NVIDIA GeForce GTX 690 vs NVIDIA GeForce RTX 3070 Comparison

NVIDIA GeForce RTX 3070 and NVIDIA GeForce GTX 690 occupy different corners of the GPU landscape, separated by an eight-month launch gap in the Fact Pack but by nearly a decade of architectural evolution. The RTX 3070, built on the Ampere architecture with an 8 nm process, delivers an average benchmark score of 17208, while the GTX 690, a Kepler dual-GPU card from the GeForce 600 generation, scores 17037. The overall percentile rankings are close—61st for the RTX 3070 versus 60th for the GTX 690—but the head-to-head data reveals a far more lopsided contest in specific workloads, driven by fundamental differences in compute capability, memory bandwidth, and API support.

Head-to-Head Benchmarks

The Fact Pack provides two direct head-to-head comparisons, and the RTX 3070 wins both. The most dramatic margin appears in the Geekbench OpenCL test, where the RTX 3070 scores 112,821 against the GTX 690’s 17,399. That is a delta of 548.4%, meaning the RTX 3070 delivers over six times the raw compute throughput in this general-purpose GPU workload. This is not a marginal improvement; it reflects a complete generational shift in shading unit count and FP32 throughput. The RTX 3070 carries 5,888 shading units and reaches 20.31 TFLOPS of FP32 performance, while the GTX 690 is built around 1,536 shading units and 3.130 TFLOPS. The OpenCL benchmark rewards that raw parallel compute, and the RTX 3070’s lead is absolute.

The second head-to-head test, Geekbench Vulkan, is closer but still favors the RTX 3070. The RTX 3070 posts 21,022 points versus the GTX 690’s 16,675, a delta of 26.1%. Vulkan is a lower-level API that can leverage the GTX 690’s dual-GPU design more effectively than OpenCL, yet the RTX 3070 still pulls ahead by more than a quarter. The GTX 690’s Vulkan support is listed as version 1.2.175, while the RTX 3070 runs Vulkan 1.4, and the newer architecture’s driver optimizations and hardware scheduling appear to overcome the older card’s multi-chip configuration. Across these two tests, the RTX 3070 holds a clean 2-0 win record, with no benchmark in the Fact Pack where the GTX 690 takes the lead.

Beyond the direct comparisons, the broader benchmark suite reinforces the RTX 3070’s dominance. In Passmark G3D, the RTX 3070 scores 22,214, and in Passmark GPU Compute it reaches 11,195. The GTX 690 has no corresponding scores in the Fact Pack, but its average benchmark score of 17,037 is dragged down by the lack of modern test coverage. The RTX 3070’s percentile rank of 61 places it above the GTX 690’s 60th percentile, and its nearest rivals include the AMD Radeon RX 7600 XT (0.7% higher) and the NVIDIA Tesla K40c (1.5% lower). The GTX 690’s nearest rival list includes the RTX 3070 itself, with a delta of -1%, confirming that the two cards are statistically close in aggregate but diverge sharply in modern workloads.

FAQ

Q: Which card has the higher average benchmark score?

A: The NVIDIA GeForce RTX 3070 has an average benchmark score of 17,208, compared to the NVIDIA GeForce GTX 690’s 17,037, a difference of 171 points or roughly 1%.

Q: How much faster is the RTX 3070 in Geekbench OpenCL?

A: The RTX 3070 scores 112,821 in Geekbench OpenCL, while the GTX 690 scores 17,399, making the RTX 3070 548.4% faster in that specific test.

Q: Does the GTX 690 win any benchmark in the head-to-head data?

A: No. The Fact Pack lists two head-to-head benchmarks—Geekbench OpenCL and Geekbench Vulkan—and the RTX 3070 wins both, with winsA equal to 2 and winsB equal to 0.

Q: What is the memory configuration difference?

A: The RTX 3070 has 8 GB of GDDR6 memory on a 256-bit bus with 448.0 GB/s bandwidth, while the GTX 690 has 2 GB of GDDR5 memory on a 256-bit bus with 192.3 GB/s bandwidth.

Q: Which card supports newer DirectX features?

A: The RTX 3070 supports DirectX 12 Ultimate (12_2), while the GTX 690 supports DirectX 12 (11_0), meaning the RTX 3070 is compatible with the latest feature levels.

Q: How do their transistor counts compare?

A: The RTX 3070 has 17,400 million transistors on a 392 mm² die, while the GTX 690 has 3,540 million transistors on a 294 mm² die, giving the RTX 3070 a much higher transistor density.

Architecture Differences

The two cards are built on entirely different architectures, nodes, and design philosophies. The RTX 3070 uses the GA104 chip on the Ampere architecture, fabricated by Samsung on an 8 nm process. It packs 17,400 million transistors into a 392 mm² die, yielding a transistor density of 44.4 million transistors per square millimeter. The GTX 690, in contrast, uses the GK104 chip on the Kepler architecture, made by TSMC on a 28 nm process, with 3,540 million transistors on a 294 mm² die—a density of just 12.0 million transistors per square millimeter. That density gap explains why the RTX 3070 can fit nearly five times as many transistors into a die that is only about a third larger physically.

Core configuration differs sharply. The RTX 3070 has 5,888 shading units, 184 texture mapping units (TMUs), and 96 raster operation pipelines (ROPs). It also includes 46 ray tracing cores and 184 tensor cores, neither of which exists on the GTX 690—those are null fields in its specification. The GTX 690 pairs 1,536 shading units with 128 TMUs and 32 ROPs. Even though the GTX 690 is a dual-GPU card—a fact implied by its dual 2 GB memory configuration and 300 W TDP—its per-GPU resources are far smaller. The RTX 3070’s pixel rate of 165.6 GPixel/s and texture rate of 317.4 GTexel/s dwarf the GTX 690’s 32.61 GPixel/s and 130.4 GTexel/s, respectively.

Memory architecture is another major divider. The RTX 3070 uses 8 GB of GDDR6 with 448.0 GB/s of bandwidth, while the GTX 690 uses 2 GB of GDDR5 with 192.3 GB/s. The bus width is identical at 256-bit, but the newer memory type and higher effective clock speed (14 Gbps versus 6 Gbps) give the RTX 3070 more than double the bandwidth. API support also diverges: the RTX 3070 lists DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, while the GTX 690 offers DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The RTX 3070 also uses a PCIe 4.0 x16 interface, while the GTX 690 is limited to PCIe 3.0 x16.

The Verdict

The data is unambiguous: the RTX 3070 is the superior card in every measured category. Its average benchmark score is higher (17,208 versus 17,037), it wins both head-to-head tests, and it holds a higher percentile rank (61st versus 60th). The Geekbench OpenCL result—a 548.4% lead—is the single most telling statistic, as it reflects the RTX 3070’s overwhelming advantage in raw FP32 compute: 20.31 TFLOPS versus 3.130 TFLOPS. The Vulkan win of 26.1% shows that even in a more balanced workload, the newer architecture’s efficiency and feature set carry the day.

For a buyer choosing between these two today, the RTX 3070 is the only rational pick from the data. Its 8 GB memory capacity is four times larger, its bandwidth is more than double, and it supports modern DirectX 12 Ultimate features that the GTX 690 cannot match. The GTX 690’s launch MSRP was 999 USD, while the RTX 3070 launched at 499 USD, but pricing is not the focus here—performance is. The GTX 690’s dual-GPU design may have been competitive in 2012, but the Fact Pack shows no benchmark where it outperforms the RTX 3070. The verdict is a clean sweep for the Ampere card.

Specification Differences

The following fields differ between the two cards, based exclusively on the Fact Pack:

• Process Node: RTX 3070 is 8 nm (Samsung); GTX 690 is 28 nm (TSMC).
• Transistors: RTX 3070 has 17,400 million; GTX 690 has 3,540 million.
• Die Size: RTX 3070 is 392 mm²; GTX 690 is 294 mm².
• Transistor Density: RTX 3070 is 44.4M / mm²; GTX 690 is 12.0M / mm².
• Base Clock: RTX 3070 runs at 1500 MHz; GTX 690 at 915 MHz.
• Boost Clock: RTX 3070 boosts to 1725 MHz; GTX 690 to 1019 MHz.
• Memory Clock: RTX 3070 uses 1750 MHz (14 Gbps effective); GTX 690 uses 1502 MHz (6 Gbps effective).
• Memory Size: RTX 3070 has 8 GB; GTX 690 has 2 GB.
• Memory Type: RTX 3070 uses GDDR6; GTX 690 uses GDDR5.
• Memory Bandwidth: RTX 3070 reaches 448.0 GB/s; GTX 690 reaches 192.3 GB/s.
• Shading Units: RTX 3070 has 5,888; GTX 690 has 1,536.
• TMUs: RTX 3070 has 184; GTX 690 has 128.
• ROPs: RTX 3070 has 96; GTX 690 has 32.
• RT Cores: RTX 3070 has 46; GTX 690 has none.
• Tensor Cores: RTX 3070 has 184; GTX 690 has none.
• Pixel Rate: RTX 3070 is 165.6 GPixel/s; GTX 690 is 32.61 GPixel/s.
• Texture Rate: RTX 3070 is 317.4 GTexel/s; GTX 690 is 130.4 GTexel/s.
• FP32: RTX 3070 is 20.31 TFLOPS; GTX 690 is 3.130 TFLOPS.
• FP16: RTX 3070 is 20.31 TFLOPS (1:1); GTX 690 has no listed FP16.
• TDP: RTX 3070 is 220 W; GTX 690 is 300 W.
• Power Connectors: RTX 3070 uses 1x 12-pin; GTX 690 uses 2x 8-pin.
• Suggested PSU: RTX 3070 recommends 550 W; GTX 690 recommends 700 W.
• Bus Interface: RTX 3070 is PCIe 4.0 x16; GTX 690 is PCIe 3.0 x16.
• Display Outputs: RTX 3070 has 1x HDMI 2.1 and 3x DisplayPort 1.4a; GTX 690 has 3x DVI and 1x mini-DisplayPort 1.2.
• DirectX Support: RTX 3070 is 12 Ultimate (12_2); GTX 690 is 12 (11_0).
• Vulkan Support: RTX 3070 is 1.4; GTX 690 is 1.2.175.
• Dimensions: RTX 3070 is 242 mm long and 112 mm high; GTX 690 is 279 mm long, 111 mm high, and 38 mm wide.
• Release Date: RTX 3070 launched 2020-08-31; GTX 690 launched 2012-05-02.
• Generation: RTX 3070 is GeForce 30; GTX 690 is GeForce 600.
• Predecessor/Successor: RTX 3070 follows GeForce 20 and precedes GeForce 40; GTX 690 follows GeForce 500 and precedes GeForce 700.

Where Each One Wins

The RTX 3070 wins in every scenario where the Fact Pack provides data. In Geekbench OpenCL, its 548.4% lead makes it the clear choice for compute-heavy tasks like rendering, physics simulation, or any workload that scales with FP32 throughput. In Geekbench Vulkan, its 26.1% advantage makes it the better option for modern games and applications that use low-level graphics APIs. The RTX 3070’s 8 GB memory capacity and 448.0 GB/s bandwidth also position it for high-resolution textures and larger datasets, while its DirectX 12 Ultimate support ensures compatibility with the latest graphical features.

The GTX 690 has no benchmark wins in the Fact Pack, so there is no measured scenario where it comes out ahead. Its only advantage lies in its physical configuration—a 300 W TDP and dual 8-pin connectors suggest it was designed for high-end multi-GPU setups, but its 2 GB memory per GPU and 192.3 GB/s bandwidth are severe bottlenecks in modern workloads. The GTX 690’s 60th percentile ranking is respectable, but it is still one point below the RTX 3070’s 61st percentile. For legacy DirectX 9 or DirectX 10 applications, the GTX 690’s Passmark scores of 247 and 150, respectively, are not directly comparable to the RTX 3070’s scores in those same tests (247 versus 150 for the GTX 690, though the RTX 3070 does not list those exact tests). The data simply does not support any use case where the GTX 690 outperforms the RTX 3070.