AMD Radeon Pro 5700 vs NVIDIA GeForce GTX 690 Comparison

The AMD Radeon Pro 5700 and NVIDIA GeForce GTX 690 represent two distinct eras of GPU design, separated by nearly a decade of architectural evolution. The data shows a clear performance hierarchy, but the comparison reveals more than just raw speed; it highlights fundamental shifts in process technology, memory design, and compute capabilities. The Radeon Pro 5700, built on 7 nm RDNA 1.0, dominates the shared benchmark suite, while the GTX 690, a dual-GPU Kepler flagship, remains a historical artifact with a higher launch MSRP.

Where Each One Wins

The benchmark results provide a one-sided picture in this comparison. The AMD Radeon Pro 5700 wins both head-to-head tests, securing 2 wins against 0 for the NVIDIA GeForce GTX 690. The most decisive victories occur in compute-oriented workloads. For instance, the Radeon Pro 5700 achieves a Geekbench OpenCL score of 52787, which is 203.4% higher than the GTX 690’s 17399. Similarly, in Vulkan, the Radeon Pro 5700 scores 51289, outperforming the GTX 690’s 16675 by 207.6%.

The GTX 690’s wins are absent from the data, meaning that for any workload where a comparison exists, the AMD part comes out ahead. However, the scope of testing is limited to two Geekbench compute tests. The GTX 690 has no entries for DirectX, PassMark, or Metal benchmarks, so its performance in gaming or legacy APIs is not quantified here. The Radeon Pro 5700, conversely, shows a broader benchmark profile, including a PassMark G3D score of 11583 and a PassMark GPU Compute score of 4961, indicating its strength in both rasterization and general-purpose compute. The GTX 690’s average benchmark score of 17037 places it near its nearest rivals, such as the AMD Radeon HD 7970M (17019, a 0.1% delta), while the Radeon Pro 5700’s average of 18189 sits slightly below the Intel Arc A770M (18383, a -1.1% delta).

Architecture Differences

The architectural gap between these two GPUs is vast. The Radeon Pro 5700 uses the Navi 10 chip with RDNA 1.0 architecture, fabricated on a 7 nm process at TSMC. It packs 10,300 million transistors into a 251 mm² die, yielding a transistor density of 41.0M per mm². In contrast, the GTX 690 uses the GK104 chip with Kepler architecture on a 28 nm process, also from TSMC. It contains 3,540 million transistors on a larger 294 mm² die, resulting in a much lower density of 12.0M per mm².

Memory configurations differ sharply. The Radeon Pro 5700 features 8 GB of GDDR6 memory on a 256-bit bus, delivering 384.0 GB/s of bandwidth. The GTX 690 has only 2 GB of GDDR5 memory on the same 256-bit bus, but with 192.3 GB/s bandwidth—exactly half the throughput. Clock speeds also favor the newer part: the Radeon Pro 5700 runs at a base of 1243 MHz and boost of 1350 MHz, whereas the GTX 690 operates at 915 MHz base and 1019 MHz boost. Memory clocks show a similar split, with 12 Gbps effective for the Radeon Pro 5700 versus 6 Gbps for the GTX 690.

Compute resources are fundamentally different in scale. The Radeon Pro 5700 has 2304 shading units, 144 TMUs, and 64 ROPs, producing a pixel rate of 86.40 GPixel/s and a texture rate of 194.4 GTexel/s. The GTX 690, being a dual-GPU board, has 1536 shading units, 128 TMUs, and 32 ROPs per GPU, but the listed numbers aggregate to a pixel rate of 32.61 GPixel/s and a texture rate of 130.4 GTexel/s. FP32 performance is 6.221 TFLOPS for the Radeon Pro 5700, more than double the GTX 690’s 3.130 TFLOPS. The Radeon Pro 5700 also supports FP16 at 12.44 TFLOPS (2:1), while the GTX 690 has no listed FP16 capability.

Head-to-Head Benchmarks

The head-to-head data is unambiguous. In Geekbench OpenCL, the Radeon Pro 5700 scores 52787 versus the GTX 690’s 17399, a delta of 203.4%. This massive margin reflects the combination of higher clocks, doubled FP32 throughput, and superior memory bandwidth. The Vulkan test shows a similar story: 51289 for the Radeon Pro 5700 against 16675 for the GTX 690, a 207.6% advantage.

To contextualize these scores, the Radeon Pro 5700’s average benchmark score of 18189 places it among modern mid-range parts. Its nearest rival, the NVIDIA GeForce RTX 3060 Mobile, scores 18159, a negligible 0.2% difference. The GTX 690’s average of 17037 is closer to the AMD Radeon HD 7970M (17019, 0.1% delta) and the NVIDIA Tesla M4 (16932, 0.6% delta). The performance gap between the two cards in this comparison is not incremental; it is a generational leap. The Radeon Pro 5700’s 62nd percentile versus the GTX 690’s 60th percentile further illustrates that even the older card holds a respectable position in the broader distribution, yet the head-to-head tests show a threefold performance advantage for the newer architecture.

FAQ

Q: Which GPU has higher raw compute performance?

A: The AMD Radeon Pro 5700, with FP32 throughput of 6.221 TFLOPS, is roughly twice the NVIDIA GeForce GTX 690’s 3.130 TFLOPS.

Q: How much faster is the Radeon Pro 5700 in OpenCL?

A: The Radeon Pro 5700 scores 52787 in Geekbench OpenCL, which is 203.4% higher than the GTX 690’s 17399.

Q: What is the memory bandwidth difference?

A: The Radeon Pro 5700 provides 384.0 GB/s, exactly double the GTX 690’s 192.3 GB/s, using GDDR6 versus GDDR5 memory.

Q: Does the GTX 690 support Vulkan 1.3 or higher?

A: No. The GTX 690 supports Vulkan 1.2.175, while the Radeon Pro 5700 supports Vulkan 1.4.

Q: How do their transistor densities compare?

A: The Radeon Pro 5700 has a density of 41.0M transistors per mm², over three times the GTX 690’s 12.0M per mm².

Q: Which card has a higher average benchmark score?

A: The Radeon Pro 5700 averages 18189, compared to the GTX 690’s 17037, a difference of roughly 6.8%.

The Verdict

Based strictly on the data, the AMD Radeon Pro 5700 is the superior product for any compute task measured. It wins every head-to-head benchmark by a wide margin, offers more than double the FP32 performance, and provides four times the memory capacity (8 GB versus 2 GB) with double the bandwidth. The GTX 690, despite its dual-GPU design, cannot overcome the architectural efficiency of the 7 nm RDNA 1.0 part. The Radeon Pro 5700’s percentile ranking (62nd) is also higher than the GTX 690’s (60th), reinforcing its position in the overall performance distribution.

For a user prioritizing modern API support, the Radeon Pro 5700 is the clear choice, with Vulkan 1.4 and DirectX 12 (12_1) versus the GTX 690’s Vulkan 1.2.175 and DirectX 12 (11_0). The GTX 690 does have a higher launch MSRP of 999 USD, but that figure is irrelevant to performance analysis. The Radeon Pro 5700 also requires less power (130 W TDP versus 300 W) and a smaller PSU (300 W versus 700 W), though those are secondary considerations. The verdict is simple: the Radeon Pro 5700 is the better GPU, and the GTX 690 should only be considered for legacy compatibility, not performance.

Specification Differences

| Specification | AMD Radeon Pro 5700 | NVIDIA GeForce GTX 690 |

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

| Architecture | RDNA 1.0 | Kepler |

| Process Node | 7 nm | 28 nm |

| Transistors | 10,300 million | 3,540 million |

| Die Size | 251 mm² | 294 mm² |

| Transistor Density | 41.0M / mm² | 12.0M / mm² |

| Base Clock | 1243 MHz | 915 MHz |

| Boost Clock | 1350 MHz | 1019 MHz |

| Memory Size | 8 GB GDDR6 | 2 GB GDDR5 |

| Memory Bus Width | 256 bit | 256 bit |

| Memory Bandwidth | 384.0 GB/s | 192.3 GB/s |

| Shading Units | 2304 | 1536 |

| TMUs | 144 | 128 |

| ROPs | 64 | 32 |

| Pixel Rate | 86.40 GPixel/s | 32.61 GPixel/s |

| Texture Rate | 194.4 GTexel/s | 130.4 GTexel/s |

| FP32 | 6.221 TFLOPS | 3.130 TFLOPS |

| FP16 | 12.44 TFLOPS (2:1) | None |

| TDP | 130 W | 300 W |

| Slot Width | IGP | Dual-slot |

| Power Connectors | None | 2x 8-pin |

| Suggested PSU | 300 W | 700 W |

| Bus Interface | PCIe 4.0 x16 | PCIe 3.0 x16 |

| Display Outputs | No outputs | 3x DVI, 1x mini-DisplayPort 1.2 |

| DirectX Version | 12 (12_1) | 12 (11_0) |

| Vulkan Version | 1.4 | 1.2.175 |

| Release Date | 2020-08-03 | 2012-05-02 |

| Launch MSRP | None | 999 USD |