NVIDIA GeForce GTX 690 vs NVIDIA GeForce RTX 5090 D V2 Comparison

The NVIDIA GeForce GTX 690 and the NVIDIA GeForce RTX 5090 D V2 represent two distinct eras of GPU design, separated by over a decade of architectural evolution. The GTX 690, a dual-GPU Kepler flagship from 2012, and the RTX 5090 D V2, a single-chip Blackwell 2.0 behemoth from 2025, offer benchmark results that reflect their radically different design philosophies and target workloads.

Head-to-Head Benchmarks

The available benchmark data shows that the two cards do not share a single common test, making direct score-to-score comparisons impossible. However, the average benchmark scores and percentile rankings provide a useful proxy for overall performance. The GTX 690 achieves an average benchmark score of 17,037, while the RTX 5090 D V2 scores 16,504. This places the GTX 690 in the 60th percentile of all GPUs, with the RTX 5090 D V2 sitting at the 59th percentile. In terms of raw average score, the GTX 690 is 3.2% ahead of the RTX 5090 D V2.

The GTX 690’s individual benchmark results show a Geekbench OpenCL score of 17,399 and a Geekbench Vulkan score of 16,675. The RTX 5090 D V2’s only recorded benchmark is a 3DMark Steel Nomad DX12 score of 16,504. When examining the nearest rivals for each card, the GTX 690’s closest competitor is the NVIDIA GeForce RTX 3070, which has an average score of 17,208, putting the GTX 690 1% behind it. The GTX 690 also sits 0.3% behind the AMD Radeon RX 7600 XT (17,083) and 0.1% ahead of the AMD Radeon HD 7970M (17,019). For the RTX 5090 D V2, its closest rival is the NVIDIA T400 with an identical average score of 16,508 (0% delta). It also sits 0.5% ahead of the AMD Radeon PRO W7500 (16,415) and 0.6% ahead of the NVIDIA RTX PRO 6000 Blackwell (16,408).

The data indicates that the GTX 690’s dual-GPU configuration excels in compute-oriented workloads like OpenCL, while the RTX 5090 D V2’s single monolithic die delivers competitive DX12 performance. The percentile rankings, which are nearly identical, suggest that both cards occupy a similar tier of overall benchmark performance, despite the massive generational gap in their specifications. The lack of overlapping tests means that the 3.2% average score difference is not a definitive measure of victory in any specific workload, but rather a statistical summary of their respective benchmarking suites.

Architecture Differences

The GTX 690 is built on the Kepler architecture, utilizing the GK104 chip manufactured on a 28 nm process at TSMC. It contains 3,540 million transistors on a 294 mm² die, resulting in a transistor density of 12.0M per mm². In contrast, the RTX 5090 D V2 uses the Blackwell 2.0 architecture with the GB202 chip, fabricated on a 5 nm process, also at TSMC. This newer chip packs 92,200 million transistors into a 750 mm² die, achieving a transistor density of 122.9M per mm²—over ten times the density of the Kepler chip.

The RTX 5090 D V2 brings hardware features that the GTX 690 completely lacks: 170 ray tracing cores and 680 tensor cores. These dedicated units enable real-time ray tracing and AI-accelerated workloads, capabilities that did not exist in the Kepler generation. The GTX 690 is limited to DirectX 12 (11_0) support, while the RTX 5090 D V2 supports DirectX 12 Ultimate (12_2), along with Vulkan 1.4 compared to the GTX 690’s Vulkan 1.2.175. Both cards support OpenGL 4.6.

The memory subsystems are drastically different. The GTX 690 uses 2 GB of GDDR5 on a 256-bit bus with 192.3 GB/s bandwidth. The RTX 5090 D V2 uses 24 GB of GDDR7 on a 384-bit bus, delivering 1.34 TB/s bandwidth—a nearly sevenfold increase. The RTX 5090 D V2 also features a substantially larger compute configuration with 21,760 shading units, 680 TMUs, and 176 ROPs, compared to the GTX 690’s 1,536 shading units, 128 TMUs, and 32 ROPs. This translates to a peak FP32 throughput of 104.8 TFLOPS for the RTX 5090 D V2 versus 3.130 TFLOPS for the GTX 690, a 33-fold increase. The RTX 5090 D V2 also offers FP16 performance at a 1:1 ratio with FP32, while the GTX 690 has no listed FP16 capability.

FAQ

Q: Which card has the higher average benchmark score?

A: The GTX 690 has an average benchmark score of 17,037, while the RTX 5090 D V2 has an average score of 16,504. The GTX 690 is 3.2% higher.

Q: How do the two cards compare in terms of memory capacity and type?

A: The GTX 690 features 2 GB of GDDR5 memory, while the RTX 5090 D V2 features 24 GB of GDDR7 memory. The RTX 5090 D V2 also has a wider 384-bit bus compared to the GTX 690’s 256-bit bus.

Q: Does the RTX 5090 D V2 support ray tracing?

A: Yes, the RTX 5090 D V2 includes 170 ray tracing cores. The GTX 690 does not have any ray tracing cores.

Q: What is the difference in their process nodes?

A: The GTX 690 is manufactured on a 28 nm process, while the RTX 5090 D V2 is manufactured on a 5 nm process. Both are produced by TSMC.

Q: Which card has a higher pixel rate?

A: The RTX 5090 D V2 has a pixel rate of 423.6 GPixel/s, compared to the GTX 690’s 32.61 GPixel/s.

Q: What are the launch MSRPs of these two cards?

A: The GTX 690 had a launch MSRP of 999 USD, while the RTX 5090 D V2 had a launch MSRP of 2,299 USD.

Specification Differences

The two cards differ across nearly every specification field. The GTX 690 uses the Kepler architecture on a 28 nm process, while the RTX 5090 D V2 uses Blackwell 2.0 on a 5 nm process. Transistor counts are 3,540 million versus 92,200 million, and die sizes are 294 mm² versus 750 mm². The GTX 690 has a base clock of 915 MHz and a boost clock of 1019 MHz, compared to the RTX 5090 D V2’s 2017 MHz base and 2407 MHz boost. Memory clocks are 1502 MHz (6 Gbps effective) for the GTX 690 versus 1750 MHz (28 Gbps effective) for the RTX 5090 D V2.

Memory capacity differs from 2 GB to 24 GB, with GDDR5 versus GDDR7 types. The bus width expands from 256-bit to 384-bit, and bandwidth jumps from 192.3 GB/s to 1.34 TB/s. Shading units increase from 1,536 to 21,760, TMUs from 128 to 680, and ROPs from 32 to 176. The RTX 5090 D V2 adds 170 RT cores and 680 tensor cores, which the GTX 690 lacks. Pixel rate rises from 32.61 GPixel/s to 423.6 GPixel/s, texture rate from 130.4 GTexel/s to 1,636.8 GTexel/s, and FP32 performance from 3.130 TFLOPS to 104.8 TFLOPS.

The TDP differs significantly, with the GTX 690 rated at 300 W and the RTX 5090 D V2 at 575 W. Power connectors change from 2x 8-pin to 1x 16-pin, and suggested PSU rises from 700 W to 950 W. The bus interface upgrades from PCIe 3.0 x16 to PCIe 5.0 x16. Display outputs change from 3x DVI and 1x mini-DisplayPort 1.2 to 1x HDMI 2.1b and 3x DisplayPort 2.1b. Dimensions also differ, with the GTX 690 measuring 279 mm by 111 mm by 38 mm, while the RTX 5090 D V2 measures 304 mm by 137 mm by 48 mm. Production status differs from End-of-life to Active, with release dates of 2012-05-02 versus 2025-08-14.

Where Each One Wins

The GTX 690 wins in the OpenCL compute benchmark, scoring 17,399, which is its highest recorded result. This suggests that its dual-GPU configuration is well-suited to certain parallel compute workloads. It also holds a slight edge in average benchmark score over the RTX 5090 D V2, indicating that its overall benchmark suite performance is marginally higher. The GTX 690 achieves a 60th percentile ranking, one point above the RTX 5090 D V2’s 59th percentile.

The RTX 5090 D V2 wins in raw architectural capabilities. It offers a 33-fold higher FP32 throughput, a 13-fold higher pixel rate, and a 12.5-fold higher texture rate compared to the GTX 690. Its memory bandwidth of 1.34 TB/s is nearly seven times higher. The inclusion of RT cores and tensor cores gives it capabilities in ray tracing and AI workloads that the GTX 690 cannot perform at all. The RTX 5090 D V2 also supports newer API versions, including DirectX 12 Ultimate and Vulkan 1.4.

In terms of power efficiency relative to performance, the RTX 5090 D V2 delivers substantially more compute per watt, despite its higher absolute TDP. The GTX 690’s 3.130 TFLOPS over 300 W yields approximately 10.4 GFLOPS/W, while the RTX 5090 D V2’s 104.8 TFLOPS over 575 W yields approximately 182.3 GFLOPS/W—a 17.5-fold improvement in efficiency.

The Verdict

The data paints a clear picture of two cards built for different purposes. The GTX 690, with its higher average benchmark score and 60th percentile ranking, remains competitive in legacy compute benchmarks like OpenCL and Vulkan. Its dual-GPU design from the Kepler era delivers respectable scores even against modern hardware, which is remarkable for a card released in 2012.

The RTX 5090 D V2, despite a slightly lower average benchmark score and 59th percentile ranking, offers a fundamentally more capable feature set. Its 104.8 TFLOPS FP32 performance, 24 GB of GDDR7 memory, and dedicated RT and tensor cores make it the clear choice for modern gaming, ray tracing, AI inference, and high-resolution workloads. The GTX 690’s 2 GB memory capacity and lack of hardware ray tracing render it unsuitable for contemporary AAA titles.

For users seeking raw compute performance in legacy APIs, the GTX 690’s benchmark scores indicate it can still hold its own. However, for anyone building a system today, the RTX 5090 D V2’s architectural advantages, superior efficiency, and future-proofing make it the logical selection. The data shows that while the GTX 690 wins on average benchmark score, the RTX 5090 D V2 wins on every meaningful technical specification that determines real-world performance in modern applications. The choice is clear: pick the GTX 690 for retro benchmarking or niche compute tasks, and the RTX 5090 D V2 for anything requiring current-generation performance and features.