GeForce RTX 4070 GDDR6 Comparison
AMD Radeon RX 560
GeForce RTX 4070 GDDR6
PERFORMANCE BENCHMARKS
Analysis: AMD Radeon RX 560 vs NVIDIA GeForce RTX 4070 GDDR6
The AMD Radeon RX 560 and the NVIDIA GeForce RTX 4070 GDDR6 occupy opposite ends of the GPU spectrum, separated by seven years of architecture evolution and a massive gulf in intended use. The RX 560 is a 2017 entry-level card built on a 14 nm process, while the RTX 4070 GDDR6 is a 2024 high-end offering on a 5 nm node. This analysis uses only the provided benchmark data, which is limited to a single 3DMark test for the RTX 4070 GDDR6 and a broader suite for the RX 560, so direct comparisons must be interpreted cautiously.
The Verdict
The data paints a stark picture of generational and performance-class divide. The RTX 4070 GDDR6, despite being an end-of-life product, is the clear choice for anyone needing modern, high-end rasterization and ray tracing performance. Its single benchmark result in 3DMark Steel Nomad DX12 is 4334.5, which is nearly 18% higher than the RX 560’s highest recorded score of 3671 in Passmark G3D. The RTX 4070 GDDR6 also brings 12 GB of GDDR6 memory, 5888 shading units, 46 RT cores, and 184 tensor cores, making it suitable for 1440p and 4K gaming, ray tracing, and AI workloads.
The RX 560, however, is not without its niche. Its 75 W TDP and lack of power connectors make it a drop-in upgrade for older, pre-built systems with weak power supplies. Its 4 GB of GDDR5 memory and 112.0 GB/s bandwidth are enough for 1080p esports titles and light productivity. Its Passmark scores across multiple DirectX versions (16 in DX10, 25 in DX11, 21 in DX12, 57 in DX9) show it handles legacy APIs better than modern ones, which is typical for its era. The RTX 4070 GDDR6, in contrast, supports DirectX 12 Ultimate (12_2) and Vulkan 1.4, positioning it for current and future software.
In short: pick the RTX 4070 GDDR6 for any serious gaming or creative build. Pick the RX 560 only if you have a constrained power budget, a legacy system, and modest 1080p expectations. The benchmark data shows the RTX 4070 GDDR6 is the superior performer, but the RX 560 remains a functional, low-power option.
FAQ
Q: Which GPU has the higher average benchmark score?
A: The AMD Radeon RX 560 has a higher average benchmark score of 4569, compared to the NVIDIA GeForce RTX 4070 GDDR6’s 4335. However, this is misleading. The RX 560’s average is computed from nine diverse tests (Geekbench Metal/OpenCL and Passmark DX9-12, G2D, G3D, compute), while the RTX 4070 GDDR6 has only one 3DMark Steel Nomad DX12 score, so the averages are not directly comparable.
Q: How do their percentiles compare?
A: The RX 560 sits at the 26th percentile of all GPUs, while the RTX 4070 GDDR6 is at the 25th percentile. This means both are in the bottom quarter of the performance distribution, which is surprising for the RTX 4070 GDDR6. This is likely due to the single benchmark result pulling its average down, and the percentile data may be skewed by the limited test set.
Q: What are the nearest rivals for each card?
A: The RX 560’s nearest rivals include the AMD Radeon R5 M230 (avg 4577, delta -0.2%), Intel HD Graphics P530 (avg 4560, delta 0.2%), NVIDIA Quadro M3000M (avg 4621, delta -1.1%), and NVIDIA GeForce GTX 970M (avg 4628, delta -1.3%). The RTX 4070 GDDR6’s rivals are the Intel Iris Pro Graphics 5200 (avg 4360, delta -0.6%), AMD FirePro W2100 (avg 4295, delta 0.9%), NVIDIA GeForce 930M (avg 4388, delta -1.2%), and NVIDIA GeForce GTX 460M (avg 4282, delta 1.2%). The RTX 4070 GDDR6’s rivals are all much older, lower-end cards, highlighting how its single benchmark result is an outlier.
Q: Which card supports ray tracing?
A: Only the NVIDIA GeForce RTX 4070 GDDR6 has dedicated RT cores (46 of them) and tensor cores (184), enabling ray tracing and DLSS. The AMD Radeon RX 560 has no RT or tensor cores, so it cannot accelerate these workloads.
Q: What are the launch MSRPs?
A: The AMD Radeon RX 560 launched at 99 USD. The NVIDIA GeForce RTX 4070 GDDR6 launched at 599 USD.
Q: Which card has a higher memory bandwidth?
A: The RTX 4070 GDDR6 has a memory bandwidth of 480.0 GB/s, which is over four times higher than the RX 560’s 112.0 GB/s. This is a direct result of the RTX 4070 GDDR6’s 192-bit bus and GDDR6 memory running at 20 Gbps effective, versus the RX 560’s 128-bit bus and GDDR5 at 7 Gbps effective.
Architecture Differences
The two GPUs are built on fundamentally different architectures. The RX 560 uses AMD’s GCN 4.0 architecture on a 14 nm process at GlobalFoundries. It is based on the Polaris 21 chip, which packs 3,000 million transistors into a 123 mm² die, giving a transistor density of 24.4 million per square millimeter. The GCN 4.0 design is a mature, scalar-based architecture that lacks dedicated hardware for ray tracing or tensor operations. Its 1024 shading units, 64 texture mapping units, and 16 ROPs are arranged in a traditional GCN layout, and it supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3.
In contrast, the RTX 4070 GDDR6 is based on NVIDIA’s Ada Lovelace architecture, fabricated on a 5 nm process at TSMC. The AD104 chip contains 35,800 million transistors on a 294 mm² die, yielding a density of 121.8 million per square millimeter — five times denser than the RX 560. Ada Lovelace is a modern, deeply pipelined architecture designed for real-time ray tracing, AI acceleration, and high-throughput compute. It includes 46 RT cores and 184 tensor cores, which are absent on the RX 560. The RTX 4070 GDDR6 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, adding features like mesh shaders and variable rate shading that the RX 560 cannot handle.
The process node difference alone explains much of the performance gap. The 5 nm node allows the RTX 4070 GDDR6 to run at a boost clock of 2475 MHz, nearly double the RX 560’s 1275 MHz boost, while also fitting 5888 shading units. The architectural leap from GCN 4.0 to Ada Lovelace also brings improvements in instruction-level parallelism, cache hierarchy, and scheduling efficiency, which are not captured in raw specs but contribute to the benchmark gap.
Specification Differences
The most obvious differences are in core configuration and memory. The RTX 4070 GDDR6 has 5888 shading units, 184 TMUs, and 64 ROPs, versus the RX 560’s 1024 shading units, 64 TMUs, and 16 ROPs. This is a 5.75x increase in shading units and a 4x increase in ROPs. The RTX 4070 GDDR6 also has 46 RT cores and 184 tensor cores; the RX 560 has neither.
Memory is another major differentiator. The RTX 4070 GDDR6 comes with 12 GB of GDDR6 on a 192-bit bus, delivering 480.0 GB/s of bandwidth. The RX 560 has 4 GB of GDDR5 on a 128-bit bus, with 112.0 GB/s. The RTX 4070 GDDR6’s memory clock is 2500 MHz (20 Gbps effective), while the RX 560’s is 1750 MHz (7 Gbps effective).
Power and physical characteristics differ drastically. The RX 560 has a 75 W TDP, requires no power connectors, and suggests a 250 W PSU. It is 170 mm long, dual-slot, and uses a PCIe 3.0 x8 interface. The RTX 4070 GDDR6 has a 200 W TDP, needs a 1x 16-pin power connector, and recommends a 550 W PSU. It is 240 mm long, 110 mm tall, 40 mm wide, dual-slot, and uses PCIe 4.0 x16. The RTX 4070 GDDR6 also has more display outputs: 1x HDMI 2.1 and 3x DisplayPort 1.4a, versus the RX 560’s 1x DVI, 1x HDMI 2.0b, and 1x DisplayPort 1.4a.
Clock speeds are also far apart. The RTX 4070 GDDR6 boosts to 2475 MHz from a 1920 MHz base, while the RX 560 boosts to 1275 MHz from a 1175 MHz base. Pixel and texture rates follow: the RTX 4070 GDDR6 delivers 158.4 GPixel/s and 455.4 GTexel/s, versus the RX 560’s 20.40 GPixel/s and 81.60 GTexel/s. FP32 compute is 29.15 TFLOPS for the RTX 4070 GDDR6 and 2.611 TFLOPS for the RX 560, an 11x difference.
Head-to-Head Benchmarks
The direct head-to-head benchmark list is empty, so we must rely on each card’s individual scores. The RTX 4070 GDDR6’s sole result is in 3DMark Steel Nomad DX12, where it scores 4334.5. The RX 560 does not have a result in this test, so no direct comparison is possible there.
The RX 560’s best results come from Passmark G3D (3671) and Geekbench Metal (18941), with Geekbench OpenCL at 16472. Its Passmark compute score is 1437, and its DirectX-specific scores are low: 16 in DX10, 25 in DX11, 21 in DX12, and 57 in DX9. The G2D score is 485. These numbers show the RX 560 performs best in legacy DirectX 9 workloads and general 3D rasterization, but struggles with modern compute tasks.
The RTX 4070 GDDR6’s 4334.5 in Steel Nomad is a heavier, more modern workload than any of the RX 560’s Passmark tests. Steel Nomad is designed to stress DX12 features and high-resolution rendering. Given that the RX 560’s Passmark DX12 score is only 21, the RTX 4070 GDDR6 is likely several orders of magnitude faster in comparable modern DX12 scenarios. Even against the RX 560’s best Passmark G3D score of 3671, the RTX 4070 GDDR6’s 4334.5 is 18% higher, despite Steel Nomad being a more demanding test.
In terms of compute, the RTX 4070 GDDR6’s 29.15 TFLOPS FP32 dwarfs the RX 560’s 2.611 TFLOPS, an 11.2x advantage. The RTX 4070 GDDR6’s texture rate of 455.4 GTexel/s is 5.6x the RX 560’s 81.60 GTexel/s. Pixel rate is 7.8x higher. These raw throughput numbers align with the benchmark gap.
Where Each One Wins
The RTX 4070 GDDR6 wins in every modern, performance-intensive scenario. Its 12 GB VRAM and 480.0 GB/s bandwidth make it suitable for 1440p and 4K gaming, high-resolution texture packs, and creative workloads like video editing or 3D rendering. The 46 RT cores enable ray tracing in supported titles, and the 184 tensor cores accelerate DLSS and AI-based features. Its 200 W TDP is manageable for a high-end card, and the 550 W PSU recommendation is standard for this class.
The RX 560 wins exclusively in low-power and legacy scenarios. Its 75 W TDP and lack of power connectors mean it can run in systems with a 250 W PSU and no auxiliary power. This makes it an ideal drop-in upgrade for old office desktops or HTPCs. Its 4 GB VRAM is enough for 1080p esports titles like League of Legends or CS:GO, and its strong Passmark DX9 score (57) indicates it handles older games well. Its compact 170 mm length fits in small cases, and it uses PCIe 3.0 x8, which is backward-compatible with older motherboards.
In terms of raw benchmark wins, the RTX 4070 GDDR6’s single 3DMark score is higher than any of the RX 560’s scores, but the RX 560 has more benchmark coverage. The RTX 4070 GDDR6 wins on absolute performance, memory capacity, bandwidth, and feature set. The RX 560 wins on power efficiency, physical size, and compatibility with legacy systems. For a modern gaming rig, the RTX 4070 GDDR6 is the obvious choice. For a budget, low-power secondary system, the RX 560 remains functional and relevant.