AMD Radeon Pro 5700 XT vs NVIDIA GeForce RTX 2070 SUPER Comparison

The benchmark data presents a clear hierarchy between the NVIDIA GeForce RTX 2070 SUPER and the AMD Radeon Pro 5700 XT. Across nine head-to-head tests, the NVIDIA card secures nine wins, with the AMD card failing to take a single benchmark. The average benchmark score for the RTX 2070 SUPER is 20,282, placing it at the 65th percentile of all GPUs, while the Radeon Pro 5700 XT averages 18,685, sitting at the 63rd percentile. This page breaks down the specific performance deltas, architectural origins, and use-case implications derived strictly from the provided data.

Head-to-Head Benchmarks

The most decisive victory for the NVIDIA GeForce RTX 2070 SUPER comes in the Geekbench Vulkan test, where it scores 90,637 against the AMD’s 56,593. This represents a 60.2% delta, the largest margin in the entire comparison. The NVIDIA card’s advantage is substantial in this cross-platform graphics API, indicating a significant performance gap in modern rendering workloads.

DirectX 11 performance shows a similar pattern of dominance. The RTX 2070 SUPER achieves a Passmark score of 151, while the Radeon Pro 5700 XT trails at 85, a 77.6% difference. This is closely followed by the DirectX 10 test, where NVIDIA leads 132 to 76, a 73.7% delta. These legacy API results suggest that the NVIDIA architecture maintains a strong edge in older, more CPU-bound rendering paths.

The Passmark G3D score, often a proxy for overall gaming performance, also heavily favors NVIDIA. The RTX 2070 SUPER posts 18,169, while the Radeon Pro 5700 XT scores 12,547, a 44.8% advantage. In compute-oriented workloads, the Geekbench OpenCL test shows NVIDIA ahead with 83,358 versus 59,467, a 40.2% delta, while the Passmark GPU Compute test shows a narrower but still decisive 30.6% lead (7,557 vs 5,787).

The closest contest is in Passmark DirectX 12, where the RTX 2070 SUPER scores 67 and the AMD card scores 59, a 13.6% delta. While this remains a win for NVIDIA, it is the smallest gap in the dataset, suggesting that the AMD RDNA architecture is comparatively more competitive in the latest DirectX 12 feature set. The 2D performance delta is also modest, with NVIDIA leading 878 to 810, an 8.4% difference.

Architecture Differences

The two cards are built on fundamentally different architectures and process nodes. The NVIDIA GeForce RTX 2070 SUPER utilizes the TU104 chip based on the Turing architecture, fabricated on a 12 nm process at TSMC. This chip contains 13,600 million transistors on a die size of 545 mm², resulting in a transistor density of 25.0M / mm². In contrast, the AMD Radeon Pro 5700 XT uses the Navi 10 chip based on the RDNA 1.0 architecture, built on a more advanced 7 nm process, also at TSMC. The AMD chip packs 10,300 million transistors into a much smaller 251 mm² die, achieving a higher transistor density of 41.0M / mm².

These architectural differences extend to compute resources. Both cards feature 2,560 shading units and 160 texture mapping units (TMUs), along with 64 render output units (ROPs). However, the NVIDIA card includes 40 dedicated RT cores and 320 tensor cores, features entirely absent from the AMD specification, which lists null for both. This indicates a hardware-level focus on ray tracing and AI-accelerated workloads for the Turing card, while the RDNA 1.0 architecture lacks these dedicated blocks.

Clock speeds and memory configurations also diverge significantly. The NVIDIA card operates at a base clock of 1605 MHz and a boost clock of 1770 MHz, while the AMD card runs lower at 1243 MHz base and 1499 MHz boost. In terms of memory, the RTX 2070 SUPER has 8 GB of GDDR6 on a 256-bit bus, delivering 448.0 GB/s of bandwidth with a memory clock of 1750 MHz (14 Gbps effective). The Radeon Pro 5700 XT doubles the capacity to 16 GB of GDDR6 but on the same 256-bit bus, resulting in lower bandwidth of 384.0 GB/s due to its slower 1500 MHz (12 Gbps effective) memory clock.

The power and interface profiles are starkly different. NVIDIA lists a 215 W TDP with a dual-slot cooler and requires both a 6-pin and 8-pin power connector, suggesting a 550 W power supply. The AMD card is far more power-efficient with a 130 W TDP, and it is designated as an IGP (integrated graphics processor) with no power connectors and no display outputs, requiring only a 300 W power supply. The bus interface also differs, with NVIDIA using PCIe 3.0 x16 and AMD utilizing the newer PCIe 4.0 x16.

Where Each One Wins

Based on the benchmark results, the NVIDIA GeForce RTX 2070 SUPER is the clear winner across every test category included in the dataset. The data shows no scenario where the AMD Radeon Pro 5700 XT outperforms NVIDIA. The most significant wins for NVIDIA are in synthetic graphics tests like Passmark G3D and legacy DirectX APIs (9, 10, 11), where the deltas exceed 44%. This points to a strong advantage in traditional rasterized gaming workloads and compatibility with older software stacks.

The NVIDIA card also wins decisively in compute benchmarks, including Geekbench OpenCL and Vulkan, as well as Passmark GPU Compute. The 40.2% and 60.2% deltas in Geekbench tests suggest that for general-purpose GPU compute and modern API utilization, the RTX 2070 SUPER is substantially more capable. The presence of tensor cores and RT cores, while not directly benchmarked here, further separates the cards for AI and ray tracing tasks, areas where the AMD card has no hardware support.

The only area where the AMD card shows relative strength is in DirectX 12, where the delta shrinks to 13.6%. While still a loss, this indicates that the RDNA architecture is less disadvantaged in the latest Microsoft API. However, this single data point does not translate into a win. For any user choosing between these two based strictly on the provided benchmark scores, the NVIDIA card is the only option that delivers superior performance across the board.

FAQ

Q: Which card has the higher average benchmark score?

A: The NVIDIA GeForce RTX 2070 SUPER has an average benchmark score of 20,282, which is higher than the AMD Radeon Pro 5700 XT's average of 18,685.

Q: What is the largest performance gap between the two cards?

A: The largest gap is in the Geekbench Vulkan test, where the NVIDIA card scores 90,637 against AMD's 56,593, resulting in a 60.2% delta in favor of NVIDIA.

Q: Does the AMD card have any hardware support for ray tracing?

A: No. The AMD Radeon Pro 5700 XT lists null for both RT cores and tensor cores, while the NVIDIA card features 40 RT cores and 320 tensor cores.

Q: How do the memory capacities differ?

A: The AMD Radeon Pro 5700 XT has 16 GB of GDDR6 memory, while the NVIDIA GeForce RTX 2070 SUPER has 8 GB of GDDR6 memory. However, the NVIDIA card has higher memory bandwidth at 448.0 GB/s compared to AMD's 384.0 GB/s.

Q: What is the difference in power consumption?

A: The NVIDIA card has a 215 W TDP, while the AMD card has a 130 W TDP. The AMD card also requires no power connectors and has a 300 W suggested PSU, whereas NVIDIA needs a 6-pin and 8-pin connector and a 550 W PSU.

Q: Which card has a higher transistor density?

A: The AMD Radeon Pro 5700 XT has a higher transistor density of 41.0M / mm² due to its 7 nm process, compared to the NVIDIA card's 25.0M / mm² on a 12 nm process.

Specification Differences

The specifications reveal distinct design philosophies. The NVIDIA GeForce RTX 2070 SUPER uses a 12 nm process with a 545 mm² die, while the AMD Radeon Pro 5700 XT uses a 7 nm process with a 251 mm² die. Clock speeds differ, with NVIDIA boosting to 1770 MHz and AMD to 1499 MHz. Memory configurations show AMD providing double the capacity (16 GB vs 8 GB) but NVIDIA delivering higher bandwidth (448.0 GB/s vs 384.0 GB/s).

Compute rates favor NVIDIA, with a pixel rate of 113.3 GPixel/s and a texture rate of 283.2 GTexel/s, compared to AMD's 95.94 GPixel/s and 239.8 GTexel/s. FP32 performance is 9.062 TFLOPS for NVIDIA versus 7.675 TFLOPS for AMD. The cards also differ in power requirements, with NVIDIA at 215 W TDP requiring external power connectors, while AMD is at 130 W TDP with none. The form factor differs as well: NVIDIA is dual-slot with dimensions of 267mm x 116mm x 35mm, whereas AMD is listed as an IGP with no dimensions or display outputs. The NVIDIA card supports DirectX 12 Ultimate (12_2), while the AMD card supports DirectX 12 (12_1).

The Verdict

The data is unambiguous. The NVIDIA GeForce RTX 2070 SUPER outperforms the AMD Radeon Pro 5700 XT in every single benchmark recorded. With nine wins and zero losses, the NVIDIA card offers a 40.2% lead in OpenCL, a 60.2% lead in Vulkan, and a 44.8% lead in the Passmark G3D test. For users prioritizing raw performance in gaming, compute, and legacy API compatibility, the RTX 2070 SUPER is the superior choice.

The AMD Radeon Pro 5700 XT does have advantages outside of raw speed. Its 16 GB of memory is double the NVIDIA card's 8 GB, which could be relevant for capacity-sensitive workloads. It also consumes significantly less power (130 W vs 215 W TDP) and uses a more advanced 7 nm process, making it a more efficient compute unit. However, these advantages do not translate into a single benchmark win, and the card's lack of display outputs and dedicated RT/tensor cores limits its versatility for end-user graphics tasks.

The choice depends on the use case. If the primary goal is maximum frame rates and compute throughput, the data points exclusively to the NVIDIA GeForce RTX 2070 SUPER. If the requirement is a low-power, high-capacity memory solution for a specialized compute environment where performance per watt matters more than absolute speed, the AMD card holds specification-level appeal, but its benchmark scores lag significantly. The verdict from the numbers is that NVIDIA wins the performance crown in this comparison, while AMD offers a more power-efficient, higher-capacity alternative that does not compete on speed.