NVIDIA GeForce GTX 690 vs NVIDIA GeForce RTX 2070 Comparison

Where Each One Wins

The recorded benchmark data splits cleanly between these two NVIDIA cards, and the outcome is unambiguous. The RTX 2070 wins every head-to-head test in the database, with margins that are not close. In Geekbench OpenCL, the RTX 2070 scores 79,966 against the GTX 690's 17,399, a 359.6% advantage. In Geekbench Vulkan, the RTX 2070 posts 82,521 against 16,675, a 394.9% lead. That is a fourfold performance gap in the Vulkan test, and it reflects far more than just clock speed differences.

The GTX 690 has no winning benchmark in the database. Its two recorded tests, both Geekbench entries, place it far behind the newer card. Looking at the broader average benchmark scores, the RTX 2070 sits at 18,789 while the GTX 690 averages 17,037. The percentile rankings tell a similar story: the RTX 2070 ranks in the 63rd percentile of all GPUs, while the GTX 690 sits at the 60th percentile. That gap in percentile is modest, but the head-to-head deltas are enormous, which suggests the GTX 690's average is propped up by tests that favor its dual-GPU configuration while falling apart in modern API workloads.

For gaming and compute tasks that leverage Vulkan or OpenCL, the RTX 2070 is the clear choice. The GTX 690's architecture, built for a different era of DirectX, simply cannot keep pace in these cross-platform APIs. The data indicates the RTX 2070 is not just faster, it is in a different performance class altogether for these specific workloads.

Architecture Differences

The architectural gap between these two cards spans a full hardware generation and then some. The RTX 2070 uses the TU106 chip on TSMC's 12 nm process, packing 10,800 million transistors into a 445 mm² die. The GTX 690 uses two GK104 chips on TSMC's 28 nm process, with 3,540 million transistors per chip on a 294 mm² die. That means the RTX 2070 has roughly three times the transistor count of a single GTX 690 chip, and the transistor density tells the story: 24.3 million transistors per square millimeter for the RTX 2070 versus 12.0 million for the GTX 690.

The RTX 2070 is built on the Turing architecture, which introduces features the Kepler-based GTX 690 simply does not have. The RTX 2070 includes 36 RT cores and 288 tensor cores, dedicated hardware for ray tracing and AI acceleration. The GTX 690 has neither, as those features did not exist in the Kepler generation. This is a fundamental capability difference, not just a clock speed difference.

Memory configurations also diverge sharply. The RTX 2070 has 8 GB of GDDR6 memory on a 256-bit bus, delivering 448.0 GB/s of bandwidth. The GTX 690 has 2 GB of GDDR5 per GPU (the card is a dual-GPU design) on a 256-bit bus per chip, delivering 192.3 GB/s total. The RTX 2070's memory bandwidth is more than double, and the 8 GB capacity is four times the per-GPU allocation of the GTX 690.

Shader and output capabilities follow the same pattern. The RTX 2070 has 2,304 shading units, 144 texture mapping units, and 64 ROPs. The GTX 690 has 1,536 shading units, 128 TMUs, and 32 ROPs per GPU. Pixel rate is 103.7 GPixel/s for the RTX 2070 versus 32.61 GPixel/s for the GTX 690. Texture rate is 233.3 GTexel/s versus 130.4 GTexel/s. FP32 compute is 7.465 TFLOPS versus 3.130 TFLOPS, and the RTX 2070 also offers FP16 performance at 14.93 TFLOPS, a capability the GTX 690 lacks entirely.

API support is another differentiator. The RTX 2070 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The GTX 690 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The RTX 2070 also has a more modern display output set: one DVI, one HDMI 2.0, two DisplayPort 1.4a, and one USB Type-C. The GTX 690 has three DVI and one mini-DisplayPort 1.2.

Power requirements reflect the efficiency gains. The RTX 2070 has a 175 W TDP and requires a single 8-pin power connector with a suggested 450 W power supply. The GTX 690 has a 300 W TDP, needs two 8-pin connectors, and suggests a 700 W power supply. The RTX 2070 delivers far more performance while consuming nearly half the power budget of the GTX 690.

The Verdict

The data points to a decisive recommendation. For any user running Geekbench OpenCL or Vulkan workloads, the RTX 2070 is the only rational choice. It is roughly 3.5 to 4 times faster in these tests, has more than twice the memory bandwidth, four times the VRAM capacity per GPU, and adds hardware ray tracing and tensor cores that the GTX 690 cannot match. The RTX 2070 also does this while drawing 175 W versus 300 W, making it easier to power and cool.

The GTX 690's only advantage is its historical positioning. It was released in 2012 with a launch MSRP of 999 USD, and it represented the peak of NVIDIA's dual-GPU design at the time. But benchmark results indicate that peak is long past. Its 60th percentile ranking places it near the RTX 2070 in overall average score, but that average masks the catastrophic performance gap in modern APIs. The GTX 690's nearest rivals include the AMD Radeon RX 7600 XT (0.3% ahead) and the NVIDIA GeForce RTX 3070 (1% ahead), which suggests it still trades blows with mid-range cards from much later generations, but the RTX 2070 sits in a different league for the tests that matter today.

For a buyer choosing between these two today, the RTX 2070 is the clear pick. It is newer, faster, more efficient, and supports modern features. The GTX 690 is an end-of-life product with no path forward in current workloads. The database shows two wins for the RTX 2070 and zero for the GTX 690, and that record is decisive.

FAQ

Q: Which card is faster in Vulkan benchmarks?

A: The NVIDIA GeForce RTX 2070 scores 82,521 in Geekbench Vulkan, while the NVIDIA GeForce GTX 690 scores 16,675. The RTX 2070 leads by 394.9%.

Q: Does the GTX 690 support ray tracing?

A: No. The GTX 690 is built on the Kepler architecture and has no RT cores. The RTX 2070, built on Turing, includes 36 RT cores and 288 tensor cores.

Q: How much memory does each card have?

A: The RTX 2070 has 8 GB of GDDR6 memory on a 256-bit bus with 448.0 GB/s bandwidth. The GTX 690 has 2 GB of GDDR5 per GPU on a 256-bit bus per chip, with 192.3 GB/s total bandwidth.

Q: What is the power draw difference?

A: The RTX 2070 has a 175 W TDP and requires a single 8-pin connector with a suggested 450 W power supply. The GTX 690 has a 300 W TDP, needs two 8-pin connectors, and suggests a 700 W power supply.

Q: Which card has better API support?

A: The RTX 2070 supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The GTX 690 supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175.

Q: Are these cards still in production?

A: No. Both the RTX 2070 and the GTX 690 are marked as end-of-life in the database. The RTX 2070 was released in 2018, and the GTX 690 was released in 2012.

Head-to-Head Benchmarks

The head-to-head results are stark. The database records two direct comparisons, and the RTX 2070 wins both by enormous margins.

In Geekbench OpenCL, the RTX 2070 scores 79,966 against the GTX 690's 17,399. The delta is 359.6%, meaning the RTX 2070 is roughly 4.6 times faster in raw compute throughput. This test stresses general-purpose GPU compute, and the Turing architecture's 2,304 shading units and 7.465 TFLOPS of FP32 performance simply overwhelm the Kepler design's 1,536 shading units and 3.130 TFLOPS. The GTX 690's dual-GPU configuration does not help here, as OpenCL workloads often fail to scale across separate GPUs, and the 192.3 GB/s of total bandwidth is a severe bottleneck compared to the RTX 2070's 448.0 GB/s.

In Geekbench Vulkan, the gap widens further. The RTX 2070 scores 82,521, while the GTX 690 manages only 16,675. The delta is 394.9%, nearly a fivefold difference. Vulkan is a low-overhead API that favors modern architectures with efficient command processing and asynchronous compute capabilities. The GTX 690's Kepler architecture, with its DirectX 12 (11_0) feature level and older Vulkan 1.2.175 support, is at a fundamental disadvantage. The RTX 2070's DirectX 12 Ultimate and Vulkan 1.4 support, combined with its dedicated tensor cores and RT cores, allow it to execute these workloads far more efficiently.

Looking at the broader benchmark context, the RTX 2070's average benchmark score of 18,789 places it just 0.4% behind the NVIDIA Tesla K80 and 0.9% ahead of the AMD Radeon RX 560X. The GTX 690's average of 17,037 is 0.1% behind the AMD Radeon HD 7970M and 1% behind the NVIDIA GeForce RTX 3070. These averages suggest the two cards are comparable in mixed workloads, but the head-to-head tests reveal that the RTX 2070's strengths lie precisely in the modern APIs that matter most for current software.

The GTX 690's pixel rate of 32.61 GPixel/s and texture rate of 130.4 GTexel/s are less than a third of the RTX 2070's 103.7 GPixel/s and 233.3 GTexel/s. Even the GTX 690's boost clock of 1019 MHz, which is respectable for its era, cannot compensate for the architectural deficits. The RTX 2070's boost clock of 1620 MHz is 59% higher, and when combined with the newer process node and larger memory pipeline, the performance gap becomes insurmountable. The recorded data shows no scenario in which the GTX 690 overtakes the RTX 2070 in these benchmarks.