AMD Radeon Pro 570 vs AMD Radeon RX 590 GME Comparison

The Verdict

The data is unambiguous: the AMD Radeon RX 590 GME is the faster card in every benchmark where the two are compared directly, and it does so while offering double the memory. The Pro 570’s only statistical advantage is a slight win in average benchmark score—33,207 versus 32,601—which is a 1.9% margin that falls within the noise of the nearest-rival deltas (the Pro 570 sits 0.1% ahead of the RTX 3050 Mobile, while the RX 590 GME sits 0.2% ahead of the FirePro S9300 X2). If you need raw compute performance and don’t care about the physical form factor, the RX 590 GME is the clear pick. If you specifically require an integrated (IGP) card that draws no external power connectors and fits in a portable chassis, the Pro 570 is the only option that matches that description. For anyone building a desktop workstation with space for a dual-slot card and an 8-pin connector, the RX 590 GME’s 47.2% lead in Vulkan is decisive.

FAQ

Q: Which card is faster in Geekbench OpenCL?

A: The AMD Radeon RX 590 GME scores 36,294 versus the Pro 570’s 27,702, a 23.7% advantage.

Q: How big is the Vulkan performance gap?

A: The RX 590 GME scores 60,508 in Geekbench Vulkan, while the Pro 570 scores 31,974. That is a 47.2% difference in favor of the RX 590 GME.

Q: Do both cards use the same architecture?

A: Yes, both are GCN 4.0 on a 14 nm process from GlobalFoundries, with the same 5,700 million transistors on a 232 mm² die.

Q: What is the memory difference?

A: The RX 590 GME has 8 GB of GDDR5 with 256.0 GB/s bandwidth, while the Pro 570 has 4 GB of GDDR5 with 217.0 GB/s bandwidth. Both use a 256-bit bus.

Q: Are there any benchmark results where the Pro 570 wins?

A: In the head-to-head data, the Pro 570 wins zero tests. Its overall average benchmark score (33,207) is higher than the RX 590 GME’s (32,601), but that is not from a direct comparison test.

Q: Can the Pro 570 be installed in a standard desktop case?

A: The Pro 570 is listed as an IGP (integrated graphics processor) with no power connectors and "portable device dependent" display outputs. The RX 590 GME is a dual-slot card with a 241 mm length and a single 8-pin connector.

Architecture Differences

Both GPUs share the same GCN 4.0 architecture, built on the same 14 nm GlobalFoundries process, with identical transistor counts of 5,700 million and identical die sizes of 232 mm². The transistor density is the same at 24.6M per mm². The fundamental silicon is the same, but the configurations diverge significantly.

The Pro 570 uses the Ellesmere chip, while the RX 590 GME uses Polaris 20. The Pro 570 belongs to the "Radeon Pro Mac (500 Series)" generation, whereas the RX 590 GME is part of the "Polaris (RX 500)" generation. The RX 590 GME has a predecessor listed as "Arctic Islands" and a successor as "Vega," while the Pro 570 has no listed predecessor or successor.

The shading unit count is the most striking architectural difference: the RX 590 GME has 2,304 shading units, versus 1,792 on the Pro 570. That is a 28.6% increase in raw shader hardware. The texture mapping units (TMUs) also scale up from 112 to 144, while the render output units (ROPs) remain identical at 32. Neither card features dedicated ray tracing or tensor cores. Both support DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3, so API feature parity is complete.

The memory subsystem is where the architectural choices show up in practice. Both use GDDR5 on a 256-bit bus, but the RX 590 GME runs its memory at 2000 MHz (8 Gbps effective) versus the Pro 570’s 1695 MHz (6.8 Gbps effective). That yields 256.0 GB/s versus 217.0 GB/s of bandwidth, an 18% advantage for the RX 590 GME. The RX 590 GME also doubles the capacity to 8 GB.

Specification Differences

The clocks are substantially different. The Pro 570 has a base clock of 1000 MHz and a boost of 1105 MHz. The RX 590 GME starts at 1257 MHz base and boosts to 1420 MHz. That is a 25.7% higher base clock and a 28.5% higher boost clock.

The compute rates scale accordingly. The RX 590 GME delivers 6.543 TFLOPS of FP32 and FP16 (1:1), while the Pro 570 delivers 3.960 TFLOPS for both. That is 65.2% more compute throughput. Pixel rate goes from 35.36 GPixel/s on the Pro 570 to 45.44 GPixel/s on the RX 590 GME, a 28.5% lift. Texture rate jumps from 123.8 GTexel/s to 204.5 GTexel/s, a 65.2% improvement.

Power and physical design differ sharply. The Pro 570 is rated at 150 W TDP, is an IGP slot width, has no power connectors, and has "portable device dependent" display outputs. The RX 590 GME is rated at 175 W TDP, is dual-slot, requires a single 8-pin power connector, recommends a 450 W PSU, and has 1x HDMI 2.0b plus 3x DisplayPort 1.4a outputs. The RX 590 GME is 241 mm (9.5 inches) long.

Release dates differ: the Pro 570 launched on June 4, 2017, and the RX 590 GME launched on March 8, 2020. Both are end-of-life. Neither has a launch MSRP listed.

Head-to-Head Benchmarks

The direct comparison data covers two tests, and the RX 590 GME wins both. In Geekbench OpenCL, the RX 590 GME scores 36,294 against the Pro 570’s 27,702. The deltaPct is -23.7%, meaning the Pro 570 is 23.7% behind. That is a significant gap but not a catastrophic one—it suggests the RX 590 GME’s extra shading units and higher clocks translate into a solid but not overwhelming compute lead in OpenCL workloads.

The Vulkan result is far more lopsided. The RX 590 GME scores 60,508, while the Pro 570 manages only 31,974. The deltaPct here is -47.2%, meaning the Pro 570 delivers just over half the Vulkan performance. This is a massive difference, indicating that the RX 590 GME’s architecture scales much better under Vulkan’s lower-overhead API. The Vulkan gap is roughly double the OpenCL gap in percentage terms, which points to the RX 590 GME’s higher memory bandwidth and shading unit count being more fully utilized in modern graphics workloads.

The overall win tally is 2-0 in favor of the RX 590 GME. The Pro 570’s average benchmark score of 33,207 versus the RX 590 GME’s 32,601 is the only counterpoint, but that average includes different test suites—the Pro 570 has a Geekbench Metal score of 39,945, which the RX 590 GME cannot compete in because it lacks that specific benchmark result. The RX 590 GME has a 3DMark Steel Nomad DX12 score of 1,001, which the Pro 570 lacks.

Where Each One Wins

The RX 590 GME wins in every category that matters for gaming and general compute. It is 23.7% faster in OpenCL and 47.2% faster in Vulkan. It has 8 GB of memory versus 4 GB, which matters for modern textures and large datasets. It has higher bandwidth (256.0 GB/s versus 217.0 GB/s), higher pixel rate (45.44 GPixel/s versus 35.36 GPixel/s), and higher texture rate (204.5 GTexel/s versus 123.8 GTexel/s). It also has 65.2% more FP32 throughput. For any workload that uses Vulkan—which includes most modern Windows games and many compute APIs—the RX 590 GME is the definitive choice.

The Pro 570 has exactly one niche where it wins: physical integration. It is an IGP with no power connectors, meaning it can be embedded in a portable or all-in-one system where a dual-slot card with an 8-pin connector cannot fit. Its display outputs are "portable device dependent," confirming it is designed for laptop or compact chassis use. Its lower TDP (150 W versus 175 W) also makes it easier to cool in constrained environments. If you are building a small-form-factor or mobile workstation and cannot accommodate a 241 mm dual-slot card, the Pro 570 is the only option between these two.

In terms of percentile ranking, both cards are nearly identical—the Pro 570 sits at the 78th percentile of all GPUs, while the RX 590 GME sits at the 77th. The nearest rivals confirm they are in the same performance class: the Pro 570 trades blows with the RTX 3050 Mobile (0.1% delta) and the T550 Mobile (0.1% delta), while the RX 590 GME is 0.2% ahead of the FirePro S9300 X2 and 0.4% ahead of the RX 7900 GRE. The practical takeaway is that for desktop users with standard power and cooling, the RX 590 GME is strictly superior. For embedded and portable systems, the Pro 570 is the only viable pick, and you accept the performance penalty as the cost of the form factor.