AMD Radeon RX 5700 vs AMD Radeon RX 590 Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon RX 590 has an average benchmark score of 24,744, while the AMD Radeon RX 5700 scores 22,170. The RX 590 sits at the 70th percentile of all GPUs, compared to the RX 5700's 67th percentile.

Q: How do the two cards compare in the 3DMark Steel Nomad DX12 test?

A: The RX 5700 scores 1,862, which is 41.9% higher than the RX 590's 1,081. This is a substantial lead for the newer architecture in this DirectX 12 workload.

Q: What is the difference in Geekbench Metal performance?

A: The RX 5700 records 86,079 points versus the RX 590's 48,407 points. The RX 5700 is 43.8% ahead in this Apple Metal API benchmark.

Q: Which GPU has more shading units and texture mapping units?

A: Both cards have identical counts: 2,304 shading units and 144 TMUs. The difference lies elsewhere, such as in ROPs and memory technology.

Q: What are the memory bandwidth figures for each card?

A: The RX 590 uses GDDR5 memory with a 256-bit bus and achieves 256.0 GB/s. The RX 5700 uses GDDR6 with the same 256-bit bus but reaches 448.0 GB/s, a 75% bandwidth advantage.

Q: Do both cards support the same DirectX version?

A: No. The RX 590 supports DirectX 12 (12_0), while the RX 5700 supports DirectX 12 (12_1). The RX 5700 also supports Vulkan 1.4 versus Vulkan 1.3 on the RX 590.

Architecture Differences

The two GPUs represent distinct architectural generations from AMD. The RX 590 is built on the Polaris 30 chip using GCN 4.0 architecture, fabricated on a 12 nm process at GlobalFoundries. The RX 5700 uses the Navi 10 chip with RDNA 1.0 architecture, manufactured on a 7 nm process at TSMC. This process shrink is significant: the RX 5700 packs 10,300 million transistors into a 251 mm² die, while the RX 590 fits 5,700 million transistors into a 232 mm² die. The transistor density tells the story: 41.0M per mm² for the RX 5700 versus 24.6M per mm² for the RX 590.

Clock behavior also differs markedly. The RX 590 has a base clock of 1469 MHz and a boost clock of 1545 MHz, with no separate game clock listed. The RX 5700 starts at a slightly lower 1465 MHz base but boosts to 1725 MHz, with a dedicated game clock of 1625 MHz. This higher sustained boost is a direct consequence of the more efficient RDNA design.

The memory subsystem is another major architectural divergence. The RX 590 pairs its 8 GB GDDR5 memory with a 2000 MHz base clock (8 Gbps effective) for 256.0 GB/s bandwidth. The RX 5700 uses 8 GB of GDDR6 at 1750 MHz (14 Gbps effective) on the same 256-bit bus, yielding 448.0 GB/s. The RX 5700 also doubles the ROP count from 32 to 64, which directly affects pixel throughput.

Compute capabilities differ beyond raw counts. The RX 590 delivers 7.119 TFLOPS FP32 and the same 7.119 TFLOPS FP16 (1:1 ratio). The RX 5700 reaches 7.949 TFLOPS FP32 but offers 15.90 TFLOPS FP16 (2:1 ratio), indicating a different approach to half-precision math. Pixel rate jumps from 49.44 GPixel/s on the RX 590 to 110.4 GPixel/s on the RX 5700, while texture rate improves from 222.5 GTexel/s to 248.4 GTexel/s.

The RX 5700 also supports PCIe 4.0 x16, while the RX 590 is limited to PCIe 3.0 x16. API support shows the RX 5700 with DirectX 12 (12_1) and Vulkan 1.4, versus DirectX 12 (12_0) and Vulkan 1.3 on the RX 590. Both share OpenGL 4.6 and identical display outputs (1x HDMI 2.0b, 3x DisplayPort 1.4a).

Head-to-Head Benchmarks

The database records two direct comparison benchmarks between these GPUs, and the RX 5700 wins both decisively. In 3DMark Steel Nomad DX12, the RX 590 scores 1,081 while the RX 5700 scores 1,862. The delta of -41.9% means the RX 590 trails by nearly 42% in this modern DirectX 12 workload. This is not a marginal gap; it is a generational leap in rasterization efficiency.

The Geekbench Metal test shows an even wider margin. The RX 590 records 48,407 points, while the RX 5700 reaches 86,079 points. The -43.8% delta indicates the RX 5700 is roughly 78% faster in absolute terms. Metal is a low-level API that exposes hardware capabilities directly, and the RDNA architecture clearly extracts more from its compute units.

Notably, the RX 590 has no wins across any benchmark in this comparison. The winsA count is 0, and winsB is 2. The only other benchmark data available for the RX 590 are its average score and percentile, which are higher than the RX 5700's despite the RX 5700 winning every head-to-head test. This paradox deserves attention.

The average benchmark score of 24,744 for the RX 590 versus 22,170 for the RX 5700 suggests the RX 590 performs better across a broader suite of older or less demanding tests. The RX 5700's additional benchmark results (Geekbench OpenCL at 69,603, Vulkan at 64,166, and various Passmark tests) show strong scores in compute-heavy workloads but a lower overall average, possibly due to inclusion of more diverse tests.

In the Passmark suite, the RX 5700 scores 14,314 in G3D, 6,517 in GPU compute, and 877 in G2D. Its DirectX 9 score of 209 is notably higher than DirectX 11 (101) and DirectX 12 (54), which hints at driver maturity or architectural optimization for legacy APIs.

Specification Differences

The most obvious specification differences between these two cards are as follows:

• Process Node: RX 590 uses 12 nm GlobalFoundries; RX 5700 uses 7 nm TSMC
• Transistors: 5,700 million versus 10,300 million
• Die Size: 232 mm² versus 251 mm²
• Transistor Density: 24.6M / mm² versus 41.0M / mm²
• Base Clock: 1469 MHz versus 1465 MHz
• Boost Clock: 1545 MHz versus 1725 MHz
• Game Clock: Not listed versus 1625 MHz
• Memory Clock: 2000 MHz (8 Gbps effective) versus 1750 MHz (14 Gbps effective)
• Memory Type: GDDR5 versus GDDR6
• Memory Bandwidth: 256.0 GB/s versus 448.0 GB/s
• ROPs: 32 versus 64
• Pixel Rate: 49.44 GPixel/s versus 110.4 GPixel/s
• Texture Rate: 222.5 GTexel/s versus 248.4 GTexel/s
• FP32: 7.119 TFLOPS versus 7.949 TFLOPS
• FP16: 7.119 TFLOPS (1:1) versus 15.90 TFLOPS (2:1)
• TDP: 175 W versus 180 W
• Power Connectors: 1x 8-pin versus 1x 6-pin + 1x 8-pin
• Bus Interface: PCIe 3.0 x16 versus PCIe 4.0 x16
• DirectX: 12 (12_0) versus 12 (12_1)
• Vulkan: 1.3 versus 1.4
• Dimensions: 241 mm length versus 268 mm length, 111 mm height, 36 mm width
• Release Date: 2018-11-14 versus 2019-07-06
• Launch MSRP: 279 USD versus 349 USD

Both share 8 GB memory, 256-bit bus width, 2,304 shading units, 144 TMUs, dual-slot width, 450 W suggested PSU, and identical display outputs. The RX 5700 is longer and physically larger, requiring more case clearance.

Where Each One Wins

The RX 5700 wins in every directly compared benchmark, making it the clear choice for modern workloads. Its advantages in 3DMark Steel Nomad DX12 and Geekbench Metal are substantial, with 41.9% and 43.8% leads respectively. The doubled ROP count and higher memory bandwidth directly benefit fill-rate-bound scenarios and high-resolution texturing. The 2:1 FP16 ratio suggests strong performance in compute tasks that utilize half precision, such as certain AI inference or image processing workloads.

The RX 5700's PCIe 4.0 interface provides double the bandwidth to the host system, which can matter for data-intensive compute tasks or when VRAM is insufficient and system memory must be used. Its higher boost clock of 1725 MHz and game clock of 1625 MHz indicate better sustained performance under load. The newer DirectX 12 (12_1) and Vulkan 1.4 support means better compatibility with upcoming titles that leverage these features.

The RX 590 wins in the broader average benchmark score (24,744 versus 22,170) and percentile ranking (70th versus 67th). This suggests it performs better across legacy DirectX 9, 10, and 11 workloads, or in tests that favor its higher base clock and simpler architecture. The RX 590 also has a shorter physical length (241 mm versus 268 mm), making it easier to fit in compact cases. Its lower TDP (175 W versus 180 W) and simpler single 8-pin power connector reduce power supply requirements, though both share a 450 W suggested PSU.

The RX 590's higher average score relative to its direct benchmark losses indicates that its performance profile is more consistent across a wide range of tests, whereas the RX 5700's average is pulled down by additional compute tests that may not reflect typical gaming scenarios.

The Verdict

The data clearly favors the AMD Radeon RX 5700 for users prioritizing modern gaming and compute performance. It wins both head-to-head benchmarks by margins exceeding 40%, offers more than 75% additional memory bandwidth, doubles the pixel rate, and provides superior API support. The 7 nm RDNA architecture delivers higher boost clocks and better efficiency per transistor. The RX 5700 is the pick for DirectX 12 titles, Vulkan applications, and any workload that leverages FP16 compute.

However, the RX 590's higher average benchmark score and percentile ranking cannot be dismissed. It outperforms the RX 5700 across the broader benchmark suite, likely due to strength in older APIs and less demanding workloads. For users whose software library consists primarily of DirectX 9, 10, or 11 titles, or who need a shorter card for a compact build, the RX 590 remains a viable option. Its lower launch MSRP of 279 USD versus 349 USD also reflects its position as a more accessible entry point, though both cards are now end-of-life.

The choice ultimately depends on workload priorities. The RX 5700 is the superior GPU for current and future games, particularly those using DirectX 12 or Vulkan. The RX 590 offers better average performance across the database's full test suite, which may indicate more consistent behavior in diverse applications. The RX 5700's 41.9% lead in Steel Nomad and 43.8% lead in Metal are decisive for anyone targeting modern rendering pipelines. The RX 590's 70th percentile versus the RX 5700's 67th percentile suggests a slight edge in overall compatibility, but the magnitude of the RX 5700's wins in the latest benchmarks makes it the stronger recommendation for forward-looking buyers.